How to test a temperature controlled differential transistor pair?

[diamonddustin]

I'm working on a Roland SH-09 synth, I can't get a steady note from it. I've recapped, replaced some crusty trim resistors, cleaned the keyboard bed, torn down the faders and cleaned them out, each step has improved my measurements, but it's still not rock steady.

At this point I feel I've isolated the issue down to a Fairchild µa726HC temperature controlled differential pair. The control voltage coming in is steady, but the wave form coming out will bounce above and below the note frequency by up to 5Hz, and after 10 minutes of being on the frequency generated will drift up to 50Hz.

I tested the transistors in the package, both seem to be fine, test method was using the diode tester on my meter (Fluke 87V if it matters). I tested the resistor that is tied to the temperature control pin, it reads dead on. I was wondering if the AC in my shop was effecting it, so I placed a piece of foam on the package like a hat, same issues.

Any thoughts on how to properly test it? Are these known to fail with age?

I would just buy a NOS one and swap it to see if that fixes the issue, but the cheapest I can find them are on eBay for $95 plus $25 shipping, which is a bit much for a potential part that just sits in a part bin until I die.

Link to µa726HC datasheet: https://www.alldatasheet.com/datasheet-pdf/pdf/54850/FAIRCHILD/UA726.html

Link to SH-09 service manual (package can be found on page 4 in the VCO section): https://www.synthxl.com/wp-content/uploads/2018/02/Roland-SH-09-Service-Note.pdf

[Manul]

I guess you could test the heater circuit. One way would be to start it cold and observe heater current. It should slowly decrease and stabilise. Cooling the package should increase the current, heating the package should reduce current. If the temperature regulation seems to work, then the part is likely ok (you already tested the function of transistors).

[Audiorepair]

Does the 726 look original?

It could be someones replaced it with a fake.

Or an equivalent manufactured later that isn't actually an equivalent in that particular circuit.

[shakalnokturn]

I would just buy a NOS one and swap it to see if that fixes the issue, but the cheapest I can find them are on eBay for $95 plus $25 shipping, which is a bit much for a potential part that just sits in a part bin until I die.

Unfortunately more and more of those old matched transistors are getting hard to find, when available on eBay there's still the question of trusting the supplier and in this case the cost...

In this situation considering the μA726's schematic is in the datasheet I'd probably go for a handmade replacement with SOT23-6 or smaller dual transistors and SMD thermal jumpers. The layout should be quite challenging, the original packaging's pin layout doesn't help at all.

Edit:
As for testing... Over the years I've run into that many differential pairs that had intermittent faults on audio amps that I wouldn't trust it with a quick diode test. I'd probably substitute it into an audio amp where I could listen to just the matched pair for glitches for hours on the run. Once those are ruled out I'd do as suggested by Manul, monitoring the Temp Adj. voltage if there is any active control on this, power supply voltage and current draw (insert small series resistance) with an oscilloscope.
Considering the symptoms my suspicions would be more on the matched pair than the temperature control circuitry.

[perieanuo]

hi,
maybe the problem it's not the ua726, frequency output is also depending on IC6 IC7 area
you have also ic4 involved in vco translation

[perieanuo]

I tested the resistor that is tied to the temperature control pin, it reads dead on. I was wondering if the AC in my shop was effecting it, so I placed a piece of foam on the package like a hat, same issues.

i don't understand this. that thing should be R38 120kohm, that's it?
and i doubt this is the pb, at limit some bad soldering, but i doubt this too

[Audiorepair]

This may be of interest.

The Moog prodigy uses a transistor array IC as a temperature compensated differential pair, using one of the transistors as a heater, and another as a temperature sensor.  The whole thing can be calibrated, as opposed to the 726 which is all done in house.

http://dl.lojinx.com/analoghell/MoogProdigy-TechnicalServiceManual.pdf


Bear in mind, though, that recent versions of the 3046 chip do not work in this circuit, it is designed around an old version which has different parameters, so you will need to build and test your own version, not copy this one.

[diamonddustin]

I guess you could test the heater circuit. One way would be to start it cold and observe heater current. It should slowly decrease and stabilise. Cooling the package should increase the current, heating the package should reduce current. If the temperature regulation seems to work, then the part is likely ok (you already tested the function of transistors).

Good call. I have a meter reading the current on the V+ pin, it's sitting idle at 0.05mA, and taking a temperature reading of the package I'm getting around 25C, 2C above room temp. This seems to not agree with what the data sheet shows on it's current to ambient temp graph, which if I'm reading it correctly, 0.05mA is what it should be drawing at around 120C. When I cooled it, the current did rise, but only to around 0.11mA.

2C above ambient doesn't seem right at all, I would think to get stability across the transistor pair, you'd want it over 50C (pure guess). It would be nice if in the data sheet they included a chart or formula for what temperature you'd get from a resistance value.

[diamonddustin]

Does the 726 look original?

It could be someones replaced it with a fake.

Or an equivalent manufactured later that isn't actually an equivalent in that particular circuit.

It belongs to the original owner, and they've never had it serviced, so I'm fairly certain it's the original. But still a possibility.

[diamonddustin]

This may be of interest.

The Moog prodigy uses a transistor array IC as a temperature compensated differential pair, using one of the transistors as a heater, and another as a temperature sensor.  The whole thing can be calibrated, as opposed to the 726 which is all done in house.

http://dl.lojinx.com/analoghell/MoogProdigy-TechnicalServiceManual.pdf


Bear in mind, though, that recent versions of the 3046 chip do not work in this circuit, it is designed around an old version which has different parameters, so you will need to build and test your own version, not copy this one.

I did find this, which seems like the way to move forward since it was designed to replace the 726. Fingers crossed.

http://www.portabellabz.be/pa726.html#to100

[Manul]

Good call. I have a meter reading the current on the V+ pin, it's sitting idle at 0.05mA, and taking a temperature reading of the package I'm getting around 25C, 2C above room temp. This seems to not agree with what the data sheet shows on it's current to ambient temp graph, which if I'm reading it correctly, 0.05mA is what it should be drawing at around 120C. When I cooled it, the current did rise, but only to around 0.11mA.

It seems the heating circuit failed, but just to be sure, I would try to measure supply voltage and replace the Radj resistor, maybe try smaller value.

[Audiorepair]

I did find this, which seems like the way to move forward since it was designed to replace the 726. Fingers crossed.

http://www.portabellabz.be/pa726.html#to100

Ah, someones done all the work then!


I may have to use that one day.
Thanks.