This is probably more of a "what you feel comfortable with" type question. Anyhow, The transistors in the VAS stage of an amplifier (current source and amplifying) heat up to 140f / 60c at idle. Since the VAS is class A, they don't heat up anymore with a signal. Of course, this is far from the danger zone but no one would run transistors at max junction temp continuously. I know the cooler the better, so at what temperature would you consider heat sinking them?
TO-92 thermal resistance from junction to ambient is 200 deg C per watt.
If you install a heatsink between the transistor and the ambient air, you need to remember that TO-92 thermal resistance from junction to case is 84 deg C per watt.
If you want to keep the junction temperature below 110 C, while operating the device in 45 C ambient air, using NO HEATSINK:
If you fit an Aavid Thermalloy TO-92 heatsink (Mouser part number 532-575200B00), whose case to ambient thermal resistance is 60 deg C per watt, then:
Naturally you can change the assumed junction temperature and the assumed ambient air temperature, and calculate again. Dubai hit 49 C this week and Pakistan's all time high is 54 C.
_
If you install a heatsink between the transistor and the ambient air, you need to remember that TO-92 thermal resistance from junction to case is 84 deg C per watt.
If you want to keep the junction temperature below 110 C, while operating the device in 45 C ambient air, using NO HEATSINK:
- 110 > 45 + (Power * 200) -----> solution: Power < 0.325 watts
- 110 > 45 + (Power * (84 + 60)) -----> solution: Power < 0.451 watts
_
Last edited:
TO-92 is underrated for VAS.
There were some TO92 with a bare metal flange that Allen organ used for VAS & drivers, zetex 2n6715 2n6727 but no longer available and the NTE "copies" have no heat sink.
TO-39 has same current rating but wastes the heat into a sink a lot better because of metal package. Newark usually has a pnp & npn to-39 or TO-18 in stock. I bought multicomp 2n5682 120 Vceo rated, but their numbers change a lot.
I've used 30 mhz rated TO-220 package for vas and my unit sounded very good on top octave piano & tinkly bells. GE D44R4 (no longer available, but MJE15028/29 is). The Ccb boogerbear did not eat my highs in an AX6 circuit.
There were some TO92 with a bare metal flange that Allen organ used for VAS & drivers, zetex 2n6715 2n6727 but no longer available and the NTE "copies" have no heat sink.
TO-39 has same current rating but wastes the heat into a sink a lot better because of metal package. Newark usually has a pnp & npn to-39 or TO-18 in stock. I bought multicomp 2n5682 120 Vceo rated, but their numbers change a lot.
I've used 30 mhz rated TO-220 package for vas and my unit sounded very good on top octave piano & tinkly bells. GE D44R4 (no longer available, but MJE15028/29 is). The Ccb boogerbear did not eat my highs in an AX6 circuit.
Last edited:
I use heatsinked TO-220 MJE15032 and MJE15033 for VAS and it works fine and sounds great. The limiting factor for bandwidth and distortion is almost always the output transistors, and if it's not, probably something is wrong with the design.
I would like to stick with cheap to-92 for vas, unless the power rail is so high that the temp is too high to be comfortable.
The major benefit is that to-92 inherits good ft and less parasitic capacitance, in general.
The major benefit is that to-92 inherits good ft and less parasitic capacitance, in general.
Zetex (now Diodes Inc.) E-line packages can handle a bit more power, and they still put planar transistors with a reasonable fT in them. See the ZTX458 datasheet, for example. If you can accept small emitter resistors, you can also parallel a few TO-92 devices (each with a small emitter resistor to get a stable current distribution).
The fT may be high enough on 50 watt TO-220 devices, but output capacitances of 50 pF or more are getting up there. If the Miller capacitance is nonlinear you get more distortion. It is better with a smaller device and an added collector-base capacitor. That was the big draw fir TO-126 VAS devices - similar die sizes to the TO-39’s and some higher power To-92’s, and could handle a watt or more with no heat sink - and be easy to add a heat sink to if you wanted to. But these are now getting few and far between unless you want dogs like the modern version of the BD139/140 (might as well use MJE1503x).
TO-92’s work fine on say +/-30 volt supplies, where you don’t need 20 some odd mA in the VAS and can get by with 6. I guess paralleling is another option - I do it with TO-126’s when the dissipation gets in the 2-3 watt range and I want a decently low Cob.
TO-92’s work fine on say +/-30 volt supplies, where you don’t need 20 some odd mA in the VAS and can get by with 6. I guess paralleling is another option - I do it with TO-126’s when the dissipation gets in the 2-3 watt range and I want a decently low Cob.
I've had no issues with operation up to 90C die (about 60C on the package) with ~150mW dissipation in input stages.
Of course, I later added a small piece of aluminum to sink some of the heat. there were no measured performance differences but India is pretty hot, so some safety margin is always welcome.
Of course, I later added a small piece of aluminum to sink some of the heat. there were no measured performance differences but India is pretty hot, so some safety margin is always welcome.