[grasshpr]

Hi Everyone,

Question for those who have put together their dynalo's. I recently purchased a set of TO-92 heatsinks made by Aavid Thermalloy (digikey part# HS251-ND) for 2SA1015 and 2SC1815 transistors. I bought them in hopes to reduce transistor temperatures in my Hammond enclosure for my dynalo.

I wanted to know if it was necessary for thermal bonding of these heatsinks. I know that these transistor gains depend on temperature and if you thermally bond these transistors you hopefully will have a common temperature thus providing a common gain. However, the way I see it, since the ambient temperature is constant (or slowly increasing), each heatsink will have similar natural convection characteristics, which will allow each transistor to be similar in temperature (give or take several degrees). What do you guys think? BTW, my dynalo rocks! Its as good as my tube amp!

grasshpr

 

[Pars]

I would think you could just slap them on (Digikey product info under mounting says "press fit"). Let us know how these work... do you notice a more constant offset with them? You've probably already discovered that while watching the offset on a meter, just putting your finger on one bank of the output trannies will cause the offset to change.

 

[Garbz]

My first design had a heatspreader accross all of them to bond them together, but despite how hot they get they wern't any where near cooking temp. My current design therefore is just plain open.

 

[dgardner]

Remember that the purpose of the thermal tracking diodes is to make sure the output stage bias stays more constant over the range of operating temperatures. You might want to try to attach the bank of four transistors to the adjacent thermal diode with something thermally conductive like a small strip of metal or heatsink. I little bit of two-part epoxy might do the trick. The goal is have the diode's component body at the same temperature of the output stage components' bodies.

 

[grasshpr]

Quote:

Originally Posted by dgardner Remember that the purpose of the thermal tracking diodes is to make sure the output stage bias stays more constant over the range of operating temperatures.

I'm not able to touch my diodes but, do they increase temperature as drastically as transistors?

I'm thinking that if transistors are at a cooler temp and the diodes are also cool, then my output stage bias won't be very large. Also, I'm hesitant to connect anything to the diode mainly because its small, hard to reach, and could ruin my amp if I squeeze the epoxy at the wrong location (like at the terminals).

My major concern for the transistors is to cool them down alittle. When I put my finger on a set for more than 5 seconds, I feel alot of pain. If this is going to be in an enclosed body then I can assume these temps will increase another 10 degrees after constant use.

 

[Pars]

Quote:

Originally Posted by grasshpr I'm not able to touch my diodes but, do they increase temperature as drastically as transistors?

No the diodes will not get hot (by themselves). The idea of thermally bonding them to the output transistors was so that their junction drop would vary with the transistors, thereby equalizing the offset and gain changes as the transistors got hotter.

Quote:

Originally Posted by grasshpr I'm thinking that if transistors are at a cooler temp and the diodes are also cool, then my output stage bias won't be very large. Also, I'm hesitant to connect anything to the diode mainly because its small, hard to reach, and could ruin my amp if I squeeze the epoxy at the wrong location (like at the terminals). My major concern for the transistors is to cool them down alittle. When I put my finger on a set for more than 5 seconds, I feel alot of pain. If this is going to be in an enclosed body then I can assume these temps will increase another 10 degrees after constant use.

What is your output current? Measure the voltage drop across the output resistors (avg) on each bank and divide by the resistance. For example, if you measure 0.44V across the emitter resistors, with a resistance of (nominal) 24.9 ohms, you get 0.0176A, or 17.6mA. Anything over 18 or so is starting to get a bit excessive. This can be changed by either increasing the 499 ohm resistors in the front end, or decreasing the resistor between the diodes. You could of course also increase the emitter resistor value, but thats alot of resistors...

 

[grasshpr]

I don't have my dynalo handy right now (and won't be able to measure any electrical quantities since its at the lab right now) but I do not think I am exceeding the temp specs of the transistors (and I doubt I'm drawing anything that will fry the amp).

I am thinking in terms of long term operation and how prolonged exposure to high temps will effect the amp/transistors. Mainly, I don't feel comfortable having components in an enclosure run at 40-50 degrees celsius. So, if I can reduce transistor temp without adversely effecting the output stage bias then I will be very happy

Dan, thanks for giving me a good explanation for the purpose of having thermal bonding of transistors and diode.

 

[dgardner]

Quote:

Originally Posted by grasshpr Mainly, I don't feel comfortable having components in an enclosure run at 40-50 degrees celsius.

Oh, you'll be running with TO-92 case temperature higher than 40-50 C for sure. No worries. You won't exceed maximum junction temperature if you stay inside the measurement the Pars has listed above. This unit should run for years without premature failure. And remember, transistors are $0.10 each.

 

[grasshpr]

Yeah, there cheap, but I can't stand to pay for shipping for these things. 1 lb of transistors and MCM charged me 7-8 dollars!

 

[grasshpr]

Wouldn't it be interesting if someone could place some active cooling elements (using a separate power supply) on the output transistors, bond a thermocouple on the 1N4148 diode and transistors and try to do some temperature control and see if this gives better results than thermal bonding?

From what I know about thermal bonding, temperature distribution may vary considerably, spatial, due to the choice of heat conducting material and geometry of the element.

What do you guys think? Is this worth an effort?

 

[Garbz]

well in order to control the offset as the temp changes no i don't think it's worth the effort given that the design has a servo in place to compensate for this.

 

[amb]

Quote:

Originally Posted by Garbz well in order to control the offset as the temp changes no i don't think it's worth the effort given that the design has a servo in place to compensate for this.

The primary thing being controlled by thermally-coupling the bias diodes to the output transistors is not DC offset. It is to keep the quiescent current in check (because BJTs have positive tempco, and thus propensity to thermal runaway).