Post by thread:[PIC:] PIC-based thermal airspeed meter

I would build a "quick and dirty" PIC-based thermal anemometer with minimal external parts, similar (but less complex) than Microchip TB044 "Sensing Air Flow with the PIC16C781". I need only a "rough" measure of the air flow (20% accuracy maybe acceptable) but a possible problem is the speed range, that in my case may vary from 1 to 90 ft/sec. and the highest value seems a bit outside of typical "hot-wire" anemometers applications. However, I would try this approach: - use of one 16F628 comparator (my PIC) - first comparator input connected to an external voltage divider with 2 NTC of same value. - second input of the comparator connected to the internal VRef. - one NTC is coupled with an heating resistor managed by another PIC pin, by a transistor. The measure is executed by heating the NTC (start of a timer or counter) and when the voltage exceed a defined threshold the timer/counter is read, disabling the heater. The second NTC (exposed to the airflow but not heated) is for temperature compensation. I will appreciate any suggestion on possible errors and/or a better approach to this problem. thanks MarcoG. ____________________________________________________________ Libero ADSL: navighi gratis a 1.2 Mega, senza canone e costi di attivazione. Abbonati subito su http://www.libero.it ____________________________________________

An NTC thermistor is a finicky little beast for several reasons. First, they are nonlinear, so somewhere in your program you have to have a table for this. If you have the table already out of an app note, it is no extra work. Most any sensor will need a lookup table or calibration of some kind. Second, cheap ones don't all measure the same, expensive ones do. Don't expect much accuracy out of a cheap thermistor. I have used these a lot, don't get me wrong, but they can be problematic and calibration can eat up a lot of time. For low accuracy applications it might work pretty well. Some hot wire anemometers use a nichrome loop, which is essentially an NTC device, and can be made quite hot. The idea is to regulate a constant current through it, then measure the voltage drop and therefore resistance at temperature. A drop in temperature means airflow is stealing heat. I remember when I read up on these that the greater the difference between your sensor temperature and the air temperature, the greater accuracy and low speed sensing you can achieve, but I can't quite remember the principle behind it. OK, you still need a lookup table. This is, of course, a shameless plug, as my daughter has a very small online business selling nichrome wire http://www.lile.biz This is in leiu of an allowance, and at 18 she's learned more about business than I had at 30. -- Lawrence Lile, P.E. Electrical and Electronic Solutions Project Solutions Companies http://www.projsolco.com > {Original Message removed}

>> First, they are nonlinear.. Yes, you are right. But one good feature of thermistors seems the sensitivity. When the airflow speed become high, it is necessary much more current to warm the "hot" sensor and if it is a thermistor, probably I can use less power than other detectors at the same sensitivity (?). I don't want to implement a lookup table but analyze the raw data on an Excel worksheet with the right formula: in this case I'm guessing if it is necessary to measure also the air temperature... Another solution, perhaps, may be to "linearize" the NTC in hw, but the sensitivity is much reduced >> Don't expect much accuracy out of a cheap thermistor. OK. I would use 2% tolerance thermistors, but they become not too "cheap"... However, since I have to made only 5-6 of these flowmeters, a more expensive NTC may be acceptable. >> The idea is to regulate a constant current through it, then measure the voltage >>drop and >> therefore resistance at temperature. Microchip TB044 seems to use a similar principle, but adapted to the peripherals of a PIC16C781. As I see, it is a bit complicated for a "quick and dirty" PIC solution. thanks for the reply, Lile MarcoG ____________________________________________________________ Libero ADSL: navighi gratis a 1.2 Mega, senza canone e costi di attivazione. Abbonati subito su http://www.libero.it ____________________________________________

> I would build a "quick and dirty" PIC-based thermal anemometer with > minimal external parts, similar (but less complex) than Microchip TB044 > "Sensing Air Flow with the PIC16C781". Hi Marco, Back in 1994 or '96 there was a clever design idea in Electronic Design Magazine for an anemometer. It had a pulse width or frequency output, I forget which. Some quick googling turned up a later article by the same author, Stephen Woodward. This version is several times more complicated in order to make it linear, which you could of course use the PIC to do. Perhaps you'll have some luck finding the original. It was very simple. If you can't find it, I have it somewhere and can scan it for you. Here's a link to the current version, with references at the end to the original simple version: http://www.reed-electronics.com/ednmag/contents/images/91902di.pdf I did build the original version, and found it slower than I needed. It uses transistors rather than a hot wire. Cheerful regards, Bob ____________________________________________

Bob Blick wrote: > This looks like the original: > > www.reed-electronics.com/ednmag/archives/1996/031496/06di3.htm > > -Bob Thanks Bob, this isn't exactly my desired solution but is seems simple and the output is a frequency, so easy to read with any PIC... >> I did build the original version, and found it slower than I needed. It >> uses transistors rather than a hot wire. How much slower..? A possible problem may be power consumption at highest speed (1watt is somewhat high, as the flowmeter circuit is for a battery-operating datalogger). I will try to build this layout on the breadboard and do some tests.. many thanks Bob! MarcoG ____________________________________________________________ Libero ADSL: navighi gratis a 1.2 Mega, senza canone e costi di attivazione. Abbonati subito su http://www.libero.it ____________________________________________

>>> I did build the original version, and found it slower than I needed. It >>> uses transistors rather than a hot wire. > > How much slower..? A possible problem may be power consumption at highest > speed (1watt is somewhat high, as the flowmeter circuit is for a > battery-operating datalogger). > I will try to build this layout on the breadboard and do some tests.. It took more than a second to settle. Since you don't need great accuracy, if you had an A/D converter I would suggest simply using a high temperature PTC thermistor, just measure the current through it. Adding compensation for temperature could make it very accurate over temperature. Cheerful regards, Bob ____________________________________________

marco.genovesi@libero.it writes: > > Bob Blick wrote: > > > This looks like the original: > > > > www.reed-electronics.com/ednmag/archives/1996/031496/06di3.htm > > > > -Bob > > > Thanks Bob, this isn't exactly my desired solution but is seems simple and the output is a frequency, so easy to read with any PIC... > I gave this circuit a 30 second look and the first thing I thought was... I wonder if the two transistors could both be simply driven via PWM off two PIC pins, dramatically reducing parts count and allowing automatic calibration of the system by balancing the setup while stationary. I really have no need for this myself, so I didn't give it much more thought than that. Comments? -p. ____________________________________________

>> > http://www.reed-electronics.com/ednmag/archives/1996/031496/06di3.htm > > I gave this circuit a 30 second look and the first thing I thought > was... I wonder if the two transistors could both be simply driven via > PWM off two PIC pins, dramatically reducing parts count and allowing > automatic calibration of the system by balancing the setup while > stationary. I really have no need for this myself, so I didn't give > it much more thought than that. Comments? Absolutely. I think using a comparator in the 16F628 would make it a natural match for the application. The original circuit takes quite a while to stabilize, and then a second or two to track changes. We'll have to let Marco decide, it's his application, and I don't know what that is. Cheerful regards, Bob ____________________________________________

Bob Blick writes: > Absolutely. I think using a comparator in the 16F628 would make it a > natural match for the application. > > The original circuit takes quite a while to stabilize, and then a second > or two to track changes. > > We'll have to let Marco decide, it's his application, and I don't know > what that is. Indeed! I've just started the process of teaching myself electronics/EE, and I've been learning a *lot* just by following the discussions here. When I saw the sample circuit, I thought that if I were ever to do something like that in a circuit where I was already using a microcontroller, it would seem to make a lot of sense to roll the complexity of the analog circuit right into the uC. When Marco implied that he was going to use that circuit more-or-less directly and then simply process the output, I just had to know if there was some reason why my idea wouldn't work, or there was a specific advantage to using the circuit presented directly. (My general tendancy is to assume that I know less than anyone else on this list...) -p. ____________________________________________

John Pearson wrote: >> I missed the first few posts on this subject so I have no idea what the goal is-... Sorry John for this late reply, below there is my original post. regards MarcoG ========================================= I would build a "quick and dirty" PIC-based thermal anemometer with minimal external parts, similar (but less complex) than Microchip TB044 "Sensing Air Flow with the PIC16C781". I need only a "rough" measure of the air flow (20% accuracy maybe acceptable) but a possible problem is the speed range, that in my case may vary from 1 to 90 ft/sec. and the highest value seems a bit outside of typical "hot-wire" anemometers applications. However, I would try this approach: - use of one 16F628 comparator (my PIC) - first comparator input connected to an external voltage divider with 2 NTC of same value. - second input of the comparator connected to the internal VRef. - one NTC is coupled with an heating resistor managed by another PIC pin, by a transistor. The measure is executed by heating the NTC (start of a timer or counter) and when the voltage exceed a defined threshold the timer/counter is read, disabling the heater. The second NTC (exposed to the airflow but not heated) is for temperature compensation. I will appreciate any suggestion on possible errors and/or a better approach to this problem. ========================================= ____________________________________________________________ Libero ADSL: navighi gratis a 1.2 Mega, senza canone e costi di attivazione. Abbonati subito su http://www.libero.it ____________________________________________

I haven't much experience with electronics so excuse me for possible errors in valuation the right approach: this is the the reason for which I have asked for suggestions :-) As I said in the original post, I would prefer a PIC-based solution with few external components. Yes, the two-transistor circuit don't match with my requirements, however seem to me an interesting idea: relatively simple, cheap and since the output is a frequency it can be easily interfaced also with micro without ad A/D. Actually I have only some 16F628 (the logger is based on it) and 12F629: these PICs don't have an AD input but only comparators, so my first idea was to use this internal peripherals for made a simple "hot wire" aenemometers. An important "requirement" for me is the low cost: I have to made some of these, that must be abandoned with the logger in some caves (not guarded) near the entrance. Therefore there is a probability that they are damaged or removed for vandalism: unfortunately this is possible in some areas. The whole cost of logger + anemometer doesn't have to exceed 10-15$, as this cost is only mine.. This is the reason for which I was finding for a "quick and dirty" solution. Current consumption may be a problem, but if I use "C" or "D" cells (instead of AA as actually) perhaps it may be tolerated. I'm still estimating which it is the minimal indispensable interval between the measures of airspeed. Regards MarcoG ---------- Initial Header ----------- >From : .....piclist-bouncesKILLspam@spam@mit.edu To : "Microcontroller discussion list - Public." piclistKILLspammit.edu Cc : Date : Tue, 26 Oct 2004 12:08:51 -0700 (PDT) Subject : Re: [PIC:] PIC-based thermal airspeed meter {Quote hidden}> > >> > www.reed-electronics.com/ednmag/archives/1996/031496/06di3.htm > > > > > I gave this circuit a 30 second look and the first thing I thought > > was... I wonder if the two transistors could both be simply driven via > > PWM off two PIC pins, dramatically reducing parts count and allowing > > automatic calibration of the system by balancing the setup while > > stationary. I really have no need for this myself, so I didn't give > > it much more thought than that. Comments? > > Absolutely. I think using a comparator in the 16F628 would make it a > natural match for the application. > > The original circuit takes quite a while to stabilize, and then a second > or two to track changes. > > We'll have to let Marco decide, it's his application, and I don't know > what that is. > > Cheerful regards, > > Bob > > > > ______________________________________________

Sorry Mauricio, I had forgotten your mail! Well, as I know ultrasound speed is a good choice for this kind of measures. In some places of my hard disk I must have some references related to precise airflow measurements in caves with this method. If I remember right, it is good also with low speed airflows. But as I have explained in reply to Bob, at the moment I would want a very low-cost circuit and I suspect that it is difficult to be in this limitation with ultrasounds. Or not? However, thanks for the suggestion. regards MarcoG ---------- Initial Header ----------- >From : .....piclist-bouncesKILLspam.....mit.edu To : "Microcontroller discussion list - Public." EraseMEpiclistspam_OUTTakeThisOuTmit.edu Cc : Date : Tue, 26 Oct 2004 16:43:50 -0300 Subject : RE: [PIC:] PIC-based thermal airspeed meter {Quote hidden}> What about ultrasound speed meters? I don't know about your app Marco, but > it might be useful... Perhaps someone can illustrate us on how difficult is > to implement this kind of measurement... > > Regards > > Mauricio Jancic > Janso Desarrollos > Microchip Consultant Program Member > (54) 11-4542-3519 > infospam_OUTjanso.com.ar > http://www.janso.com.ar ____________________________________________________________ Libero ADSL: navighi gratis a 1.2 Mega, senza canone e costi di attivazione. Abbonati subito su http://www.libero.it ____________________________________________

> As I said in the original post, I would prefer a PIC-based solution with > few external components. Yes, the two-transistor circuit don't match with > my requirements, however seem to me an interesting idea: relatively > simple, cheap and since the output is a frequency it can be easily > interfaced also with micro without ad A/D. > Actually I have only some 16F628 (the logger is based on it) and 12F629: > these PICs don't have an AD input but only comparators, so my first idea > was to use this internal peripherals for made a simple "hot wire" > aenemometers. Hi Marco, If you stripped the circuit down to just the fundamentals, what you have is two transistors, a few resistors, and a comparator (inside the PIC). Make a free running loop alternately comparing the two transistors, and heating one of them. Try to keep the hot transistor hot(preprogrammed with the low-value resistor on the reference transistor) by watching switching of the comparator, and keep track of the timing for your intermediate result. Compare that value with a lookup table and you have it. Using a SOT-23 package transistor (not mounted to the circuit board) you should be able to get faster response. You'd use three pins on the pic: two comparator inputs and another pin to drive the heated transistor. Cheerful regards, Bob ____________________________________________

Thanks Bob, this is a good input! I'm not so able to manage a SOT-23 component..., altough I have seen on the PicList many useful methods for this. At the moment I would prefer to start with a "normal" transistor and see if the results are acceptable for my use. regards MarcoG ---------- Initial Header ----------- {Quote hidden}> Hi Marco, > > If you stripped the circuit down to just the fundamentals, what you have > is two transistors, a few resistors, and a comparator (inside the PIC). > Make a free running loop alternately comparing the two transistors, and > heating one of them. Try to keep the hot transistor hot(preprogrammed with > the low-value resistor on the reference transistor) by watching switching > of the comparator, and keep track of the timing for your intermediate > result. Compare that value with a lookup table and you have it. Using a > SOT-23 package transistor (not mounted to the circuit board) you should be > able to get faster response. > > You'd use three pins on the pic: two comparator inputs and another pin to > drive the heated transistor. > > Cheerful regards, > > Bob > > > ______________________________________________

Yes, TO18 metal package can be an idea.. Thanks Jinx! MarcoG ---------- Initial Header ----------- >From : @spam@piclist-bouncesKILLspammit.edu To : "Microcontroller discussion list - Public." KILLspampiclistKILLspammit.edu Cc : Date : Fri, 29 Oct 2004 09:57:07 +1300 Subject : Re: [PIC:] PIC-based thermal airspeed meter > > I would prefer to start with a "normal" transistor and see if the results > > are acceptable for my use > > I wonder - if you got a metal can, eg TO18, and took the top off, > (like we used to do to make an opto-transistor) that would expose > the silicon. Not something to consider for production but thought it > would be an interesting experiment to test thermal response > > ______________________________________________

At 10:26 AM 10/26/2004 -0700, you wrote: > >>> I did build the original version, and found it slower than I needed. It > >>> uses transistors rather than a hot wire. > > > > How much slower..? A possible problem may be power consumption at highest > > speed (1watt is somewhat high, as the flowmeter circuit is for a > > battery-operating datalogger). > > I will try to build this layout on the breadboard and do some tests.. > >It took more than a second to settle. The key to getting fast response out of this kind of instrument is to servo the temperature of the sensing element to a constant and to measure the power going into the element. Kind of analogous to a photodiode amplifier where you want to maintain constant voltage across the high diode capacitance to keep it fast. Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" RemoveMEspeffTakeThisOuTinterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com ____________________________________________

Yes, adiabatic gradient temperature of the air varies with altitude by ABOUT 10 C. per 3000 feet (dry air) and 5-6 C. with humid air, that's the most common case. But where is the problem? Seems to me that the two transistor circuit has temperature compensation. Eventually, is the air density variation with altitude that can alter in some way the measures: it can be calculate, however. MarcoG ---------- Initial Header ----------- >From : spamBeGonepiclist-bouncesspamBeGonemit.edu To : "Microcontroller discussion list - Public." TakeThisOuTpiclistEraseMEspam_OUTmit.edu Cc : Date : Fri, 29 Oct 2004 09:03:54 -0500 Subject : Re: [PIC:] PIC-based thermal airspeed meter {Quote hidden}> PIC-based thermal airspeed meter > > You do know that temperature varies with altitude, > about 2 degrees C per a thousand feet? > > And that this is different, everyday, and also varies from > location to location, pretty much with the weather, season, > cloud cover, etc. > > If you make this work, you 'are' a god. > > Just remember, according to Rutan, when doing cutting edge > research, half the people think you are crazy. ____________________________________________________________ Libero ADSL: navighi gratis a 1.2 Mega, senza canone e costi di attivazione. Abbonati subito su http://www.libero.it ____________________________________________

Ward, David wrote: > The temperature gradient isn't a big problem as the proposed method > compares the temperature of two sensors, one of which is in the > airstream. > It's the cooling effect of the airstream that is being measured Yes, but this does vary with pressure. Think of the limiting case where you have a vacuum, which doesn't cool at all. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com ____________________________________________

----- Original Message ----- From: <RemoveMEmarco.genovesiTakeThisOuTlibero.it> To: "piclist" <piclistEraseME.....mit.edu> Sent: Friday, October 29, 2004 10:02 AM Subject: Re: [PIC:] PIC-based thermal airspeed meter {Quote hidden}> > > Yes, adiabatic gradient temperature of the air varies with altitude by > ABOUT 10 C. per 3000 feet (dry air) and 5-6 C. with humid air, that's the > most common case. But where is the problem? > Seems to me that the two transistor circuit has temperature compensation. > Eventually, is the air density variation with altitude that can alter in > some way the measures: it can be calculate, however. > > MarcoG > > I don't want to discourage you. If you have an idea for a new mouse trap, go for it. ____________________________________________

Marco, You know, with the number of units, cost limits, and the very low velocities you are measuring, I would try to keep it real simple and low tech. Perhaps a simple vane type meter is in order. A flap attached to a pot or strain gauge. Then a thermistor and an absolute pressure sensor to calculate dry air density for air corrections. Throw in a humidity sensor if you like complicated math equations. Perhaps you could adapt some old Bosh fuel injection flapper style mass air meters. Just a thought. John {Original Message removed}

Mechanical methods to measure airflow are to don't underrate, but in my (little) experience they easy fail on low flows (as rotating cups anemometer). I don't have tested something as that you suggest and may be that your solution can be efficient also at low airflow speed. However I think that isn't necessary to calculate air density by a pressure sensor, can be sufficient reading values from standard atmospheric tables and to make the corrections. Well, sorry John but I must interrupt here because now I must go out of office... Good weekend MarcoG ---------- Initial Header ----------- >From : EraseMEpiclist-bouncesmit.edu To : "Microcontroller discussion list - Public." RemoveMEpiclistEraseMEEraseMEmit.edu Cc : Date : Fri, 29 Oct 2004 09:11:39 -0700 Subject : Re: [PIC:] PIC-based thermal airspeed meter {Quote hidden}> Marco, > > You know, with the number of units, cost limits, and the very low velocities > you are measuring, I would try to keep it real simple and low tech. Perhaps > a simple vane type meter is in order. A flap attached to a pot or strain > gauge. Then a thermistor and an absolute pressure sensor to calculate dry > air density for air corrections. Throw in a humidity sensor if you like > complicated math equations. > > Perhaps you could adapt some old Bosh fuel injection flapper style mass air > meters. > > Just a thought. > John > ____________________________________________________________ Libero ADSL: navighi gratis a 1.2 Mega, senza canone e costi di attivazione. Abbonati subito su http://www.libero.it ____________________________________________

On Fri, 29 Oct 2004, Olin Lathrop wrote: > Ward, David wrote: >> The temperature gradient isn't a big problem as the proposed method >> compares the temperature of two sensors, one of which is in the >> airstream. >> It's the cooling effect of the airstream that is being measured > > Yes, but this does vary with pressure. Think of the limiting case where you > have a vacuum, which doesn't cool at all. Air heat capacity does not vary seriously with pressure until you have a serious vacuum. The thermal device should measure heat capacity per volume sampled actually, and this can be expressed as speed when corrected for pressure. Peter ____________________________________________