Post by thread:[EE]: Car power supply challenges

I think I've done my homework on this one folks. I googled. I checked the archives for this list. I checked the news archives. Here's the situation: I want to power a PC in the car. 5V has multiple solutions so it isn't of concern here. I'm particularly interested in the 12V supply. The supply will have a 12V 7Ah gel cell to act as a mini UPS. So the input voltage will be from 12 to 14 volts. I want to get a regulated supply of 12V +- 5% @ 4A. Lastly and most importantly to me: It needs to be a discrete solution with components that can be purchased at the local parts store. Think Radio Shack or Radio Spares. This requirement has a twofold purpose: I want to learn the process, and I want to be able to have readily available replacement components. From the specs the problem is obvious: How in the heck do you get a regulated 12V output from a possible 12V input? It's impossible. However with the error term regulation down to 11.4V is acceptable. So 12V in and 11.4V output screams low dropout regulator. So I started researching them. NPN and NMOS are immediately depricated. PMOS and PNP are the way to go with the series pass element. This leaves the problem of the control/error element. The obvious, a regular linear regulator, is out because the are amost universally NPN devices, which means high dropout voltages. The next choice is a zener. Simple, clean, and reasonable regulation. But each and every circuit I've seen that uses a zener control has an NPN pass element. Again high dropout voltages. So I was wondering if anyone had any ideas on how a zener can be coupled with a PNP/PMOS pass element so that reasonable regulation and low dropout is still possible. On a parallel path I did take a look at the buck switching regulators that have been discussed in the last year. However they seemed to need more voltage headroom than I was able to provide. One last issue to raise again was the input voltage spike protection. Most of the discussions that have been proposed in the past have been rather low current. I was wondering if a low valued resistor with a power shunt before the supply could work. For a shunt I was thinking along the lines of Fr. Thomas' power shunt here: http://www.infosite.com/~jkeyzer/piclist/2001/Jan/2382.html (login and password are the name of this list in lowercase) or a simple crowbar using a high valued zener driving an NPN shunted to ground. This would be helpful because I really wanted to regulate the UPS batteries input voltage to 13.8V anyway. Well this is what I've thought up so far. I'd really be interested in getting some input on how to solve these issues. BAJ -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Here's the situation: > > I want to power a PC in the car. 5V has multiple solutions so it isn't > of concern here. I'm particularly interested in the 12V supply. > > The supply will have a 12V 7Ah gel cell to act as a mini UPS. So the > input voltage will be from 12 to 14 volts. > > I want to get a regulated supply of 12V +- 5% @ 4A. > > Lastly and most importantly to me: It needs to be a discrete solution with > components that can be purchased at the local parts store. Think Radio Shack > or Radio Spares. This requirement has a twofold purpose: I want to learn > the process, and I want to be able to have readily available replacement > components. > > From the specs the problem is obvious: How in the heck do you get a regulated > 12V output from a possible 12V input? It's impossible. However with the > error term regulation down to 11.4V is acceptable. Hello BAJ I know that you would like to make this regulator with discrete components, but.... If I had a 12 V in and need a +/- 12 V output I would use a DC/DC converter. There is a company named TRACO Power that makes small nice items in different Watts. In sweden you can get this part from for example Elfa, http://www.elfa.se. If you do a search on the art. number 69-622-60 you will find a DC/DC converter with a 12V input and a +/- 12V output without any extra components. It's a 2W, but there are up to 15W, 9-18V in and +/-12 out. I have used these components alot and are very pleased with them. They are simple, they work.. But they are a litle bit to expensive. Regards Michael Pettersson -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Radio Shack has a dc power supply for computers. lighter socket input, select the proper plug output, set dial for desired voltage out. Works like a champ. voltage out can be above or below 12v ex. 18v for my compaq and toshiba. Byron A Jeff wrote: {Quote hidden}> > I think I've done my homework on this one folks. I googled. I checked > the archives for this list. I checked the news archives. > > Here's the situation: > > I want to power a PC in the car. 5V has multiple solutions so it isn't > of concern here. I'm particularly interested in the 12V supply. > > The supply will have a 12V 7Ah gel cell to act as a mini UPS. So the > input voltage will be from 12 to 14 volts. > > I want to get a regulated supply of 12V +- 5% @ 4A. > > Lastly and most importantly to me: It needs to be a discrete solution with > components that can be purchased at the local parts store. Think Radio Shack > or Radio Spares. This requirement has a twofold purpose: I want to learn > the process, and I want to be able to have readily available replacement > components. > > >From the specs the problem is obvious: How in the heck do you get a regulated > 12V output from a possible 12V input? It's impossible. However with the > error term regulation down to 11.4V is acceptable. > > So 12V in and 11.4V output screams low dropout regulator. So I started > researching them. NPN and NMOS are immediately depricated. PMOS and PNP > are the way to go with the series pass element. This leaves the problem of the > control/error element. The obvious, a regular linear regulator, is out > because the are amost universally NPN devices, which means high dropout > voltages. > > The next choice is a zener. Simple, clean, and reasonable regulation. But > each and every circuit I've seen that uses a zener control has an NPN pass > element. Again high dropout voltages. > > So I was wondering if anyone had any ideas on how a zener can be coupled with > a PNP/PMOS pass element so that reasonable regulation and low dropout is > still possible. > > On a parallel path I did take a look at the buck switching regulators that > have been discussed in the last year. However they seemed to need more voltage > headroom than I was able to provide. > > One last issue to raise again was the input voltage spike protection. Most > of the discussions that have been proposed in the past have been rather low > current. I was wondering if a low valued resistor with a power shunt before > the supply could work. For a shunt I was thinking along the lines of Fr. > Thomas' power shunt here: > > http://www.infosite.com/~jkeyzer/piclist/2001/Jan/2382.html > > (login and password are the name of this list in lowercase) > > or a simple crowbar using a high valued zener driving an NPN shunted to > ground. This would be helpful because I really wanted to regulate the UPS > batteries input voltage to 13.8V anyway. > > Well this is what I've thought up so far. I'd really be interested in getting > some input on how to solve these issues. > > BAJ > > -- > http://www.piclist.com hint: The PICList is archived three different > ways. See http://www.piclist.com/#archives for details. -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Hi Byron, I think you have already hit on the cleverest solution, using a gell cell battery allows you to use a linear series regulator, and because of the storage capacity of the battery you can regulate the output voltage up to the PEAK input voltage. Not the MINIMUM input voltage like a normal series regulator. :o) You should aim for your battery to be kept at about 12.8v, and full, so it can be charged anytime the car is above 13.0v with a simple series regulator transistor. The only times the car will be under 13.0v will be when you start it, or maybe idling with the headlights on. Anytime the motor is "driving" the car will produce about 14.2v to 14.5v. -Roman Byron A Jeff wrote: {Quote hidden}> > I think I've done my homework on this one folks. I googled. I checked > the archives for this list. I checked the news archives. > > Here's the situation: > > I want to power a PC in the car. 5V has multiple solutions so it isn't > of concern here. I'm particularly interested in the 12V supply. > > The supply will have a 12V 7Ah gel cell to act as a mini UPS. So the > input voltage will be from 12 to 14 volts. > > I want to get a regulated supply of 12V +- 5% @ 4A. > > Lastly and most importantly to me: It needs to be a discrete solution with > components that can be purchased at the local parts store. Think Radio Shack > or Radio Spares. This requirement has a twofold purpose: I want to learn > the process, and I want to be able to have readily available replacement > components. > > >From the specs the problem is obvious: How in the heck do you get a regulated > 12V output from a possible 12V input? It's impossible. However with the > error term regulation down to 11.4V is acceptable. > > So 12V in and 11.4V output screams low dropout regulator. So I started > researching them. NPN and NMOS are immediately depricated. PMOS and PNP > are the way to go with the series pass element. This leaves the problem of the > control/error element. The obvious, a regular linear regulator, is out > because the are amost universally NPN devices, which means high dropout > voltages. > > The next choice is a zener. Simple, clean, and reasonable regulation. But > each and every circuit I've seen that uses a zener control has an NPN pass > element. Again high dropout voltages. > > So I was wondering if anyone had any ideas on how a zener can be coupled with > a PNP/PMOS pass element so that reasonable regulation and low dropout is > still possible. > > On a parallel path I did take a look at the buck switching regulators that > have been discussed in the last year. However they seemed to need more voltage > headroom than I was able to provide. > > One last issue to raise again was the input voltage spike protection. Most > of the discussions that have been proposed in the past have been rather low > current. I was wondering if a low valued resistor with a power shunt before > the supply could work. For a shunt I was thinking along the lines of Fr. > Thomas' power shunt here: > > http://www.infosite.com/~jkeyzer/piclist/2001/Jan/2382.html > > (login and password are the name of this list in lowercase) > > or a simple crowbar using a high valued zener driving an NPN shunted to > ground. This would be helpful because I really wanted to regulate the UPS > batteries input voltage to 13.8V anyway. -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

On Wed, Jan 09, 2002 at 02:59:40PM -0600, Don Hyde wrote: > This might be time to examine your original specification. Ok. Let's. > > Do you need to be able to replace the power supply in any old generic PC and > make it work in a car? In this case, you need to really provide plenty of > +12 over the car's full voltage excursions -- which means working from +8V > or less during cranking the engine or when the battery's almost dead, to > 13.8V, which is most of the time when the alternator is working, but not > keeling over dead when the 40V spikes hit as the power windows and door > locks are used. The UPS battery takes care of the undervoltage. The overvoltage issues still apply. > > If you need to get SOME PC to work in a car, then examine PC's. These days > most of them don't use much +12. You might find (or configure) one that > doesn't need any at all -- it's usually stuff like sound cards that still > use +12. I'm pretty sure there are even disk drives that use only +5. Laptop drives. However it's my preference to use ordinary 3.5in drives because of their phenominal prices. Also I plan to interface a 100 MB ZIP drive to the unit which requires 12V. > If > you can get it down to a few milliamps, then you can sacrifice some > efficiency for simplicity and use an off-the-shelf and > not-extremely-expensive DC-DC converter to make the needed +12 from the +5, > which as you say, is much easier to get. > > The general solution is probably going to need a pretty sophisticated > switching boost-buck converter that can take whatever it gets and make nice > clean +12 out of it, and that's going to be expensive and complicated. I think (mind you think) that the UPS battery will preclude the need for a boost situation. The battery has quite good about providing 12V for moderate periods of time. The issue at hand is how to do LDO regulation with 0.6V of headroom. A PNP or PMOS pass transistor can do it, but how does one control it? BAJ > > > {Original Message removed}

I got a computer running using a small surplus UPS (110V). You could fool it into turning on without even being plugged in by switching the power switch on and off twice quickly. Then I used a standard PC power supply switched to 110V input. The UPS battery was replaced with leads to the car battery. To efficiently get 12V from above or below 12V without winding transformers, you can use 'SEPIC' topology converters. These are non-isolated and only use an extra inductor. You can use a standard converter chip, lots of these can do SEPIC. ----- Original Message ----- From: "Byron A Jeff" <spam_OUTbyronTakeThisOuTCC.GATECH.EDU> To: <.....PICLISTKILLspam@spam@MITVMA.MIT.EDU> Sent: Thursday, January 10, 2002 8:58 AM Subject: Re: [EE]: Car power supply challenges > On Wed, Jan 09, 2002 at 02:59:40PM -0600, Don Hyde wrote: > > This might be time to examine your original specification. > > Ok. Let's. > > > > > Do you need to be able to replace the power supply in any old generic PC and > > make it work in a car? In this case, you need to really provide plenty of > > +12 over the car's full voltage excursions -- which means working from +8V > > or less during cranking the engine or when the battery's almost dead, to > > 13.8V, which is most of the time when the alternator is working, but not > > keeling over dead when the 40V spikes hit as the power windows and door > > locks are used. > > The UPS battery takes care of the undervoltage. The overvoltage issues still > apply. > > > > > If you need to get SOME PC to work in a car, then examine PC's. These days > > most of them don't use much +12. You might find (or configure) one that > > doesn't need any at all -- it's usually stuff like sound cards that still > > use +12. I'm pretty sure there are even disk drives that use only +5. > > Laptop drives. However it's my preference to use ordinary 3.5in drives because > of their phenominal prices. > > Also I plan to interface a 100 MB ZIP drive to the unit which requires 12V. > > > If > > you can get it down to a few milliamps, then you can sacrifice some > > efficiency for simplicity and use an off-the-shelf and > > not-extremely-expensive DC-DC converter to make the needed +12 from the +5, > > which as you say, is much easier to get. > > > > The general solution is probably going to need a pretty sophisticated > > switching boost-buck converter that can take whatever it gets and make nice > > clean +12 out of it, and that's going to be expensive and complicated. > > I think (mind you think) that the UPS battery will preclude the need for a > boost situation. The battery has quite good about providing 12V for moderate > periods of time. > > The issue at hand is how to do LDO regulation with 0.6V of headroom. A PNP > or PMOS pass transistor can do it, but how does one control it? > > BAJ > > > > > {Original Message removed}

On Thu, Jan 10, 2002 at 08:44:58AM +1100, Roman Black wrote: > Hi Byron, I think you have already hit on the > cleverest solution, using a gell cell battery > allows you to use a linear series regulator, > and because of the storage capacity of the > battery you can regulate the output voltage up > to the PEAK input voltage. Not the MINIMUM > input voltage like a normal series regulator. > :o) Thanks! It was the solution to the starter dropout voltage when I was driving this equipment with a laptop power supply I got surplus from Marlon P Jones. Unfortunately they don't carry it anymore and I succeeded in shorting out the two rather fragile units I had. > > You should aim for your battery to be kept at > about 12.8v, and full, so it can be charged > anytime the car is above 13.0v with a simple > series regulator transistor. The only times the > car will be under 13.0v will be when you start > it, or maybe idling with the headlights on. Anytime > the motor is "driving" the car will produce about > 14.2v to 14.5v. Correct. But the problem is getting the 12V line to regulate when only the UPS gel cell is on line. While it'll float at 12.8V to 13V, it quickly heads towards 12V when put under load. I'm trying to ensure that the drives will not brown out during starting and when the car has short stops in the 5 to 10 minute range. The nominal system input voltage will be 13.8V. It should provide enough headroom for normal use. It's the 12V minimum input voltage that I'm trying to attack. Thanks for the encouragement. BAJ {Quote hidden}> -Roman > > > > Byron A Jeff wrote: > > > > I think I've done my homework on this one folks. I googled. I checked > > the archives for this list. I checked the news archives. > > > > Here's the situation: > > > > I want to power a PC in the car. 5V has multiple solutions so it isn't > > of concern here. I'm particularly interested in the 12V supply. > > > > The supply will have a 12V 7Ah gel cell to act as a mini UPS. So the > > input voltage will be from 12 to 14 volts. > > > > I want to get a regulated supply of 12V +- 5% @ 4A. > > > > Lastly and most importantly to me: It needs to be a discrete solution with > > components that can be purchased at the local parts store. Think Radio Shack > > or Radio Spares. This requirement has a twofold purpose: I want to learn > > the process, and I want to be able to have readily available replacement > > components. > > > > >From the specs the problem is obvious: How in the heck do you get a regulated > > 12V output from a possible 12V input? It's impossible. However with the > > error term regulation down to 11.4V is acceptable. > > > > So 12V in and 11.4V output screams low dropout regulator. So I started > > researching them. NPN and NMOS are immediately depricated. PMOS and PNP > > are the way to go with the series pass element. This leaves the problem of the > > control/error element. The obvious, a regular linear regulator, is out > > because the are amost universally NPN devices, which means high dropout > > voltages. > > > > The next choice is a zener. Simple, clean, and reasonable regulation. But > > each and every circuit I've seen that uses a zener control has an NPN pass > > element. Again high dropout voltages. > > > > So I was wondering if anyone had any ideas on how a zener can be coupled with > > a PNP/PMOS pass element so that reasonable regulation and low dropout is > > still possible. > > > > On a parallel path I did take a look at the buck switching regulators that > > have been discussed in the last year. However they seemed to need more voltage > > headroom than I was able to provide. > > > > One last issue to raise again was the input voltage spike protection. Most > > of the discussions that have been proposed in the past have been rather low > > current. I was wondering if a low valued resistor with a power shunt before > > the supply could work. For a shunt I was thinking along the lines of Fr. > > Thomas' power shunt here: > > > > www.infosite.com/~jkeyzer/piclist/2001/Jan/2382.html > > > > (login and password are the name of this list in lowercase) > > > > or a simple crowbar using a high valued zener driving an NPN shunted to > > ground. This would be helpful because I really wanted to regulate the UPS > > batteries input voltage to 13.8V anyway. > > -- > http://www.piclist.com hint: The PICList is archived three different > ways. See http://www.piclist.com/#archives for details. > -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

> The supply will have a 12V 7Ah gel cell to act as a mini UPS. So the > input voltage will be from 12 to 14 volts. > > I want to get a regulated supply of 12V +- 5% @ 4A. Be aware that a 12v gel cell "floated" off the main battery by a diode will be undercharged when the main battery is charged. The extra 0.6v or so can make quite a difference to cell life. This comment is based on practical experience with 100m + car based systems using this arrangement and a standard silicon diode. If this is one off or small volume then use of a low voltage drop diode can be useful eg Schottky diode. If you don't mind having to replace the gel cell every year or 2 this may not be too much of a problem. A suitably low Rdson FET driven hard on is also very good BUT as the next few sentences show, is surprisingly hard to use properly here. . Getting FET drive voltage for an N Channel FET can be a pain as it has to be positive ABOVE battery positive. Note that if a FET is used the gel cell will try and drive the car via the FET body diode when the car battery is low under eg cranking conditions. A solution is to use the FET "backwards" so that the body diode conducts to charge the gel cell and the on FET also conducts in the same direction. You STILL have to bias an N channel FET with the gate positive above battery even though you have reversed Drain & Source so biasing is just as annoying. A P channel FET will solve this but their Rdson is usually higher for otherwise similar specs. Use a Schottky diode :-). A small boost converter sitting on the battery rail will solve these problems. The 12v is boosted by about 2 volts or less to allow for diode drop and low battery voltage at times. Wattage required is low as it only supplies about 2 volts to add to the 12 volts - not the whole 13 volts or so that you need. Russell McMahon . -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservKILLspammitvma.mit.edu with SET PICList DIGEST in the body

Byron A Jeff wrote: {Quote hidden}> > On Thu, Jan 10, 2002 at 08:44:58AM +1100, Roman Black wrote: > > because of the storage capacity of the > > battery you can regulate the output voltage up > > to the PEAK input voltage. Not the MINIMUM > > input voltage like a normal series regulator. > > Thanks! It was the solution to the starter dropout voltage when I was driving > this equipment with a laptop power supply I got surplus from Marlon P Jones. > Unfortunately they don't carry it anymore and I succeeded in shorting out > the two rather fragile units I had. > > > You should aim for your battery to be kept at > > about 12.8v, and full, so it can be charged > > anytime the car is above 13.0v with a simple > > series regulator transistor. The only times the > > car will be under 13.0v will be when you start > > it, or maybe idling with the headlights on. Anytime > > the motor is "driving" the car will produce about > > 14.2v to 14.5v. > > Correct. But the problem is getting the 12V line to regulate when only the > UPS gel cell is on line. While it'll float at 12.8V to 13V, it quickly heads > towards 12V when put under load. No drama there, low dropout series regs are not that hard to do with a decent low-sat PNP device and a 7805 as a zener. You need to determine your max current and typical current for the 12v rail, if it's only 2A or so this will be an easy task. If it's 4A continuous the task gets slightly harder. > I'm trying to ensure that the drives will not brown out during starting and > when the car has short stops in the 5 to 10 minute range. That should not be a problem, most modern floppies i've seen either don't use the 12v anymore, (and I strip 5v motors from them) or the 12v motors are 150mA or so. Even a floppy starting up will not draw more than 0.5A from your 12v supply. Pull the cover off your PC and stick an amp meter on the 12v rail and check it out. Design from the back forward. :o) > The nominal system input voltage will be 13.8V. It should provide enough > headroom for normal use. Not so. 13.8 is an "on charge" voltage. The moment the car stops, your 13.8 sinks quickly back to a typical battery "full" voltage of about 12.8v. > It's the 12V minimum input voltage that I'm trying to attack. I don't think it's a problem. Assuming 12.8v in your gell battery when the car stops, and 2A used at 12v, you need a low dropout regulator good for 2A at 0.3v drop, and you have a headroom of 0.8v-0.3v=0.5v. So your 12.8v can drop to 12.3v before the 12v out is affected. I would maybe choose a regulated 11.6v rail, as this gives you another 0.4v headroom, and a large enough battery to supply you for 10 minutes at 2A and not drop 0.9v. Remember this will be the 12v battery only, if you need a 5v supply to run the PC at some 20A or more, that will involve a whole 'nother battery and PSU... -Roman > Thanks for the encouragement. Anytime! :o) -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email @spam@listservKILLspammitvma.mit.edu with SET PICList DIGEST in the body

Roman, Jeff, You have damn good car batteries there if you consider only 12.8 to 14.5 voltage variations ! Here a battery having 10...11V with load is considered excellent. And I'm not talking about measurement at -15C Also a defective charging relay could charge your battery up to 16V I think ( with batteries from here ) you have no chance with linear regulators. Just my personal opinion, please don't kill me... best regards and succes, Vasile On Thu, 10 Jan 2002, Roman Black wrote: {Quote hidden}> Byron A Jeff wrote: > > > > On Thu, Jan 10, 2002 at 08:44:58AM +1100, Roman Black wrote: > > > because of the storage capacity of the > > > battery you can regulate the output voltage up > > > to the PEAK input voltage. Not the MINIMUM > > > input voltage like a normal series regulator. > > > > Thanks! It was the solution to the starter dropout voltage when I was driving > > this equipment with a laptop power supply I got surplus from Marlon P Jones. > > Unfortunately they don't carry it anymore and I succeeded in shorting out > > the two rather fragile units I had. > > > > > You should aim for your battery to be kept at > > > about 12.8v, and full, so it can be charged > > > anytime the car is above 13.0v with a simple > > > series regulator transistor. The only times the > > > car will be under 13.0v will be when you start > > > it, or maybe idling with the headlights on. Anytime > > > the motor is "driving" the car will produce about > > > 14.2v to 14.5v. > > > > Correct. But the problem is getting the 12V line to regulate when only the > > UPS gel cell is on line. While it'll float at 12.8V to 13V, it quickly heads > > towards 12V when put under load. > > No drama there, low dropout series regs are not that > hard to do with a decent low-sat PNP device and a > 7805 as a zener. > > You need to determine your max current and typical > current for the 12v rail, if it's only 2A or so this > will be an easy task. If it's 4A continuous the task > gets slightly harder. > > > I'm trying to ensure that the drives will not brown out during starting and > > when the car has short stops in the 5 to 10 minute range. > > That should not be a problem, most modern floppies > i've seen either don't use the 12v anymore, (and I > strip 5v motors from them) or the 12v motors are > 150mA or so. Even a floppy starting up will not draw > more than 0.5A from your 12v supply. Pull the cover > off your PC and stick an amp meter on the 12v rail > and check it out. Design from the back forward. :o) > > > The nominal system input voltage will be 13.8V. It should provide enough > > headroom for normal use. > > Not so. 13.8 is an "on charge" voltage. The moment > the car stops, your 13.8 sinks quickly back to a > typical battery "full" voltage of about 12.8v. > > > It's the 12V minimum input voltage that I'm trying to attack. > > I don't think it's a problem. Assuming 12.8v in your > gell battery when the car stops, and 2A used at 12v, > you need a low dropout regulator good for 2A at 0.3v > drop, and you have a headroom of 0.8v-0.3v=0.5v. > So your 12.8v can drop to 12.3v before the 12v out > is affected. I would maybe choose a regulated 11.6v > rail, as this gives you another 0.4v headroom, and > a large enough battery to supply you for 10 minutes > at 2A and not drop 0.9v. > > Remember this will be the 12v battery only, if you > need a 5v supply to run the PC at some 20A or more, > that will involve a whole 'nother battery and PSU... > -Roman > > > Thanks for the encouragement. > Anytime! :o) > > -- > http://www.piclist.com#nomail Going offline? Don't AutoReply us! > email KILLspamlistservKILLspammitvma.mit.edu with SET PICList DIGEST in the body > > -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

On Thu, Jan 10, 2002 at 11:54:58AM +1300, Russell McMahon wrote: > > The supply will have a 12V 7Ah gel cell to act as a mini UPS. So the > > input voltage will be from 12 to 14 volts. > > > > I want to get a regulated supply of 12V +- 5% @ 4A. > > > Be aware that a 12v gel cell "floated" off the main battery by a diode will > be undercharged when the main battery is charged. The extra 0.6v or so can > make quite a difference to cell life. This comment is based on practical > experience with 100m + car based systems using this arrangement and a > standard silicon diode. If this is one off or small volume > then use of a low voltage drop diode can be useful eg Schottky diode. If you > don't mind having to replace the gel cell every year or 2 this may not be > too much of a problem. Neither is an issue because it is a limited run (less then 10 units total). In my prototype I did use the Schottky diode. I was always concered with overcharging. My altenator (sp?) runs about 14.2V. [FET deleted for brevity] Russell, I appreciate the pointer, but charging the gel cell is the least of my concerns in this setup. The fundamental question is how to get a reliable 12V +- 5% when the gel cell is in its main discharge curve and the output voltage is near 12V. BTW you probably are not aware that your now famous buck converter is in the middle of this mix. I plan to use it to generate my 5V supply. So you've helped me immensely already! Thanks, BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email spamBeGonelistservspamBeGonemitvma.mit.edu with SET PICList DIGEST in the body

On Thu, Jan 10, 2002 at 10:13:17AM +0200, Vasile Surducan wrote: > On Wed, 9 Jan 2002, Byron A Jeff wrote: > > > > > >From the specs the problem is obvious: How in the heck do you get a regulated > > 12V output from a possible 12V input? It's impossible. However with the > > error term regulation down to 11.4V is acceptable. > > > No is not imposible. How is possible to have regulated +5V from +3V ? But it's a different problem. I'm not trying to generate a higher voltage. If I were then I'd be looking at a boost converter. I'd also be looking at a flyback arrangement if the primary input voltage were to vary significantly around the output voltage. But neither of these are the case. I can pretty much guarantee that the input voltage will be at or above 12V. I am willing to have a regulated voltage that is below that input voltage. It's not a boost situation. > A swiching power supply could do it. There are so many solution. Take a > look to any switching supply driver producer ( Linear, Motorola etc. ) > best regards, Vasile I'll already be using a buck converter to get my 5V line. But any buck converter solutions for the 12V line will be subject to exactly the same issues as the linear regulator design: ensuring that the pass element limits the voltage drop accoss it. Let me try to pull Roman and Russell back into the conversation. IIRC Russell said in one of his posts of his 2 transistor switching PS that it required 1V of headroom. Was there a particular reason for that design spec? What happens if the input voltage goes into the dropout range of within 1V of the regulated output voltage? Especially when that voltage is right at the breakdown knee of the zener control element? I'm almost certain that this doesn't require a buck/boost or boost/buck situation. In fact it should work well as a linear situation since it only needs to drop between 0.3 and 2V. BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email TakeThisOuTlistservEraseMEspam_OUTmitvma.mit.edu with SET PICList DIGEST in the body

On Thu, Jan 10, 2002 at 10:19:20AM +0200, Vasile Surducan wrote: > On Wed, 9 Jan 2002, Byron A Jeff wrote: > > > > > The objective is to drive an ordinary AT motherboard, standard 3.5 in disk > > and a 100 MB ZIP drive. Standard, if not dated, desktop PC equipment. > > > Why don't you say that ? There are standard PC power supply powered from > 12V. And they cost $350 US apiece. Out of my price range. My price range is $30 US. So that rules out anything but surplus and homebuilt. Again I'm using this as a learning situation. Also I want to build this out of commodity components. The simplest solution is to buy a 400-500W inverter and use a standard supply. There are several dozen LDO regulators that can do the job. There are a bunch of buck/boost arrangements that solve the problem. Bob Blick probably has the cleanest solution of all using a flyback transformer to generate multiple voltages. But what I really want is a solution that is akin to Russell and Roman's switching power supply designs. Simple. Inexpensive. Easy to understand. Uses commodity components. In that way it's possible to adapt the design to different situations, and one doesn't get caught up waiting for specialty components that are usually very pricy. That why I was very specific with the design challenge. There are bunches of solutions out there. But there are none that I could find that are simple and use commodity components. BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

On Thu, 10 Jan 2002, Byron A Jeff wrote: > > > >From the specs the problem is obvious: How in the heck do you get a regulated > > > 12V output from a possible 12V input? It's impossible. However with the > > > error term regulation down to 11.4V is acceptable. > > > > > No is not imposible. How is possible to have regulated +5V from +3V ? > > But it's a different problem. I'm not trying to generate a higher voltage. > If I were then I'd be looking at a boost converter. > > But neither of these are the case. I can pretty much guarantee that the > input voltage will be at or above 12V. I am willing to have a regulated > voltage that is below that input voltage. You have much heavily design condition that you imagine, Byron: input voltage from 11 to 16 V input spikes from 0 to hundred of volts ambient temperature from -35 to +70 C ( sorry for Farenheit speakers ) plenty of vibrations You can't do it with a linear regulator ( or maybe yes with some malfunctions or choosing right the output voltage at 9 or better 8V). A simple isolated DC-DC converter will solve all your headache regarding isolating spikes. The same one will gave you +5V and any other voltage ( including 5000V for ionisation of the habitacle etc. ) By the other hand if this is only for supplying the motherboard, HDD and CDrom, then probably you know there are boards which are running well at +8V instead of +12 and without -12 and/or -5V. Also newest CDroms doesn't need +12V at all, only +5V. > Let me try to pull Roman and Russell back into the conversation. IIRC Russell > said in one of his posts of his 2 transistor switching PS that it required > 1V of headroom. Was there a particular reason for that design spec? What > happens if the input voltage goes into the dropout range of within 1V of > the regulated output voltage? Especially when that voltage is right at > the breakdown knee of the zener control element? > Then the stabiliser will have the same function like a resistor. best regards, Vasile -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservEraseME.....mitvma.mit.edu with SET PICList DIGEST in the body

On Thu, Jan 10, 2002 at 02:22:30PM +0200, Vasile Surducan wrote: > On Thu, 10 Jan 2002, Byron A Jeff wrote: > > > > > >From the specs the problem is obvious: How in the heck do you get a regulated > > > > 12V output from a possible 12V input? It's impossible. However with the > > > > error term regulation down to 11.4V is acceptable. > > > > > > > No is not imposible. How is possible to have regulated +5V from +3V ? > > > > But it's a different problem. I'm not trying to generate a higher voltage. > > If I were then I'd be looking at a boost converter. > > > > But neither of these are the case. I can pretty much guarantee that the > > input voltage will be at or above 12V. I am willing to have a regulated > > voltage that is below that input voltage. > > You have much heavily design condition that you imagine, Byron: > > input voltage from 11 to 16 V At 11V the gel cell is dead anyway. The minimum input voltage is 12V. I agree on the 16V. > input spikes from 0 to hundred of volts That's a seperate issue that has been addressed in many forums. A combination of a fuse, MOV/TVS, and a shunt regulator can handle the spike issues. > ambient temperature from -35 to +70 C ( sorry for Farenheit speakers ) The unit is in the cabin. Ambient is more like -5 to 40 C. Hasn't been a problem up to now. > plenty of vibrations I haven't succeeded in getting the unit to glitch yet. > > You can't do it with a linear regulator ( or maybe yes with some > malfunctions or choosing right the output voltage at 9 or better 8V). I think the gel cell UPS battery eliminates a lot of the environmental issues you raise. It creates a high power, low impeadance 12V source that can be depended on. Isolation from the main battery will prevent the voltage dips you refer to. > A simple isolated DC-DC > converter will solve all your headache regarding isolating spikes. The > same one will gave you +5V and any other voltage ( including 5000V for > ionisation of the habitacle etc. ) I may agree with that assessment if one can show a reasonable discrete design. But those designs always end up being "well you need a Max something or other or a linear tech part, plus some specialized flyback transformer that has to be hand wound using a core that you can get from a scrap microwave..." They end up being complex, laden with specialized, expensive, hard to get parts. A PNP transistor, even a low saturation one, isn't a specialized hard to get part. > By the other hand if this is only for supplying the motherboard, HDD > and CDrom, then > probably you know there are boards which are running well at +8V instead > of +12 and without -12 and/or -5V. Also newest CDroms doesn't need +12V > at all, only +5V. The only problem I have with this approach is that you have to specify some specific component. I'd prefer a supply that works for all standard MB, HD, CDROMs, and ZIPs. And certainly if they work with 8V, then 11.4V should be absolutely no problem. BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email EraseMElistservmitvma.mit.edu with SET PICList DIGEST in the body

On Thu, Jan 10, 2002 at 08:39:21PM +1100, Roman Black wrote: {Quote hidden}> Byron A Jeff wrote: > > > > On Thu, Jan 10, 2002 at 08:44:58AM +1100, Roman Black wrote: > > > because of the storage capacity of the > > > battery you can regulate the output voltage up > > > to the PEAK input voltage. Not the MINIMUM > > > input voltage like a normal series regulator. > > > > Thanks! It was the solution to the starter dropout voltage when I was driving > > this equipment with a laptop power supply I got surplus from Marlon P Jones. > > Unfortunately they don't carry it anymore and I succeeded in shorting out > > the two rather fragile units I had. > > > > > You should aim for your battery to be kept at > > > about 12.8v, and full, so it can be charged > > > anytime the car is above 13.0v with a simple > > > series regulator transistor. The only times the > > > car will be under 13.0v will be when you start > > > it, or maybe idling with the headlights on. Anytime > > > the motor is "driving" the car will produce about > > > 14.2v to 14.5v. > > > > Correct. But the problem is getting the 12V line to regulate when only the > > UPS gel cell is on line. While it'll float at 12.8V to 13V, it quickly heads > > towards 12V when put under load. > > No drama there, low dropout series regs are not that > hard to do with a decent low-sat PNP device and a > 7805 as a zener. Now we're getting somewhere. This is the issue I want to address. Please take a stab at explaining the controller using the 7805. I know that it's possible to use the GND pin as an adjust pin. However I just don't see how to get over the dropout voltage issue. I just took a look at the National Semi LM340 datasheet, and even with an output current of 0A it starts to drop out about 1.5V above the regulation voltage. So how can it be used as a control element when Vin is in the dropout region? > > You need to determine your max current and typical > current for the 12v rail, if it's only 2A or so this > will be an easy task. If it's 4A continuous the task > gets slightly harder. That's measuable. If I get a chance I'll do it today. {Quote hidden}> > > I'm trying to ensure that the drives will not brown out during starting and > > when the car has short stops in the 5 to 10 minute range. > > That should not be a problem, most modern floppies > i've seen either don't use the 12v anymore, (and I > strip 5v motors from them) or the 12v motors are > 150mA or so. Even a floppy starting up will not draw > more than 0.5A from your 12v supply. Pull the cover > off your PC and stick an amp meter on the 12v rail > and check it out. Design from the back forward. :o) Agreed on that. I'm not planning on having a floppy but I'll test the draw on the hard disk and internal Iomega ZIP I plan to use. > > > The nominal system input voltage will be 13.8V. It should provide enough > > headroom for normal use. > > Not so. 13.8 is an "on charge" voltage. The moment > the car stops, your 13.8 sinks quickly back to a > typical battery "full" voltage of about 12.8v. That's true. But under load it quickly heads towards 12V. {Quote hidden}> > > It's the 12V minimum input voltage that I'm trying to attack. > > I don't think it's a problem. Assuming 12.8v in your > gell battery when the car stops, and 2A used at 12v, > you need a low dropout regulator good for 2A at 0.3v > drop, and you have a headroom of 0.8v-0.3v=0.5v. > So your 12.8v can drop to 12.3v before the 12v out > is affected. I would maybe choose a regulated 11.6v > rail, as this gives you another 0.4v headroom, and > a large enough battery to supply you for 10 minutes > at 2A and not drop 0.9v. And with a low saturation PNP, it should be possible to get in the 0.2V range. But how do I control the silly thing? One article I was reading last night: http://www.msdmag.com/1999/02/9902feat4.htm discusses using an opamp as a control element. But while it discussed the theory of operation, it didn't really give enough detail to build a working prototype. This is where I start to lose understanding. I understand transistors as swiches. But I'm still learning about transistors operating in the linear region and how to generate appropriate feedback. Let start with something concrete that we can discuss: A simple zener regulator: Vin -----C E--------Load | B | | | | +-R-+ | | | Z | | | GND GND It's operation is simple. As Vin rises, the NPN transistor starts to conduct passing current to the Load. Once Vin reaches the breakdown knee of the zener the zener starts to conduct, which draws current away from the base of the transistor. The last thing is that there is one CE diode drop across the transistor, so the voltage from the emitter can be no more than 0.6V than Vin. Lastly because the transistor will not conduct until the BE junction differential is 0.6V, the zener must 0.6V higher than the regulation voltage. So if I wanted 12V I'd use a 12.6V zener. That's my understanding of this. Please feel free to correct any errors or omissions I have made. The only problem with this perfectly simple circuit is that it requires 1.2V of headroom to regulate properly. Which means that Vin must be 13.2V or higher to regulate to 12V. Now switch the transistor to a PNP. Vin -----E C--------Load B | | | GND GND Now this isn't regulated. However I want to use it to illustrate how a LDO regulator can be built. This PNP transistor is turned fully on, it's in full saturation. The load sees a voltage that's Vin - Vce (sat), which is the saturation voltage of the PNP. This creates a low dropout when the Vce (sat) is low. PNP with saturation voltages of 0.2V are easily obtainable. So that takes care of the dropout region of the regulator, If the input voltage gets too low, simply ground the base and pass everything to the load. I fall down understanding the control element for normal regulation. Another picture: Vin -----E C--------Load B | | | ??? GND | GND What goes in the mystery box? An opamp? A zener? Roman mentioned a 7805. How would it fit in? That's the gap in my understanding. And it's compounded by the fact that whatever goes in the mystery box must operate like the first circuit when Vin falls into the dropout region. > > Remember this will be the 12v battery only, if you > need a 5v supply to run the PC at some 20A or more, > that will involve a whole 'nother battery and PSU... Actually the normal draw on the 5V line is about 4A. I measured it during testing of Russell's 2 transistor SMPS. BTW the MB would consistently boot from that power supply. BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservEraseMEEraseMEmitvma.mit.edu with SET PICList DIGEST in the body

Byron A Jeff wrote: > Now we're getting somewhere. This is the issue I want to address. Please > take a stab at explaining the controller using the 7805. I know that it's > possible to use the GND pin as an adjust pin. However I just don't see how > to get over the dropout voltage issue. I just took a look at the National > Semi LM340 datasheet, and even with an output current of 0A it starts to > drop out about 1.5V above the regulation voltage. So how can it be used as a > control element when Vin is in the dropout region? Wow! That was a long post. :o) Here's my views on building a low dropout series regulator as simple as possible. First thing is separate the thing into two parts: * power (obviously a low sat power device) * control ONLY the power device needs to be low dropout. The control circuit can be anywhere, and you put it in the best place, referenced to the zero rail. Low sat PNP Vin -------------------------E C----------- Vout (13.8v) | | B | (Reg 12v) | R | | | | | | | |-----| R1 | | | | 7805 C | |--I O---------B | G E-----| | NPN | | R2 | | | | Gnd --------------------------------------- Our power stage is just a low sat PNP like a BD204 or better part. The entire control system is referenced to Gnd, and a 7805 is used as a cheap easy precision zener, giving exactly 5v no matter what the input voltage does. R1 and R2 are a voltage divider tied to Vout. They set the NPN emitter at 5v-0.6 =4.4v whenever the output is at the correct voltage. Any less than the correct Vout, and the transistors will turn on as hard as needed to cause regulation or saturation, whatever comes first. Make sure the R1/R2 divider passes more than 10x the max expected PNP base current to keep regulation fairly accurate. If you keep this current substantially higher than the PNP base current the regulation will be close to perfect, at all voltages right down to saturation. -Roman PS. It will work just as good with a P type FET as the power part, if Vin is 12v or higher. -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservspam_OUTKILLspammitvma.mit.edu with SET PICList DIGEST in the body

(sorry re-posted to fix the ascii circuit) Byron A Jeff wrote: > Now we're getting somewhere. This is the issue I want to address. Please > take a stab at explaining the controller using the 7805. I know that it's > possible to use the GND pin as an adjust pin. However I just don't see how > to get over the dropout voltage issue. I just took a look at the National > Semi LM340 datasheet, and even with an output current of 0A it starts to > drop out about 1.5V above the regulation voltage. So how can it be used as a > control element when Vin is in the dropout region? Wow! That was a long post. :o) Here's my views on building a low dropout series regulator as simple as possible. First thing is separate the thing into two parts: * power (obviously a low sat power device) * control ONLY the power device needs to be low dropout. The control circuit can be anywhere, and you put it in the best place, referenced to the zero rail. Low sat PNP Vin -------------------------E C--------- Vout (13.8v) | | B | (Reg 12v) | R | | | | | | | |-----| R1 | | | | 7805 C | |--I O---------B | G E-----| | NPN | | R2 | | | | Gnd ------------------------------------- Our power stage is just a low sat PNP like a BD204 or better part. The entire control system is referenced to Gnd, and a 7805 is used as a cheap easy precision zener, giving exactly 5v no matter what the input voltage does. R1 and R2 are a voltage divider tied to Vout. They set the NPN emitter at 5v-0.6 =4.4v whenever the output is at the correct voltage. Any less than the correct Vout, and the transistors will turn on as hard as needed to cause regulation or saturation, whatever comes first. Make sure the R1/R2 divider passes more than 10x the max expected PNP base current to keep regulation fairly accurate. If you keep this current substantially higher than the PNP base current the regulation will be close to perfect, at all voltages right down to saturation. -Roman PS. It will work just as good with a P type FET as the power part, if Vin is 12v or higher. -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservTakeThisOuTspammitvma.mit.edu with SET PICList DIGEST in the body

On Fri, Jan 11, 2002 at 05:03:04AM +1100, Roman Black wrote: {Quote hidden}> Byron A Jeff wrote: > > > Now we're getting somewhere. This is the issue I want to address. Please > > take a stab at explaining the controller using the 7805. I know that it's > > possible to use the GND pin as an adjust pin. However I just don't see how > > to get over the dropout voltage issue. I just took a look at the National > > Semi LM340 datasheet, and even with an output current of 0A it starts to > > drop out about 1.5V above the regulation voltage. So how can it be used as a > > control element when Vin is in the dropout region? > > Wow! That was a long post. :o) > Here's my views on building a low dropout series > regulator as simple as possible. > > First thing is separate the thing into two parts: > * power (obviously a low sat power device) > * control > > ONLY the power device needs to be low dropout. The > control circuit can be anywhere, and you put it in > the best place, referenced to the zero rail. > > > Low sat PNP > Vin -------------------------E C----------- > Vout > (13.8v) | | B | (Reg 12v) > | R | | > | | | | > | |-----| R1 > | | | > | 7805 C | > |--I O---------B | > G E-----| > | NPN | > | R2 > | | > | | > Gnd > --------------------------------------- > > > Our power stage is just a low sat PNP like a BD204 > or better part. The entire control system is referenced > to Gnd, and a 7805 is used as a cheap easy precision > zener, giving exactly 5v no matter what the input > voltage does. > > R1 and R2 are a voltage divider tied to Vout. They > set the NPN emitter at 5v-0.6 =4.4v whenever the > output is at the correct voltage. Any less than the > correct Vout, and the transistors will turn on as hard > as needed to cause regulation or saturation, whatever > comes first. So that's the "magic" formula!! The R1/R2 ratio is just like the divider on an LM317. So if I calculate R1/R2 so that the midpoint gives 4.4V when the PNP collector is 12V. The transistors will lock in to put the PNP collector to 12V. Young grasshoppa finally understands. Thanks. So I did a quick calc using Vo = (R2/(R1+R2))*Vin with 12.0 for Vin and 4.4 for Vout and got a ratio of R1 = 1.73*R2. > > Make sure the R1/R2 divider passes more than 10x > the max expected PNP base current to keep regulation > fairly accurate. If you keep this current substantially > higher than the PNP base current the regulation will > be close to perfect, at all voltages right down to > saturation. And only at the cost of the base current which is the one item I have no concern over in this circuit. And just to verify the base current will be 10 * (Iout/beta) right? So let's put it all together. I went poking around the Philips NTE/ECG replacements parts site, because there are several Atlanta Distributors. I picked the NTE2314. 15A max, Vce(sat) of 0.25 typ, 0.5 max at 8A draw. Min beta of 30 at 8A. R2 would need to draw 10 * (2/30) -> 0.6A of current @ 4.4V. That works out to 7 ohms. If I use R1=12 ohms and R2=7 ohms it'll set the output voltage to 11.92V. Only one missing piece: what's the purpose of the emitter to base resistor on the PNP. A pullup to somehow force complete cutoff of the transistor? What value might it need to be? Another question: can R1/R2 be fed into a voltage follower? At that low a value the resistors will burn more than a half amp all the time. > -Roman > > PS. It will work just as good with a P type FET > as the power part, if Vin is 12v or higher. Then the Rds determines the dropout. This is a winner! I'll start working on it immediately. Thanks for all your help. BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email EraseMElistservspamspamBeGonemitvma.mit.edu with SET PICList DIGEST in the body

Hi Byron, re your posted comments below about the parts values, my simple circuit is just an example, and not really designed for 8 amps! It was designed for absolute simplicity, not as a working model for your high power needs. ;o) If you are expecting 2A as in your calcs I would suggest some simple circuit changes. Some caps in the right spots, a resistor network which doesn't need as much current, some thought about bad things like overcurrent and spike protection, etc. It's probably going to cost another transistor to fix the resistor network issue and a few discrete parts to get this circuit looking realistic at 2A. I do think your calcs are very conservative, i've been using low sat transistors a bit lately and would suggest a 15A transistor beta at 2A will saturate quite nicely at about 60-80mA. I do advise testing the sat voltage yourself because it varies from type to type and some are better than others. Is there any reason your ruled out the FET?? At 2A the FET sat volts will be very low, and at 12v in and 7805 5v you still get close to 12-4.4 =7.6v to turn on your p channel FET. Then your resistor ladder will be quite cool. ;o) Then you just need 2 of these regulators, one from the vehicle battery to your gell cell, keeping it charged to optimum, and the other regulator from gell cell to the PC 12v rail. -Roman Byron A Jeff wrote: {Quote hidden}> > On Fri, Jan 11, 2002 at 05:03:04AM +1100, Roman Black wrote: > > Byron A Jeff wrote: > > > > > Now we're getting somewhere. This is the issue I want to address. Please > > > take a stab at explaining the controller using the 7805. I know that it's > > > possible to use the GND pin as an adjust pin. However I just don't see how > > > to get over the dropout voltage issue. I just took a look at the National > > > Semi LM340 datasheet, and even with an output current of 0A it starts to > > > drop out about 1.5V above the regulation voltage. So how can it be used as a > > > control element when Vin is in the dropout region? > > > > Wow! That was a long post. :o) > > Here's my views on building a low dropout series > > regulator as simple as possible. > > > > First thing is separate the thing into two parts: > > * power (obviously a low sat power device) > > * control > > > > ONLY the power device needs to be low dropout. The > > control circuit can be anywhere, and you put it in > > the best place, referenced to the zero rail. > > > > > > Low sat PNP > > Vin -------------------------E C----------- > > Vout > > (13.8v) | | B | (Reg 12v) > > | R | | > > | | | | > > | |-----| R1 > > | | | > > | 7805 C | > > |--I O---------B | > > G E-----| > > | NPN | > > | R2 > > | | > > | | > > Gnd > > --------------------------------------- > > > > > > Our power stage is just a low sat PNP like a BD204 > > or better part. The entire control system is referenced > > to Gnd, and a 7805 is used as a cheap easy precision > > zener, giving exactly 5v no matter what the input > > voltage does. > > > > R1 and R2 are a voltage divider tied to Vout. They > > set the NPN emitter at 5v-0.6 =4.4v whenever the > > output is at the correct voltage. Any less than the > > correct Vout, and the transistors will turn on as hard > > as needed to cause regulation or saturation, whatever > > comes first. > > So that's the "magic" formula!! The R1/R2 ratio is just like the divider on > an LM317. So if I calculate R1/R2 so that the midpoint gives 4.4V when the > PNP collector is 12V. The transistors will lock in to put the PNP collector > to 12V. > > Young grasshoppa finally understands. Thanks. > > So I did a quick calc using Vo = (R2/(R1+R2))*Vin with 12.0 for Vin and 4.4 > for Vout and got a ratio of R1 = 1.73*R2. > > > > > Make sure the R1/R2 divider passes more than 10x > > the max expected PNP base current to keep regulation > > fairly accurate. If you keep this current substantially > > higher than the PNP base current the regulation will > > be close to perfect, at all voltages right down to > > saturation. > > And only at the cost of the base current which is the one item I have no > concern over in this circuit. And just to verify the base current will be > 10 * (Iout/beta) right? > > So let's put it all together. I went poking around the Philips NTE/ECG > replacements parts site, because there are several Atlanta Distributors. > I picked the NTE2314. 15A max, Vce(sat) of 0.25 typ, 0.5 max at 8A draw. > Min beta of 30 at 8A. > > R2 would need to draw 10 * (2/30) -> 0.6A of current @ 4.4V. That works out > to 7 ohms. If I use R1=12 ohms and R2=7 ohms it'll set the output voltage > to 11.92V. > > Only one missing piece: what's the purpose of the emitter to base resistor > on the PNP. A pullup to somehow force complete cutoff of the transistor? > What value might it need to be? > > Another question: can R1/R2 be fed into a voltage follower? At that low a > value the resistors will burn more than a half amp all the time. > > > -Roman > > > > PS. It will work just as good with a P type FET > > as the power part, if Vin is 12v or higher. > > Then the Rds determines the dropout. > > This is a winner! I'll start working on it immediately. > > Thanks for all your help. > > BAJ > > -- > http://www.piclist.com#nomail Going offline? Don't AutoReply us! > email RemoveMElistservKILLspammitvma.mit.edu with SET PICList DIGEST in the body -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservSTOPspamspam_OUTmitvma.mit.edu with SET PICList DIGEST in the body

On Fri, Jan 11, 2002 at 07:37:19AM +1100, Roman Black wrote: > Hi Byron, re your posted comments below about the > parts values, my simple circuit is just an example, > and not really designed for 8 amps! I wasn't planning on 8A! I just pulled parameters from the data sheet. > > It was designed for absolute simplicity, not as > a working model for your high power needs. ;o) > > If you are expecting 2A as in your calcs I would > suggest some simple circuit changes. Some caps > in the right spots, a resistor network which doesn't > need as much current, some thought about bad things > like overcurrent and spike protection, etc. spike is a separate issue. I was thinking that a shunt regulator would be appropriate. As for overcurrent all that's required is a sense resistor that trips a switch when the voltage drop gets too high across the resistor. Right? > > It's probably going to cost another transistor > to fix the resistor network issue and a few discrete > parts to get this circuit looking realistic at 2A. But the details are a significant problem when designing blind. The one part I did get was a high valued tantalum (25V. I hope that's high enough) for the output resistor. I read that a high valued low ESR cap was needed to prevent the circuit from oscillating. > > I do think your calcs are very conservative, i've > been using low sat transistors a bit lately and > would suggest a 15A transistor beta at 2A will > saturate quite nicely at about 60-80mA. I do advise > testing the sat voltage yourself because it varies > from type to type and some are better than others. And I presume I do this simply by grounding the base and drawing 2A through the collector? Or do I use a resistor on the base to limit the amount of current drawn? > > Is there any reason your ruled out the FET?? Nope. In fact I have a IRF 9540 that should do quite nicely. 0.16 RoDS. 17A part. I'll use it instead. > At 2A the FET sat volts will be very low, and at > 12v in and 7805 5v you still get close to 12-4.4 > =7.6v to turn on your p channel FET. Then your > resistor ladder will be quite cool. ;o) OK. Jut making sure I understand. The 7805, NPN, and resistor ladder are completely unchanged right? However because the FET is voltage controlled instead of current controlled it's not necessary to sink a bunch of current from the FET's gate. The only mitigating factor is the FET's gate capacitance which determines how fast that you can "move the gate". However I'm now have total cluelessness as to how to calculate how much current is required at the gate to get everything to work. Any rule of thumb for that? BTW since a lower gate voltage turns the FET on more fully, is there any reason not to use a lower reference voltage. With an LM317 you could get down to below 2V. While the LM317 would drop more voltage. It's not going to be drawing a significant amount of current. > > Then you just need 2 of these regulators, one from > the vehicle battery to your gell cell, keeping it > charged to optimum, and the other regulator from > gell cell to the PC 12v rail. And both can share the same voltage reference saving a part. BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email EraseMElistservEraseMEmitvma.mit.edu with SET PICList DIGEST in the body

On Fri, Jan 11, 2002 at 12:53:06AM +0200, Peter L. Peres wrote: > Imho build a switcher for all the voltages starting with (10.4V - wiring > losses) for minimum input voltage. You want to aim for 100W-ish probably, > which will run RMS currents of 10+A in the supply side, thus the note on > the wiring losses. If the major power drain is on 5V you could make a buck > switcher to make 5V and feed a boost switcher off of its output to make > 12V (and prolly -12V - are you using the serial ports on board ? ;-). I tried this tack before. I ended up with $60 of simple switcher parts that never exactly worked right. The 12V was a buck boost converter pair. I'm not feeling it. The 10.4V is totally unnecessary. The gel provides sufficient voltage level so that a 12V minimum can be presumed. > > I once made a SMPSU prototype to feed a PC main board (386) using a TL494 > and a transformer (rewound 50W halogen lamp transformer at 800Hz). It kind > of worked. The 800Hz hum is plesant ;-). > > Note that 7Ah gel batteries are not designed to supply 10A for any length > of time and it may get hot enough to cause trouble in the 5-10 minutes it > will last like this. Generally it's only online by itself for very short periods of time. Only during starting. Also the main car battery is backing it up. In the end I'm going to have a PIC based display and control system that will shut down the PC a short period of time after the engine is off. Like I said in my other post, I had a working system using a laptop power supply. I'd still be using it if they weren't so fragile. BAJ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email @spam@listserv@spam@spam_OUTmitvma.mit.edu with SET PICList DIGEST in the body

Roman, the nice structure you've post here can be replaced with 4 pin low-drop ( typical 0.25, max 0.5V) liner regulator KA78R12 from Motorola, in TO220 package, at current lower than 1A. Jeff, I'm realy interested about your experiment results at 8A with this schematic/structure. regards, Vasile On Fri, 11 Jan 2002, Roman Black wrote: {Quote hidden}> Byron A Jeff wrote: > > > Now we're getting somewhere. This is the issue I want to address. Please > > take a stab at explaining the controller using the 7805. I know that it's > > possible to use the GND pin as an adjust pin. However I just don't see how > > to get over the dropout voltage issue. I just took a look at the National > > Semi LM340 datasheet, and even with an output current of 0A it starts to > > drop out about 1.5V above the regulation voltage. So how can it be used as a > > control element when Vin is in the dropout region? > > Wow! That was a long post. :o) > Here's my views on building a low dropout series > regulator as simple as possible. > > First thing is separate the thing into two parts: > * power (obviously a low sat power device) > * control > > ONLY the power device needs to be low dropout. The > control circuit can be anywhere, and you put it in > the best place, referenced to the zero rail. > > > Low sat PNP > Vin -------------------------E C----------- > Vout > (13.8v) | | B | (Reg 12v) > | R | | > | | | | > | |-----| R1 > | | | > | 7805 C | > |--I O---------B | > G E-----| > | NPN | > | R2 > | | > | | > Gnd > --------------------------------------- > > > Our power stage is just a low sat PNP like a BD204 > or better part. The entire control system is referenced > to Gnd, and a 7805 is used as a cheap easy precision > zener, giving exactly 5v no matter what the input > voltage does. > > R1 and R2 are a voltage divider tied to Vout. They > set the NPN emitter at 5v-0.6 =4.4v whenever the > output is at the correct voltage. Any less than the > correct Vout, and the transistors will turn on as hard > as needed to cause regulation or saturation, whatever > comes first. > > Make sure the R1/R2 divider passes more than 10x > the max expected PNP base current to keep regulation > fairly accurate. If you keep this current substantially > higher than the PNP base current the regulation will > be close to perfect, at all voltages right down to > saturation. > -Roman > > PS. It will work just as good with a P type FET > as the power part, if Vin is 12v or higher. > > -- > http://www.piclist.com#nomail Going offline? Don't AutoReply us! > email TakeThisOuTlistservKILLspamspammitvma.mit.edu with SET PICList DIGEST in the body > > -- http://www.piclist.com hint: To leave the PICList .....piclist-unsubscribe-requestRemoveMEmitvma.mit.edu

> > > Actually the normal draw on the 5V line is about 4A. I measured it during > > > testing of Russell's 2 transistor SMPS. BTW the MB would consistently boot > > > from that power supply. > > > > I'd like to see the circuit you built, i'm curious > > about the regulation and also the type and performance > > of the inductor. > > Let me know what you want me to measure. > > The inductor is a J.W. Miller high current toroid pushed from RadioShack.com > IIRC it's a 50 mH 10A inductor. What would be interesting would be 1 Output voltage at a number of combinations of input voltage and output current. 2 It would also be interesting to know input current as well in each case. 3 For the really keen, the performance under sudden steps in load and input voltage would be of interest. 1. is best presented as a table with Vin and Iout as table axes and Vout for each combination. I found my original design to have amazingly good Vout regulation with Vin and Iout CONSIDERING HOW SIMPLE THE CIRCUIT IS. I "shout" this as I don't want people to think I am saying it is a high performance design in absolute terms compared with sophisticated IC based designs - just amazingly good for what it is. For example, *from memory*, with Vin ranging from about 12 volts to 200 volts I was getting Vout of better than +/- 0.1 volt. Can't remember regulation with load but can always measure with a working design or dig up notebooks if of great interest. Byron's figures would be more apposite in the present case. 2. above allows efficiency to be determined. 3. - transient performance (scope or similar needed) allows you to determine if there are any nasty surprises lurking. An apparently well performed design may provide a voltage spike on step removal of load (or on application!) or Vout may drop substantially during load changes. regards, Russell McMahon -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestEraseMEmitvma.mit.edu

I second that! I too would really like to see how a simple 2-tran circuit handles the 4A load. -Roman Russell McMahon wrote: {Quote hidden}> > > > > Actually the normal draw on the 5V line is about 4A. I measured it > during > > > > testing of Russell's 2 transistor SMPS. BTW the MB would consistently > boot > > > > from that power supply. > > > > > > I'd like to see the circuit you built, i'm curious > > > about the regulation and also the type and performance > > > of the inductor. > > > > Let me know what you want me to measure. > > > > The inductor is a J.W. Miller high current toroid pushed from > RadioShack.com > > IIRC it's a 50 mH 10A inductor. > > What would be interesting would be > > 1 Output voltage at a number of combinations of input voltage and output > current. > 2 It would also be interesting to know input current as well in each > case. > 3 For the really keen, the performance under sudden steps in load and > input voltage would be of interest. > > 1. is best presented as a table with Vin and Iout as table axes and Vout for > each combination. > > I found my original design to have amazingly good Vout regulation with Vin > and Iout CONSIDERING HOW SIMPLE THE CIRCUIT IS. I "shout" this as I don't > want people to think I am saying it is a high performance design in absolute > terms compared with sophisticated IC based designs - just amazingly good for > what it is. For example, *from memory*, with Vin ranging from about 12 volts > to 200 volts I was getting Vout of better than +/- 0.1 volt. Can't remember > regulation with load but can always measure with a working design or dig up > notebooks if of great interest. Byron's figures would be more apposite in > the present case. > > 2. above allows efficiency to be determined. > > 3. - transient performance (scope or similar needed) allows you to determine > if there are any nasty surprises lurking. An apparently well performed > design may provide a voltage spike on step removal of load (or on > application!) or Vout may drop substantially during load changes. > > regards, > > Russell McMahon -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

I'll come back to the 5V figures later. I now have a working 12V LDO regulator! It's a bit different than the one Roman posted (note. PASSWORDED site. user and password is 'piclist'): http://www.infosite.com/~jkeyzer/piclist/2002/Jan/0821.html Which I tried with no real success. But the principals now made sense to me (control at a lower voltage than dropout, use a resistor divider to sense the output voltage, and compare to a stable voltage reference) So I decided to get experimenting with opamps. I pieced together a 1458 based opamp circuit that regulates via a IRF 9530 PMOS pass transistor. Since us poor Linux guys can't use Roman's ASCII schematic tool (1/2 ;-) 1/2 ;-( ...) I'll just describe it. 1458 Opamp connections ---------------------- V+ : batt+ V- : batt GND In+: midpoint of a 470/270 voltage divider between Vout and batt GND In-: 4.3V Zener reference Out: To IRF 9530's gate. Pulled up by a 10K resistor to batt+ IRF 9530 P-channel MOSFET ------------------------- Source: Batt+ Drain: Vout (IMPORTANT: must be capactivly coupled to GND via low ESR cap) Gate: Connected to output of 1458 The cap and the gate pullup are both essential for proper regulation. My Vout is just about 11.7V, good enough for guvment work. The regulator maintains that voltage withing +-2 percent under initial testing. I tested with 3 different current draws: Iout Vin Vout ----------------------- 0A 13.00V 11.69V 0.92A 12.35V 11.66V 1.60A 12.00V 11.57V I'm extremely pleased with the initial results and really only want to get Vout bumped up a tenth or two and test in complete dropout. I'd love to see an ASCII schematic of the circuit if anyone gets a chance. BAJ -- http://www.piclist.com hint: To leave the PICList @spam@piclist-unsubscribe-requestRemoveMEEraseMEmitvma.mit.edu