Bias Monitoring of Separate Heatsinks

[DCPreamp]

In building my mono-block amplifiers, I am in need of thermally tracking power devices mounted on two separate heatsinks. My chassis consists of an aluminum center-shell with large heatsinks making up the complete L/R sides of the amp. All NPN power devices will be mounted to one sink and all PNPs to the other.

Are there any recommendations on bias tracking of the output devices on separate heatsinks?

Should I only measure one side relying on device matching and E x I being equal?
Should I fabricate an aluminum “bridge” to mechanically connect the two sides and track bias on the bridge temperature?
Is there a good way to parallel two bias transistors and connect one to each side?
Is there a way to series connect two bias transistors (I’ve seen series NPN/PNP devices as amplified diodes) but have them operate in the 2.2V range?
How about using one transistor, with the C-B junction shorted forming a diode, placed in series with the emitter of a regular bias-configured transistor? That way, either B-E junction will produce an amplified change in collector current.

Any suggestions or examples are much appreciated.

[mikeks]

Here.

[Jan Dupont]

This will do it

[GK]

I don't think the aluminium bridge idea would work too well.
For best results, the thermal sensing component should be mounted as close to the output devices as possible. You don’t want to increase the thermal time constant any more than necessary. It would most likely be quite OK to sense the temperature of just a single heatsink, but if you wanted to get fancy, you could make a Vbe multiplier with two transistors, with each transistor mounted on each respective heatsink.
My main concern with an amplifier constructed like this though wouldn’t be with regards to being able to achieve adequate thermal compensation, but with being able to keep parasitic oscillations of the output devices at bay. You haven’t stated the dimensions of your chassis, but mounting each half of a complementary output stage at opposite sides of the chassis may be asking for trouble if you’re going to use transistors with a decent fT.

Cheers,
Glen

[mikeks]

Originally posted by DCPreamp
Should I fabricate an aluminum “bridge” to mechanically connect the two sides and track bias on the bridge temperature?

Not worth the hassle 'cause you would have to heavily insulate the bridge to make it effective at passing heat to the compensating transistor.

[mikeks]

Originally posted by ACD
This will do it

Nope: only Q15 may be mounted on the sink.

[ilimzn]

Assuming your powerr ails are symetrical and you have no DC offset, if you can cool the left and right heatsink so tehre is no great difference in temperature, it is enough you sense the temperature of only one heatsink.

[DCPreamp]

Interesting and informative answers – thank you.

I very much enjoyed the paper written by Dr. Hawksford. I plan to do some research and see what else he has to offer.

My chassis is only 8” wide and with the heatsinks being approximately 9” high, 8” deep, and 3” thick (fin height), so the amp will be roughly cube shaped. Wire lengths connecting the output devices to the output terminal will be kept minimal, but still, I have added places for extra compensation onto the signal PCB (see post Test Amp - Construction Method for PCB pics) in case of oscillation problems.

The “bridge” idea was my least favorite, but since the chassis is relatively small and narrow, and everything is aluminum, the temperatures within the chassis will not likely vary much. Still, I fully understand the ideal place to detect bias temperatures are “at” the power device (if not on the die).

Output devices are 2SC3281 and 2SA1302 with an fT = 10MHz. Not exactly super-duper devices, but should do just fine for my project. I’m not one of those subjectivist, “can hear a bug fart at 50-feet” kind of guys. Just give me a good, dependable, low-noise, low distortion, good bandwidth amp, and I’m happy.

If there’s additional interest in my mono-blocks, which are currently in the parts gathering stage, I will post some pics of what I have so far and parts’ specs.

Thanks again

[AndrewT]

Hi,
two solutions.
a.) adopt the Leach style Vbe multiplier that uses a cold Vbe and temp comp is achieved by mounting the multiple diodes on the heatsink(s).

b.) adopt the dual NPN/PNP Vbe multiplier with triple resistor string (fig3 in the Hawksford link from Mikeks-note the extra resistor that D.Self advocates). Mount the transistors on both the sinks. The centre resistor is the adjustable one to set the bias voltage.

Consider puttng half the PNPs and half the NPNs on each sink. Allows closer coupling of the Re and emitters reducing wireing inductance.

I found that the four diodes shown in leach V4.5 ClassAB Iq=210mA, overcompensates the output bias current. That would give an auto compensation for hot/cold ambient conditions since ClassA hardly varies in delta temp and then only when it comes out of ClassA.

The original Toshiba 2sa1302/c3281 are specified as 30MHz devices and their 1943 replacement are similar.

[ilimzn]

Originally posted by AndrewT
Consider puttng half the PNPs and half the NPNs on each sink. Allows closer coupling of the Re and emitters reducing wireing inductance.
[/B]

Hehe just thought of that myself. Also, this way it may be unnecessary to thermally compensate both heatsinks, unless you can produce a truly extreme temperature differential between them. Of course, using the PNP/NPN Vbe multiplier already mentioned, would be the best way if you stillw ant to sense temperature on both heatsinks.