Vbe problems ... I'm going nuts !!

[ostripper]

Yes , I have read all the threads on Vbe's (bias generators). I did read all relevant info in both Self's 5th edition and Cordell's book. I am still frustrated.
I even tried Rod Elliot's more basic method of tempco.

My first try was spanning a 4 OP device amp across 2 heatsinks with the drivers and Vbe on the front heatsink (1 of the 2) , this was stupid .. I knew better. That resulted in a unbalanced OPS , 1 pair stabilized at 16mv and the other at 8mv. I did run it long enough (failure) to see that it slowly creeped up in temperature.

My second attempt was with this same "span" (stupid again) , but with a separate driver heatsink. In this case , the unbalance happened again , but the average dissipation was quite solid at 12mv (average/16mv front - 8mv back - unbalanced) this would be the case between freezer and hairdryer (-5C - 70C).

My 3rd attempt (arggggg) is the present one. It has the drivers/Vbe on the main heatsinks , each amp has it's own heatsink.The good thing is that both heatsinks/amps are the same temperature. Still , especially with the bias turned up (22mv-.22Re/100ma+) , it shows signs of thermal runaway. Bias creeps up to 30+mv within an hour and keeps on going.

I tried to increase the coefficient of the Vbe by using higher gain devices (Ksc3503-F/ 150hfe and ksc2690-Y / 300hfe !!). This improved the speed of the bias generator but the thermal "creep" was still there. I also tried Elliot's suggestion of reducing the current in the vbe resistor divider to 1/12th that of the transistor. This reduced the problem further , but I could not achieve "flatline" or negative tempco with any combo of currents or devices.

I realize I am trying to compensate the 2 device's Vbe's with a single device and the MJE15033/32 - njw0281/0302 pair might have too much positive coefficient for a simple Vbe to deal with. I am about to go "round 4" with this , if the japanese can do it ... so can I !! Any suggestions would be very helpful and would be tested out "real world" .

OS

[Lazy Cat]

These can compensate two gain stages tempco, drivers & OPS on the same heatsink.

[Lazy Cat]

For very stubborn cases this one has even too much compensation so it has to be carefully calculated.

[ostripper]

Yes , I was considering designing a different Vbe. My question to those that know is whether I should just redesign for separate driver / outputs since I have to make a new board regardless (below 1). Self and Cordell both say compensating 2 junctions is much harder.


Are separate driver - output thermal arrangements better or worse ?? What is the effect of ambient temp. on the separate drivers ? I notice most projects go in this direction ... I really want to keep the simple 1 device Vbe. The layout does not give many options (ventilation + 1 heatsink away from it's circuitry).

OS

[Tekko]

Strange. My diy amplifiers never had this problem, they rather almost compensate a bit too much, dropping the bias.

[Lazy Cat]

Separate drivers case has not even one advantage. You need extra space on PCB, separate heatsinks, not having control over temperature of drivers etc. If only BIAS current vs. temperature compensation is an issue than design correct Vbe compensation and that's it.

[effebi]

To keep the things simple: just try to improve thermal coupling or your Vbe transistor. AKSA uses to screw it directly on top of one of the OP, if I remember well. It does not look lelegant maybe, but surely it is much more better coupled that way.
Otherwise, just screw the VBE transistor without any insulation on the H/S (I suppose a SOT126 or TO220 device). This should be possible if it is the only one doing so and the H/S is insulated from chassis.

[Bonsai]

Ostripper, have you read Bob Cordells thoughts on Vbe spreaders and looked at any of his circuits in his book? I have something on my website also, but admit its a bit complicated and perhaps not entirely practical.

In my current amp I use a small SOT23 devies as a sensor in a CFP Vbe spreader and I go from 120mA to 140mA cold to very hot.

[ostripper]

Originally Posted by Lazy Cat

Separate drivers case has not even one advantage. You need extra space on PCB, separate heatsinks, not having control over temperature of drivers etc. If only BIAS current vs. temperature compensation is an issue than design correct Vbe compensation and that's it.

I absolutely agree , my triple op for sub will have a more exotic Vbe. This one is short on space and hard to layout. With separate drivers , the drivers are not thermally coupled , you only have the feedback of the OP's to Vbe. If the drivers were cold , gain is reduced ,Op's are cool ... vbe responds to just the OP's.

By tekko - Strange. My diy amplifiers never had this problem, they rather almost compensate a bit too much, dropping the bias.

Describe "my DIY amps" ..... I am interested (HS / Vbe / devices)

By effebi - just try to improve thermal coupling or your Vbe transistor. AKSA uses to screw it directly on top of one of the OP

Yes , Vbe on a short ribbon cable - I have a spare threaded hole right next to one of the OP's (less than 1cm).

OS

[Tekko]

I have posted them in this forum section during the past 6 months.