Cheap Resistors Temperature Coefficient

[Jwillis]

I'm having trouble finding information about the TC of  the cheap 1% resistors from china. I can imagine they may be fairly high (100ppm/K) but was wondering if anyone here has a better idea of actual TC
I was having trouble determining the cause of drift with a differential amp for current sense . I could not find any other source of the problem .  Slight temperature change to them causes the V_o on the differential circuit to change more than I would like.
I would like to find something more suitable but I don't want to give up a kidney to do it. Any suggestions would be appreciated. 

[Jay_Diddy_B]

I'm having trouble finding information about the TC of  the cheap 1% resistors from china. I can imagine they may be fairly high (100-300ppm/K) but was wondering if anyone here has a better idea.
I was having trouble determining the cause of drift with a differential amp for current sense . I could not find any other source of the problem .  Slight temperature change to them causes the V_o on the differential circuit to change more than I would like.
I would like to find something more suitable but I don't want to give up a kidney to do it. Any suggestions would be appreciated. 

Can you share your schematic?

There may be other issues.

Regards,
Jay_Diddy_B

[John B]

From what I have purchased, the really cheap thin legged ones that also are not 1%, measure in at 300ppm/degrees C

[Jwillis]

Sorry for the resolution . But its a part of a revised version of an E load I've been working on.
 With the popular versions I was having trouble getting very low current loads unless a larger shunt was used. The smaller the  value of shunt made it worse .  With this circuit I can get very low currents to high currents . At low currents from 1uA to around 500mA the measurements are quite solid and stable . Over 500mA  to 5 A the more noticeable the drift becomes as the current rises. It's not much , maybe a few mA , but it does become stable after a little while. The drift does originate from the differential amp. Changing the shunt resistor to different types only made a very small difference. The Maganin works the best. Shielding the  suspect resistors helped some that's why I concluded it must be R6 R7 R10 R11.

[Jwillis]

The last band (brown)says they should be 100ppm/K but they sure don't act like it.

[Jay_Diddy_B]

Hi,

There is something not right in the schematic.

As it is drawn:

U1 is a unity gain buffer, the voltage on pin 6 is the same as the voltage on pin 3

U3 with all the resistors equal to 220K it has a gain of 1

In the U2 circuit R8 limits the gain. U2 is essentially a voltage follower, so again the voltage on U2-6 equals the voltage on U2-3

The control mechanism that you have is coming from the MOSFET Id versus Vgs.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi,
The behaviour of your circuit is the same as this circuit:




Regards,
Jay_Diddy_B

[srb1954]

 

From what I have purchased, the really cheap thin legged ones that also are not 1%, measure in at 300ppm/degrees C

This has also been my experience. In a batch of so-called 1% resistors I find you can have anywhere between 20% and 90% of the batch fall outside their marked tolerance.

The cheap thin leads are also magnetic so they are probably just plated steel. This will cause thermoelectric voltages to be generated where the connection is made to the copper on the PCB. This may add to the apparent drift of these resistors with temperature. 

[Jwillis]

Hi,

There is something not right in the schematic.

As it is drawn:

U1 is a unity gain buffer, the voltage on pin 6 is the same as the voltage on pin 3

U3 with all the resistors equal to 220K it has a gain of 1

In the U2 circuit R8 limits the gain. U2 is essentially a voltage follower, so again the voltage on U2-6 equals the voltage on U2-3

The control mechanism that you have is coming from the MOSFET Id versus Vgs.

Regards,
Jay_Diddy_B

Oh of course I see that now. Thank you . To get the feed back from the sense ,R8 needs to be removed  and R5 needs to be removed or at least  lowered  and the gain on U3 needs to be much higher to have any effect. An instrumentation amp might be more effective . I see where the drift problem is now at higher current.

From what I have purchased, the really cheap thin legged ones that also are not 1%, measure in at 300ppm/degrees C

This has also been my experience. In a batch of so-called 1% resistors I find you can have anywhere between 20% and 90% of the batch fall outside their marked tolerance.

The cheap thin leads are also magnetic so they are probably just plated steel. This will cause thermoelectric voltages to be generated where the connection is made to the copper on the PCB. This may add to the apparent drift of these resistors with temperature. 

I found that the ones I have do fall within the 1% tolerance .I was mainly wondering if the temperature coefficient was accurate to the 100ppm/K .Nothing is perfect so some deviation is expected. I have no real way of testing them . All in all they work fine in most applications but may not be practical in this application. We'll see. 

[TimFox]

100 ppm/K is typical of metal-film and thick-film SMD, but carbon-film TH parts have higher tempco.

[Vovk_Z]

I'm having trouble finding information about the TC of  the cheap 1% resistors from china. I can imagine they may be fairly high (100ppm/K) but was wondering if anyone here has a better idea of actual TC

First of all 100ppm/K is good TC. Cheap resistors usually have about 300 ppm/K up to 1000ppm/K.
And it depends on the resistor value. Low resistance ones may have higher TC, carbon ones too etc.

[Vovk_Z]

I was having trouble determining the cause of drift with a differential amp for current sense . I could not find any other source of the problem .  Slight temperature change to them causes the V_o on the differential circuit to change more than I would like.
I would like to find something more suitable but I don't want to give up a kidney to do it. Any suggestions would be appreciated.

It could be easier if you tell us what resistor do you use or want to use (as shunt?).

In short:
1) For current shunt good cheap ones are OAR3, OAR5 metal wire type. They have about 10-100 ppm (it depends on value, need to look datasheet).
2) For temperature stable feedback circuit you have to use 0,1% resistors.
3) 1% and 5% ones have the same TC because it is the same. They usually have TC >=300 ppm/K.

[Vovk_Z]

The behaviour of your circuit is the same as this circuit:

- No, your circuit doesn't have feedback, so current will vary much with temperature.
It seems to me you are right, R8 makes a voltage divider, so feedback doesn't work ok.

[Vovk_Z]

Sorry for the resolution . But its a part of a revised version of an E load I've been working on.

My 5 cents about this circuit:
1) R8 rather interfere then help. I'd get rid of them. Added: R8 makes a voltage divider with R5.
2) value of C2, C5 is ok.
3) you need to lower R4 by order, to 1-10 k. The same about R5, R6, R7, R10, R11 (because of noise).
4) As R12 I use OAR3 or OAR5 type. 0,01 R value is ok for about 5-10 Amps current. If your max current is 5A then 0.02 R can be better.
5) If R6, R7, R10, R11 are close to each other so they have equal temperature then their TC doesn't influence much. That means your device TC is affected by shunt TC and opamp input voltage drift TC.
6) There is no need in U1 buffer.

[Picuino]

Differential amplifiers are difficult to adjust. I would prefer to use an integrated differential amplifier.
For example
https://www.analog.com/en/products/ad8278.html
https://www.analog.com/en/parametricsearch/11081#/ from Analog Devices.

[Picuino]

Analog Dialogue 53-05, May 2019
Discrete Difference Amplifier vs. an Integrated Solution
https://www.analog.com/media/en/analog-dialogue/raqs/raq-issue-165.pdf

[Vovk_Z]

Differential amplifiers are difficult to adjust. I would prefer to use an integrated differential amplifier.

There is no need in integrated differential amplifier for such a simple electronic load if there isn't strong need because integrated dif. amplifiers usually more expensive.
I can even tell that may be there isn't a strong need in differential amplifier at all - we may have about the same properties without U3 at all. But it will depend on the PCB layout much.

[Picuino]

It depends on the accuracy that is required in the load current, but it is true that an integrated amplifier can go up to 1 or 2 dollars. For a single project it is not much money and it takes away the need to adjust the resistors.

[Jwillis]

It could be easier if you tell us what resistor do you use or want to use (as shunt?).

In short:
1) For current shunt good cheap ones are OAR3, OAR5 metal wire type. They have about 10-100 ppm (it depends on value, need to look datasheet).
2) For temperature stable feedback circuit you have to use 0,1% resistors.
3) 1% and 5% ones have the same TC because it is the same. They usually have TC >=300 ppm/K.

I am using a home made shunt from Manganin wire. From what I understand Manganin has a TC of 0.000015 . I used a milliohm meter to get the resistance as accurate as I could.  Mail is slow right now . So I'm still waiting for the ones I ordered.

My 5 cents about this circuit:
1) R8 rather interfere then help. I'd get rid of them. Added: R8 makes a voltage divider with R5.
2) value of C2, C5 is ok.
3) you need to lower R4 by order, to 1-10 k. The same about R5, R6, R7, R10, R11 (because of noise).
4) As R12 I use OAR3 or OAR5 type. 0,01 R value is ok for about 5-10 Amps current. If your max current is 5A then 0.02 R can be better.
5) If R6, R7, R10, R11 are close to each other so they have equal temperature then their TC doesn't influence much. That means your device TC is affected by shunt TC and opamp input voltage drift TC.
6) There is no need in U1 buffer.

I will try your suggestions. I put U1 in place to see what negative effect it would have on one circuit. Later I would like to try another in parallel. I realize  it's a redundancy with one load circuit.

I have a working 20A Eload using the popular circuit but wanted to try something a little different.

Differential amplifiers are difficult to adjust. I would prefer to use an integrated differential amplifier.
For example
https://www.analog.com/en/products/ad8278.html
https://www.analog.com/en/parametricsearch/11081#/ from Analog Devices.

I have been looking into fully integrated  differential amplifiers . But it has to work rail to rail , through hole and not to expensive including shipping . But I would like to get the concept to work first . If it will work.   

[Picuino]

Another differencial instrumentation amplifier that fits your requirements:
https://www.microchip.com/wwwproducts/en/MCP6N11

You can request for samples at Microchip.
https://www.microchip.com/samples/

[Vovk_Z]

I am using a home made shunt from Manganin wire. From what I understand Manganin has a TC of 0.000015 .

So 15 ppm. I have to add that 15 ppm is average TC for manganin, as I know. It can vary depending on a concrete alloy example, but it is definitely good for a current shunt.

[sorin]

I am using a home made shunt from Manganin wire. From what I understand Manganin has a TC of 0.000015 . I used a milliohm meter to get the resistance as accurate as I could.  Mail is slow right now . So I'm still waiting for the ones I ordered.

Where did you get the Manganin wire? Im trying to buy it for my projects, but cannot find it anywhere.

[Gyro]

There are plenty of ebay listings for Manganin wire from the Ukraine. This will almost certainly be former Soviet military stock.

[Jwillis]

I am using a home made shunt from Manganin wire. From what I understand Manganin has a TC of 0.000015 . I used a milliohm meter to get the resistance as accurate as I could.  Mail is slow right now . So I'm still waiting for the ones I ordered.

Where did you get the Manganin wire? Im trying to buy it for my projects, but cannot find it anywhere.

Occasionally 1mm to 2mm  shows up on Ebay but it's rare. Mostly the Manganin wire on Ebay is less than 0.9 mm which isn't much good for my purposes.
Here's a place in the UK that has it from 0.05 to 2mm https://www.wires.co.uk/acatalog/manganin.html
I haven't tried this place yet , so I can't be certain of what kind of service they have.
You can get 1mm Constantan on Ebay much easier but I think the TC is around -0.000074 .Can't say how that would work.

[sorin]

Here's a place in the UK that has it from 0.05 to 2mm https://www.wires.co.uk/acatalog/manganin.html
I haven't tried this place yet , so I can't be certain of what kind of service they have.
You can get 1mm Constantan on Ebay much easier but I think the TC is around -0.000074 .Can't say how that would work.

Thanks for the reply. I have already buy some Constantan from Ebay, but I dont have used it because I read this:

ISOTAN® is notable for its low temperature coefficient and
high resistance to oxidation and chemical corrosion. The alloy
is non-magnetic. It is suitable for electrical resistors, potentiometers,
heating wires, heating cables and mats. Ribbons are used for heating
of bimetals. Because of its high EMF against copper it is not
suitable for electrical precision resistors, therefore we recommend
MANGANIN®, NOVENTIN® or ZERANIN®. Due to its high thermal
EMF against copper, ISOTAN® is also used for thermocouples and
compensation cables (see separate brochure or www.thermoalloys.com)

ISOTAN is just a trademark for Constantan.