Post by thread:[PIC] PT100 4-20mA load

Designing temperature measurement with a PT100 sensor. I'm using an 18F4520, with a 5V Vref+, 250 ohm load and nominal 12V for the sensor. This is what I've done in the past to get a 1V - 5V i/p range from 4-20mA Some measurements I've taken so far indicate that the sensor output equates to 38mV/degC. Rather than use a 1/38 routine on the ADC result, I thought I'd just use a lower value resistor - which I reckon to be 210 ohms - and a simple 1/32 shift. Reading the datasheet for the sensor (a Status SEM203/P), it says "The load symbol represent any other device connected in the loop, such as Monitoring equipment, panel indicators and loop isolators. The load value can range from 0 ohms to the max loop load for given supply, refer to section 3, "Max load", for more information". Section 3 provides the equation Max Load = ((Vsupply - 10)/20) kohms By that calculation, (13.8 - 10)/20 = 190 ohms. If battery voltage was low perhaps, (12 - 10)/20 = 100 ohms I've not used a PT100 before, and haven't come across a resistor limit. Sensors I've used in the past have suggested 125, 250 or even 500 ohm loads. Is the above saying that results will suffer if a load resistor > 190 ohms or 100 ohms is used ? What I've measured so far (a little crudely, comparing against an LM335) seems to be quite linear with a 250 ohm load TIA

On 13 Jan 2009 at 22:37, Jinx wrote: <snip> > "The load symbol represent any other device connected in the loop, such > as Monitoring equipment, panel indicators and loop isolators. The load > value can range from 0 ohms to the max loop load for given supply, refer > to section 3, "Max load", for more information". Section 3 provides the > equation > > Max Load = ((Vsupply - 10)/20) kohms I read all your post but the following is still a bit of a guess, apologies if I'm coming at it from the wrong angle. What you normally worry about in a 4-20mA loop is the total resistance of all the devices in the loop. You may have for example, one 4- 20mA sensor connected to several 4-20mA measuring devices all connected in series, and one power supply. If the total resistance is too high the power supply wont be able to make the required amount of current flow. It's just Ohms law.. eg, with a 24V power supply you can't have more than 1200 Ohms total resistance and expect 20mA to flow (24V/20mA=1200Ohms). Dunno what the figures 10 and 20 represent in your equation above. In designing a 4-20mA device it makes sense to make it present the lowest resistance/voltage drop as possible, thereby minimising the amount of power it takes from the loop power supply and allowing as much headroom as possible so other devices can be placed in the loop. -- Brent Brown, Electronic Design Solutions 16 English Street, St Andrews, Hamilton 3200, New Zealand Ph: +64 7 849 0069 Fax: +64 7 849 0071 Cell: +64 27 433 4069 eMail: spam_OUTbrent.brownTakeThisOuTclear.net.nz

> What you normally worry about in a 4-20mA loop is the total > resistance of all the devices in the loop Yup, I'd agree with that. And once I settle on a load I'll see what happens with Vsupply variations > Dunno what the figures 10 and 20 represent in your equation Me neither. Possibly it relates to the conversion performed by the Status unit. Judging by the manual's instructions on setting it up with its one button and LED flashes that indicate which menu you're in, there's probably a micro inside > In designing a 4-20mA device it makes sense to make it present the > lowest resistance/voltage drop as possible As this is the only 4-20mA device on that line/analogue input, there'll be no consideration needed with regard to any other serial loads. I think I'll scale the load resistor to get the 32mV/degC I'm looking for, which is a convenient divisor for the ADC result. It's going to be in the region of 200 ohms, which is within the range (IME) of common 4-20mA loads. If the results are linear with a range of Vsupply and nothing is being harmed, that's pretty much a win isn't it ?

Jinx wrote: > Some measurements I've taken so far indicate that the sensor output > equates to 38mV/degC. I thought you said it was 4-20mA current output. If so, then the voltage is a function of whatever resistor you place in the current path. > Rather than use a 1/38 routine on the ADC This is a pet pieve of mine. Why are people so obsessed with units inside a PIC matching the totally arbitrary units people might be used to? This only makes some sense when the PIC is directly responsible for displaying values to a human, although then you're likely to have the math routines already needed so that conversion is trivial compared whatever else is going on. If you're sensing temperature and turning on a fan, a heater, or sounding a alarm at certain trip points, there is absolutely no reason for the measured temperature inside the PIC to be in degrees F, degrees C, or some "nice" round fraction of those. There is nothing "nice" about human-habituated units to the PIC. The nicest units in the PIC are usually whatever you get from a A/D reading, possibly after some shifting due to low pass filtering. Let it come out however it comes out and deal with that directly. If you need to set trip points in the source code in deg F or whatever, use assembly time math to convert that to the PIC units, not run time math or design compromises to convert the PIC units to it. > result, I thought I'd just use a lower value resistor - which I > reckon to be 210 ohms - and a simple 1/32 shift. If you're asking about the current sense resistor that generates the voltage from the 4-20mA current loop, then you can make it as small as you want. The only downside is resolution and noise floor on your end, the current loop doesn't care. The current source will have some volage compliance range. You have to make sure all the devices in the loop don't cause that to be exceeded at the maximum current, but the current loop itself will be just fine with a dead short. ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000.

On 13 Jan 2009 at 23:47, Jinx wrote: > > In designing a 4-20mA device it makes sense to make it present the > > lowest resistance/voltage drop as possible > > As this is the only 4-20mA device on that line/analogue input, there'll be > no consideration needed with regard to any other serial loads. I think I'll > scale the load resistor to get the 32mV/degC I'm looking for, which is a > convenient divisor for the ADC result. It's going to be in the region of 200 > ohms, which is within the range (IME) of common 4-20mA loads. If the > results are linear with a range of Vsupply and nothing is being harmed, > that's pretty much a win isn't it ? Yeah, I knew you were on to it. 200 ohms is 4V drop @ 20mA, hard to imagine that not being acceptable in a typical application. Nice when the scaling works out convenient for the maths, or can be made to be convenient without too much compromise.-- Brent Brown, Electronic Design Solutions 16 English Street, St Andrews, Hamilton 3200, New Zealand Ph: +64 7 849 0069 Fax: +64 7 849 0071 Cell: +64 27 433 4069 eMail: .....brent.brownKILLspam@spam@clear.net.nz

> > equates to 38mV/degC. > > I thought you said it was 4-20mA current output. If so, then the voltage > is a function of whatever resistor you place in the current path Yes, it is 4-20mA. So with a 250 ohm load resistor you get an ADC i/p range of 1V - 5V. However, this thing has been calibrated or set up, not by me, to output ~38mV/degC. Which is fine, although other sensors I use have a more 'user-friendly' relationship. For example a depth sensor with a 256 ohm load produces 1mV/mm at the ADC i/p, easy on the maths and conversions with a 5.120V Vref+ > Rather than use a 1/38 routine on the ADC > > This is a pet peeve of mine. Why are people so obsessed with units I could use a 1/38 routine, no big deal. I only asked about the resistor because the datasheet had raised issues I'd not encountered before. If a resistor that scales the ADC to 32mV/degC keeps the end result in spec there's nothing wrong with that is there ? > If you're asking about the current sense resistor that generates the voltage > from the 4-20mA current loop, then you can make it as small as you want. > The only downside is resolution and noise floor on your end, the current > loop doesn't care The PT100 is being used to log temperature. More accurate is better, so causes of error will need to be assessed - resistor, Vref, Vsupply, maths, etc. The datasheet claims 0.1degC which is OK, but +/- 0.5C in the end is acceptable

> -----Original Message----- > From: piclist-bouncesKILLspammit.edu [.....piclist-bouncesKILLspam.....mit.edu] On Behalf {Quote hidden}> Of Jinx > Sent: 30 January 2009 13:40 > To: Microcontroller discussion list - Public. > Subject: Re: [PIC] PT100 4-20mA load > > Just a follow-up > > The PT100 is in service and working well. The person organising > this job/trial asked for more temperature sensors for some general > comparitive measurements around the test area. Not wanting to > spend spend spend on PT100s, I made this inexpensive circuit up > to use LM335s > > home.clear.net.nz/pages/joecolquitt/0-10mA_remote_lm335.gif > > Range of interest for us is 5-30C. Voltage at PIC pin is within 0.5C > of expected, and within spec for the LM335 > > Any suggestions for improvement/cheapening or how you did it welcome Joe, You might want to check the behaviour of your current source when driving an inductive load, e.g. a long bit of wire. I don't see any compensation components, and it's very easy to make current sources oscillate under those conditions. Simulating it first in e.g. SwitcherCad might be a good first step to allow you to check gain and phase margins. Cheers Mike ======================================================================= This e-mail is intended for the person it is addressed to only. The information contained in it may be confidential and/or protected by law. If you are not the intended recipient of this message, you must not make any use of this information, or copy or show it to any person. Please contact us immediately to tell us that you have received this e-mail, and return the original to us. Any use, forwarding, printing or copying of this message is strictly prohibited. No part of this message can be considered a request for goods or services. =======================================================================

> You might want to check the behaviour of your current source when > driving an inductive load, e.g. a long bit of wire. I don't see any > compensation components, and it's very easy to make current sources > oscillate under those conditions. Mike, currently updating LTSpice (not used it for quite some time) and will get on to a simulation when able What sort of compensation components might you install ? I'm trying to imagine what sort of a load a wire pair would be. As I see it, not wishing to be obtuse, wouldn't the wire and a static (a temperature sensor probably won't be changing abruptly) and current be fairly stable ? There is some possibility of noise induction around. It's a semi-industrial setting - a couple of smallish (1HP) motors and lighting How about shielded twin-core cable ? Signal, V+ and ground (braid)

part 1 3299 bytes content-type:text/plain; (decoded quoted-printable) > -----Original Message----- > From: EraseMEpiclist-bouncesspam_OUTTakeThisOuTmit.edu [piclist-bouncesspam_OUTmit.edu] On Behalf {Quote hidden}> Of Jinx > Sent: 30 January 2009 23:31 > To: Microcontroller discussion list - Public. > Subject: Re: [PIC] PT100 4-20mA load > > > You might want to check the behaviour of your current source when > > driving an inductive load, e.g. a long bit of wire. I don't see any > > compensation components, and it's very easy to make current sources > > oscillate under those conditions. > > Mike, currently updating LTSpice (not used it for quite some time) and > will get on to a simulation when able > > What sort of compensation components might you install ? I'm trying to > imagine what sort of a load a wire pair would be. As I see it, not wishing > to be obtuse, wouldn't the wire and a static (a temperature sensor > probably > won't be changing abruptly) and current be fairly stable ? > Sorry for late reply Joe. The wire will present an inductive load. The closer the coupling between the two wires, the lower the inductance, but since coupling is never perfect there will be some residual inductance. The problem is that an inductance likes to maintain a fixed current, so effectively "fights" against your current source. You then end up with a phase shift which, if it reaches 180 degrees whilst the loop gain is greater than unity, will cause oscillation. This is independent of the rate of change of your temperature signal. You also have the added issue of having two cascaded op-amps which is often bad news anyway since each will give some phase shift, so the circuit can reach 180 degrees of phase rather more easily. I have attached the Switcher CAD files using the TL27L4 Spice model downloaded from TI, and you can see that even with no inductive load the simulator suggests the circuit is not stable. In the words of the guy I sit next to, "It's hooting like an owl". A real circuit may not actually do this, since simulations are never 100% accurate, but it does show that the design is probably marginal. If you add any significant inductance into the output the simulation fails to converge, which is nearly always a sign of instability. If you download "BodeCad" from LT at http://www.linear.com/designtools/software/ you can use the same schematic with a minor alteration to plot the frequency/phase response to see where your problem lies. The tutorial (pdf link on same page) shows you how to do this. Cheers Mike ======================================================================= This e-mail is intended for the person it is addressed to only. The information contained in it may be confidential and/or protected by law. If you are not the intended recipient of this message, you must not make any use of this information, or copy or show it to any person. Please contact us immediately to tell us that you have received this e-mail, and return the original to us. Any use, forwarding, printing or copying of this message is strictly prohibited. No part of this message can be considered a request for goods or services. ======================================================================= part 2 1195 bytes content-type:application/octet-stream; (decode) part 3 2232 bytes content-type:application/octet-stream; (decode) part 4 35 bytes content-type:text/plain; charset="us-ascii" (decoded 7bit)

Joe, PT100 does not need shielded cable even in automotive application. Why? Because the load is supplied in DC under a current generator. If you have any noise induced by the engine, a simple electrolytic capacitor will filter all the noise on the current to voltage converter side (aka a poor resistor in your situation). 2-10mA or 4-20mA has the lower current (2 or 4 mA) just for being able to distinguish a broken wire from a minimum current in the loop (that's why with 0-10mA you will never be able to know if the problem is with your RTD or with your circuit as long the output value for a given temperature will be near 0mA). best luck, Vasile On 2/4/09, Jinx <@spam@joecolquittKILLspamclear.net.nz> wrote: {Quote hidden}> Mike, thanks ever so for the reply and files > > I've now downloaded BodeCAD and the tutorial. I'm being pulled > from all directions today so won't be able to start my learning curve > until later this evening, if I'm lucky ( unlucky ;-) ) > > 20m of shielded stereo audio cable between the sensor board and > the PIC seems to be OK, at the measurement level anyway, and the > signal is clean on the 'scope. So at least there's something to work > with on the s/w side > > When I've a minute I could measure the inductance of that cable, > although until I get BodeCAD running I wouldn't know whether > its inductance is "significant" > > -

Hi Vasile > Joe, > > PT100 does not need shielded cable even in automotive application Perhaps you missed the diagram that Mike and I have been referring to http://home.clear.net.nz/pages/joecolquitt/0-10mA_remote_lm335.gif The PT100 is being used as the important sensor, the much cheaper LM335s are for more general temperature indicators The possible problem is op-amp instability > (that's why with 0-10mA you will never be able to know if the problem > is with your RTD or with your circuit as long the output value for a > given temperature will be near 0mA). That's true. The 0mA of a 4-20mA system would be used as error or fault detection. For this little short-lived project I could go with either 0-10mA, and accept the risk, or 2-10mA

On 2/4/09, Jinx <KILLspamjoecolquittKILLspamclear.net.nz> wrote: > Hi Vasile > > > Joe, > > > > PT100 does not need shielded cable even in automotive application > > Perhaps you missed the diagram that Mike and I have been referring to > > http://home.clear.net.nz/pages/joecolquitt/0-10mA_remote_lm335.gif The "sum" and " x10" could be one operational amplifier, summing and gain in the same time with a convenable reference, also the first repeater hasn't too much sense because the divider can be choosen 10 times lower than the R (=10k) of the "sum" if you'll use 100K instead of 10K. The reference comming from a 5V1 zenner can't be considered reference. Imagine that ambient temperature around 5V1 zenner is moving between -20 to +60C... tragedy. {Quote hidden}> > The PT100 is being used as the important sensor, the much cheaper > LM335s are for more general temperature indicators > > The possible problem is op-amp instability > > > (that's why with 0-10mA you will never be able to know if the problem > > is with your RTD or with your circuit as long the output value for a > > given temperature will be near 0mA). > > That's true. The 0mA of a 4-20mA system would be used as error or > fault detection. For this little short-lived project I could go with either > 0-10mA, and accept the risk, or 2-10mA > > -

> -----Original Message----- > From: RemoveMEpiclist-bouncesTakeThisOuTmit.edu [spamBeGonepiclist-bouncesspamBeGonemit.edu] On Behalf {Quote hidden}> Of Jinx > Sent: 05 February 2009 11:35 > To: Microcontroller discussion list - Public. > Subject: Re: [PIC] PT100 4-20mA load > > Mike, at last I've had a chance to get back to the project with that > LM335 circuit > > As you predicted, it oscillates. Length of wire to a 400R load is > irrelevant. It's 8kHz @ 1Vp-p centred on the DC produced across > the load > > I've had a go at it with LTspice but ...... I really don't know what I'm > doing yet. Time is a bit tight and a crash course in LTspice is wishful > thinking > > Hoping you can suggest what components and where to place them > > Looking past the noise at the DC level, the circuit appears it would > give good results, so it's worth fixing > > TIA Hi Joe, For a quick fix try a ~100n capacitor from the output of the first op-amp (the one that supplies current to the load) to the inverting input of the same amp. Cheers Mike ======================================================================= This e-mail is intended for the person it is addressed to only. The information contained in it may be confidential and/or protected by law. If you are not the intended recipient of this message, you must not make any use of this information, or copy or show it to any person. Please contact us immediately to tell us that you have received this e-mail, and return the original to us. Any use, forwarding, printing or copying of this message is strictly prohibited. No part of this message can be considered a request for goods or services. =======================================================================

> -----Original Message----- > From: TakeThisOuTpiclist-bouncesEraseMEspam_OUTmit.edu On Behalf Of Jinx > Sent: Thursday, February 05, 2009 6:35 AM > > Mike, at last I've had a chance to get back to the project with that > LM335 circuit > > As you predicted, it oscillates. Length of wire to a 400R load is > irrelevant. It's 8kHz @ 1Vp-p centred on the DC produced across > the load I noticed the circuit uses TLC27L series op-amps, like LM324s they tend to oscillate with even small capacitive loads (<200pF). This is easy to check for by using small capacitors as loads and watching for the oscillation. The cure is to de-couple the op-amp output from the load with a small series resistance. 10 to 100 ohms in series with the output will usually cure the problem and if you take the feedback from the other end of the resistor the voltage drop in the resistor won't affect the final output. Capacitive load problem information is usually buried deep down in the data sheet (page 32 of 37 in my old TLC27L4 data sheet). Paul Hutch {Quote hidden}> > I've had a go at it with LTspice but ...... I really don't > know what I'm > doing yet. Time is a bit tight and a crash course in LTspice > is wishful > thinking > > Hoping you can suggest what components and where to place them > > Looking past the noise at the DC level, the circuit appears it would > give good results, so it's worth fixing > > TIA

> For a quick fix try a ~100n capacitor from the output of the first > op-amp (the one that supplies current to the load) to the inverting > input of the same amp Thanks, that seems to have done the job. Currently have 6m of shielded stereo cable, no problems. There's just a few mV of noise on the DC which is insignificant as 0C will be ~ 340mV and 30C wil be ~ 2650mV. Most importantly the PIC is reporting the remote temperature correctly and stably, not wobbling 4 degrees !! Vasile mentioned that the Sum and x10 functions can be combined into one amp. I notice in the TLC27L4 d/s such a configuration so the next 3 sensors will be that arrangement. I'll update the previously posted gif as and when. And once this unit is finished I'll do some temperature tests on the board to check the effect on the IC, zener etc, see if anything needs changing to suit a wider ambient range. Which isn't necessary for this test, ~ 25-30C presently, but ambient for the sensors will fall below 0C in winter

> -----Original Message----- > From: RemoveMEpiclist-bouncesTakeThisOuTmit.edu On Behalf Of Jinx > Sent: Saturday, February 07, 2009 6:43 PM > > x > | > | > x ? > > http://home.clear.net.nz/pages/joecolquitt/0-10mA_remote_lm335.gif That is what I was thinking of _but_, now that I've looked more carefully at your circuit I think I was all wet in my original suggestion. When I first looked at the circuit for some reason I thought the output lead was directly connected to an op-amp output. Now I see that the op-amp output is de-coupled from the output wire with 147 ohms already. My best guess now, without actually building the circuit, is that noise originating from the zeners (especially the Vref zener) and/or output lead is the culprit. Any noise from the reference Zener will be divided by ~2, summed with the sensor signal then multiplied by 10 so, 5X the original noise would be in the output. Paul Hutch

I'm measuring and logging temperature over 4-20mA lines. There are two types of sensor, PT100 and LM335. Both are driving 250R loads at the PIC AD pins, distance is about 10m. I now need to relay the voltages at one PIC to another PIC, a further 10m away from the sensors Using this (including the 100n feedback capacitor between pins 1 and 2) http://www.piclist.com/techref/microchip/aout.htm configured for 4-20mA. Taking the voltage at PIC1, for example AN0, feed it into the above circuit to convert it to current, and this in turn drives the 250R on PIC2's AN0. However, the copying isn't perfect. There's a slight discrepancy. eg if the original is 1.650V (6.6mA) what arrives at PIC2 is about 25mV higher, 1.675V (6.7mA). This equates to 0.5 degC in this application, which is tolerable, but I would like to trim the circuit to remove that 25mV. I'm using the LM358, which has no offset pins, if that's relevant Any suggestions ? TIA

> -----Original Message----- > From: piclist-bouncesEraseME.....mit.edu [EraseMEpiclist-bouncesmit.edu] On Behalf {Quote hidden}> Of Jinx > Sent: 28 February 2009 03:39 > To: Microcontroller discussion list - Public. > Subject: Re: [PIC] PT100 4-20mA load > > Just a couple of notes re question, and a bit of thinking out loud > > I'm using the LM358 because that's what I picked out of the box. I've > many other single and dual op-amps > > Did some more comparisons and found that the differences at 5C, 20C > and 40C are all pretty close to 1.5%. So in theory I could use a divider > on the input of the voltage-current converter to knock it down to 98.5% > of the source AN0 voltage > > If it was 1k5:100k (near enough) that would be in parallel with the 250R > load on PIC1's AN0, dropping its effective resistance value by ~ 0.25% > > The other option is to reduce the gain of the converter by 1.5%, which > was my original question > There could be two contributions, gain error and offset error. Gain error will be dictated primarily by the tolerance of your resistors (including the 250R remote sense resistor). This would be the easiest to fix; one method would be to trim the 100R current sense resistor. Possibly make it a little larger e.g. 110R and add a larger value resistor in parallel to get the gain correct. Note that any imbalance between the two arms of your diff amp will cause the behaviour of the circuit to deviate away from a perfect current source because you will degrade the common mode rejection, e.g. if you change your remote sense resistor value from 250R to say 100R (or if the cabling between units has significant resistance), you would normally expect to get the same current, but this would not be the case if the 100K and 10K resistors were not well matched. In terms of offset error, the input offset voltage of the LM358 isn't particularly good, typical 2.9mv, max 7mv. You diff amp has a gain of 5, so this would translate to a typical voltage error of 14.5 mv. Across your 100R sense resistor this would give a nominal 0.145mA, which is about the error you are seeing. This is a little more difficult to fix with minimal circuit changes since you need to add or subtract a voltage, preferably without changing the overall gain. A higher performance op-amp (i.e. one with lower input offset voltage) would reduce this error. Regards Mike ======================================================================= This e-mail is intended for the person it is addressed to only. The information contained in it may be confidential and/or protected by law. If you are not the intended recipient of this message, you must not make any use of this information, or copy or show it to any person. Please contact us immediately to tell us that you have received this e-mail, and return the original to us. Any use, forwarding, printing or copying of this message is strictly prohibited. No part of this message can be considered a request for goods or services. =======================================================================

On Mon, Mar 2, 2009 at 8:02 PM, Michael Rigby-Jones <RemoveMEMichael.Rigby-JonesEraseMEEraseMEbookham.com> wrote: > Note that any imbalance between the two arms of your diff amp will cause > the behaviour of the circuit to deviate away from a perfect current > source because you will degrade the common mode rejection, e.g. if you > change your remote sense resistor value from 250R to say 100R (or if the > cabling between units has significant resistance), you would normally > expect to get the same current, but this would not be the case if the > 100K and 10K resistors were not well matched. > > In terms of offset error, the input offset voltage of the LM358 isn't > particularly good, typical 2.9mv, max 7mv. You diff amp has a gain of > 5, so this would translate to a typical voltage error of 14.5 mv. > Across your 100R sense resistor this would give a nominal 0.145mA, which > is about the error you are seeing. This is a little more difficult to > fix with minimal circuit changes since you need to add or subtract a > voltage, preferably without changing the overall gain. A higher > performance op-amp (i.e. one with lower input offset voltage) would > reduce this error. Gain error and offset error can normally be mitigated by calibration since this is a MCU based design. Temperature drift is difficult to be calibrated away though. Xiaofan

{Quote hidden}> -----Original Message----- > From: RemoveMEpiclist-bouncesspam_OUTKILLspammit.edu [RemoveMEpiclist-bouncesTakeThisOuTspammit.edu] On Behalf > Of Xiaofan Chen > Sent: 02 March 2009 12:41 > To: Microcontroller discussion list - Public. > Subject: Re: [PIC] PT100 4-20mA load > > On Mon, Mar 2, 2009 at 8:02 PM, Michael Rigby-Jones > <EraseMEMichael.Rigby-JonesspamspamBeGonebookham.com> wrote: > > > Note that any imbalance between the two arms of your diff amp will cause > > the behaviour of the circuit to deviate away from a perfect current > > source because you will degrade the common mode rejection, e.g. if you > > change your remote sense resistor value from 250R to say 100R (or if the > > cabling between units has significant resistance), you would normally > > expect to get the same current, but this would not be the case if the > > 100K and 10K resistors were not well matched. > > > > In terms of offset error, the input offset voltage of the LM358 isn't > > particularly good, typical 2.9mv, max 7mv. You diff amp has a gain of > > 5, so this would translate to a typical voltage error of 14.5 mv. > > Across your 100R sense resistor this would give a nominal 0.145mA, which > > is about the error you are seeing. This is a little more difficult to > > fix with minimal circuit changes since you need to add or subtract a > > voltage, preferably without changing the overall gain. A higher > > performance op-amp (i.e. one with lower input offset voltage) would > > reduce this error. > > Gain error and offset error can normally be mitigated by calibration > since this is a MCU based design. True, but the temperature sensor and 4-20mA conversion is purely analog so calibration through the micro would only be useful if a remote sensor and the base unit are always paired i.e. if you changed the remote sensor, you'd have to re-calibrate. Regards Mike ======================================================================= This e-mail is intended for the person it is addressed to only. The information contained in it may be confidential and/or protected by law. If you are not the intended recipient of this message, you must not make any use of this information, or copy or show it to any person. Please contact us immediately to tell us that you have received this e-mail, and return the original to us. Any use, forwarding, printing or copying of this message is strictly prohibited. No part of this message can be considered a request for goods or services. =======================================================================

> There could be two contributions, gain error and offset error Mike, thanks for the analysis Presently I'm using 5% carbon resistors that were simply to hand, and the LM358, so in this first-draft circuit some error could have been expected. The project is a proof-of-concept demo, which is no excuse for not doing it properly, but...... I'll get some 1% and a better amp for a revision As only a resolution of whole degrees is needed, for the time being I can use the circuit as is, if I don't have time to revise, and compensate for the extra gain by changing the comparison number used for rounding in PIC2's s/w. It's there anyway, so no extra code is needed. The more important reading is at PIC1, and that's correct