7

I have a pellet stove that after few minutes from startup turns off. I discovered that the cause is the smoke temperature that is too low (~ 30 °C, read on the remote) even when it's burning. Of course it thinks there is no flame and turns it off as a failure.

I found the thermocouple and I tried a basic trouble shooting.

  1. Putting the tc inside a saucepan with boiling water, the reading on the remote does not change. I inspected the board and there is a TLC271 op-amp near the tc connector. I checked the power supply (pins 4 and 7) and it's ok (5V). Then I checked the output (pin 6) at room temperature and in boiling water: the voltage changes between 13 mV and 320 mV. The remote always shows 30 °C.

  2. the original tc seems to be a J-type:

enter image description here

but I have a K-type:

enter image description here

I'm aware they are not 100% compatible but the readings should be quite similar for low temperatures (< 100 °C). Putting both in boiling water I read 97 °C and 110 °C using a multimeter with a tc input. But connecting the K-type probe to the stove, the remote still read 30 °C.

  1. the next obvious step was to be sure the MCU side of the board was ok. When I force (with a resistor) a voltage on the op-amp output (pin 6) the remote shows an higher value (i.e. 150 °C with about 3V). Hence the ADC is working.

Now, it seems what is damaged is the amplifier stage. But in my experience if an op-amp burns the output is 0V or it's just tied somewhere by the resistor's network in the nearby.

Do you think it's possible that an op-amp reduces so much its gain after a failure?

Mark
  • 1,161
  • 6
  • 23
  • You are not alone! Things like this happened to me with discrete elements, digital ICs as well as analog ICs. – fraxinus Dec 26 '22 at 17:15
  • As a potentially interesting datapoint - may or may not be related to this particular problem: Thermocouples that rely on dissimilar materials generating a temperature-dependent voltage, are actually measuring the *difference* between the TC probe and some other temp sensor. 0V at the TC terminals means equal temperature, so if a connection is broken, then you're just reading the internal temp, wherever that is. – AaronD Dec 27 '22 at 18:59
  • The reason it works that way is because you actually have *two* TC junctions (at minimum, but the result still comes out the same), one at the measurement point and one at the input terminal. At equal temperature, those TC voltages cancel out, and you measure 0V. If the measurement point has a different temperature from the terminal, then they don't cancel. – AaronD Dec 27 '22 at 18:59
  • @AaronD very interesting. But when you talks about a "connection broken" this could be anywhere between the op-amp anche the TC junction at the measurement point, right? I mean, it could be even on the PCB (i.e. an open resistor) – Mark Dec 27 '22 at 21:06
  • @Mark Yes, it could be anywhere. Broken wire, broken PCB trace, open resistor, anything that breaks the circuit. The circuit should be simple though - I can't imagine why it would have anything beyond ESD protection and screw terminals - but there are still a few things that can go wrong. – AaronD Dec 28 '22 at 21:54

3 Answers3

11

Take a look at page 4 of the datasheet that you linked. It shows the internal circuitry of the TLC271.

The output pin of the OpAmp is connected to a totem-pole NMOS output driver consisting of MOSFETs N4 and N5. Let's assume that N5 has failed, while N4 is still operating normally. In that case, the OpAmp can still sink current and pull its output voltage down, but it can't pull its output voltage up anymore. Similarly, if N4 failed and N5 is still alive, it can still pull the output voltage up, but not down anymore.

Another point of failure is R6 - if it degrades and increases in resistance, the OpAmp will have less and less output sourcing current capability. The same thing can happen if N4 or N5 degrade instead of failing outright.

If the input stage (MOSFETs P1/P2/N1/N2) has failed instead, it will no longer provide a drive current to the output stage's miller integrator, causing the OpAmp output to be stuck at a poorly-defined intermediate voltage, likely with high output impedance.

So, in short: It is very much possible that an OpAmp becomes "weak", and it's most likely caused by its output stage having degraded because that's the part of the OpAmp that is stressed the most. (All the other MOSFETs in the TLC271 operate on a few dozen microamps at most, compared to the output stage, which has to drive milliamps.)

Jonathan S.
  • 14,865
  • 28
  • 50
4

Usually damage in an op-amp has something to do with the front end because it's most exposed to excessive inputs. Power supply issues such as transients outside the absolute maximum range tend to result in output railed (due to failed compensation capacitor, for example). In both cases, railed output is the most likely outcome. The TLC27L2 has a bit different compensation capacitor arrangement though, with it directly connected to the output and perhaps it could have such an effect. The TLC27L2 has quite a large typical offset voltage so you might need to make significant adjustments if you replace the chip.

If it turns out to not be the chip, I would look carefully at passive parts, even resistors and capacitors that look okay. Possibly a capacitor failed short or a resistor failed open (or mostly open) and that is causing your problems. Or a cracked PCB. Or a failed trimpot. A short such as a ceramic capacitor failure near the input would have a similar effect (remember that 0mV in = room temperature for a thermocouple circuit).

Thermocouples do not fail easily in general- if they measure low resistance and don't have any shorts along the length they are okay. Also color codes vary quite a bit between countries- you might have type T rather than J, for example. And red might be minus or plus, depending. And type J has never been very popular in countries that use the color code you are assuming.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • About tc colors: to test a new probe, can I safely connect it in any way, right? Until I get a consistent reading – Mark Dec 27 '22 at 07:19
  • @Mark Yes, you can safely connect it either way. You could also heat it with a lighter or whatever and measure with a multimeter on 200mV range. – Spehro Pefhany Dec 27 '22 at 07:43
3

if an op-amp burns the output is 0V or it's just tied somewhere by the resistor's network in the nearby

The op-amp didn't fail catastrophically. You seem to use the term "burn" to mean catastrophic failure that yields a "dead" part that shows no discernible function beyond being resistor.

The op-amp was perhaps stressed beyond its absolute limits, and its performance degraded. It's not an uncommon outcome when parts are designed into marginal circuits, and they only criterion for release is "hey, it seems to work now" - without adequate engineering analysis..

When I force (with a resistor) a voltage on the op-amp output (pin 6) the remote shows an higher value

What was the resistor value? Even a 741 op-amp would need a load below 1kOhm to misbehave that way, assuming it was otherwise OK. It seems that the op-amp you checked had a "weak" output drive, due to some internal damage.

Pretty much every datasheet out there mentions that operation of the part beyond the absolute limits specified, or even prolonged operation close to the absolute limits can degrade the performance of the part, and that continued operation within specifications is not guaranteed in such circumstances.

Op-amps and other ICs, both analog and digital, can fail in all sorts of non-terminal ways - they get "sick" but not "dead". E.g. think of a RAM chip with some bad bits, or a CPU chip with a weak output pin driver, etc. The retrocomputing and vintage test equipment aficionados (fans) often face such "partial" failures when troubleshooting.

Kuba hasn't forgotten Monica
  • 32,734
  • 1
  • 38
  • 103