Mr. Mooly, I have a question. I have a 2x40vac transformer. This in my city can cause a filtered DC voltage of up to 60v per rail. I am not sure about the durability due to thermal excess in Q3/Q5 and Q7/Q8.
In Q3/Q5 I think there is 6mA DC (in CCS). The question is, how much current will there be in Q7/Q8 when we bias the mosfets at 100mA?
I could risk mounting small heatsinks on these TO-92 transistors to alleviate the temperature.
For the power mosfets I could put the ECX20 dual matrix models.
Last edited:
Q3/5 can't handle that much voltage even with a heatsink. You'd have to use something like SC3503/SA1381 with a heatsink for those voltages. Untested on the bench, but does seem to work in simulations.
Q7/8 should be fine.
Q7/8 should be fine.
Q3/Q5 are transistors that support 150v of Vce. They hold up well. Due to tension, they hold up well. They are limited by temperature and power.
I would just like to know the bias current of Q6/Q7 when the mosfets are at 100mA.
I would just like to know the bias current of Q6/Q7 when the mosfets are at 100mA.
Poor wording on my part. When I say they can't handle the voltage, I meant the dissipation that results from 60V rails. Once you calculate the derated power due to junction temp rise, there is near zero safety margin on the TO-92L devices with 60V rails even with heatsinking.
Q3/Q5 are the limiting devices when it comes to the rail voltages. Dissipation in Q7/8 won't be significant. I can look at in sims later tonight if no one else answers before then.
Q3/Q5 are the limiting devices when it comes to the rail voltages. Dissipation in Q7/8 won't be significant. I can look at in sims later tonight if no one else answers before then.
Q3/Q5 I think there is 6mA by the ccs. 60x0.006 is 360mW each. Is it possible with a small heatsink work well.
Q6/Q7 how much current with 100mA in the power mosfets?
Q6/Q7 how much current with 100mA in the power mosfets?
Last edited:
Apologies - I meant Q6/7 (the drivers).
For Q3/5, my post #1856 dives into this deeper power derating.
Restating the equation from this post for 60V rails:
Dissipation: (60V x 110%) x 6.2mA = 409mW
Junction Temp Rise: 409mW x 200°C/W = +82°C
De-rated Power Rating: 625mW - (82°C x 5mW/°C) = 215mW
Without a heatsink, this is a no-go
Now let's add a heatsink. Let's use Aavid 92FG as an example. It's one of the few TO-92 heatsinks available. We're using it mainly to get a realistic estimated value for a TO-92 heatsink. It's has a stated value of 36.1°C/W.
From the 2N5551/2N5401 datasheet: Thermal Resistance, Junction−to−Case: 83.3°C/W
If we add the Aavid 92FG heatsink to this, we end up with a Junction-to-Ambient of 119.4°C/W. Putting this into the equations we get:
Dissipation: (60V x 110%) x 6.2mA = 409mW
Junction Temp Rise: 409mW x 119.4°C/W = +49°C
De-rated Power Rating: 625mW - (49°C x 5mW/°C) = 380mW
We're still running a bit too hot. The above equation has a 10% safety margin for voltage regulation. If we take this out we get:
Dissipation: 60V x 6.2mA = 372mW
Junction Temp Rise: 372mW x 119.4°C/W = +45°C
De-rated Power Rating: 625mW - (45°C x 5mW/°C) = 400mW
Without a safety margin, it would technically be within spec.
However, one other thing to consider. Heatsinks are often rated at 100°C and you need to derate them for the temp they are actually running at. Conrad Heatsink's website has a nice description of this. If this heatsink performs even a little below spec, you might have a problem. If you use a different more efficient heatsink, you might get it to run cool enough.
It could work. It's up to you if this is within your comfort zone.
For Q3/5, my post #1856 dives into this deeper power derating.
Restating the equation from this post for 60V rails:
Dissipation: (60V x 110%) x 6.2mA = 409mW
Junction Temp Rise: 409mW x 200°C/W = +82°C
De-rated Power Rating: 625mW - (82°C x 5mW/°C) = 215mW
Without a heatsink, this is a no-go
Now let's add a heatsink. Let's use Aavid 92FG as an example. It's one of the few TO-92 heatsinks available. We're using it mainly to get a realistic estimated value for a TO-92 heatsink. It's has a stated value of 36.1°C/W.
From the 2N5551/2N5401 datasheet: Thermal Resistance, Junction−to−Case: 83.3°C/W
If we add the Aavid 92FG heatsink to this, we end up with a Junction-to-Ambient of 119.4°C/W. Putting this into the equations we get:
Dissipation: (60V x 110%) x 6.2mA = 409mW
Junction Temp Rise: 409mW x 119.4°C/W = +49°C
De-rated Power Rating: 625mW - (49°C x 5mW/°C) = 380mW
We're still running a bit too hot. The above equation has a 10% safety margin for voltage regulation. If we take this out we get:
Dissipation: 60V x 6.2mA = 372mW
Junction Temp Rise: 372mW x 119.4°C/W = +45°C
De-rated Power Rating: 625mW - (45°C x 5mW/°C) = 400mW
Without a safety margin, it would technically be within spec.
However, one other thing to consider. Heatsinks are often rated at 100°C and you need to derate them for the temp they are actually running at. Conrad Heatsink's website has a nice description of this. If this heatsink performs even a little below spec, you might have a problem. If you use a different more efficient heatsink, you might get it to run cool enough.
It could work. It's up to you if this is within your comfort zone.
I think everyone has covered it nicely 🙂
Once you start pushing the rails into this sort of territory (and 45 to 60 volt is a 33% increase) I think you have to look at different transistors for the current source and VAS stage. Q5 in the original diagram in post #1 is the most important performance wise. You don't really need a complementary pair for these two. For the current sink even something like an MJE350 should be fine but as always they are all untested changes to the original design.
You also have to look at what is available where you are. Brian suggests the 2SC3503 and the 2SA1381 but as always, if you try anything like these make sure you buy known genuine parts.
As Mooly points out... what is available. Unfortunately, 2SC3503 / 2SA1381 have been unavailable for awhile now. And KSA1381 (for Q3) will be EOL Jan 2025. Most resellers are already out of stock with large minimums and long waits to get any before the deadline. KSC3503D (for Q5) is still available. I would image this would be the preferred devices for Q5 if the rails are pushed above 45V. As Mooly points out, Q3 is more forgiving. There are a number of in production devices that should work here.
Works in sim, but untested on the bench. Proceed with caution.
Works in sim, but untested on the bench. Proceed with caution.
So Q6/Q7 from the original diagram of the post would not have problems with +-60vdc.
And the problem would only be Q3/Q5. Especially Q5 that I could test with a KSC3503 and in Q3 a MJE350, both with a small heatsink.
Would these decisions be theoretically acceptable?
Even if it is just to test them...
Yep. Are you planning on using one of the PCB layouts from this thread? Be aware of the pinout differences that likely exist.
No, I will use my own pcb design, although it looks similar to one proposed in this thread. Don't worry about the position of the transistor pins, I have a lot of experience in PCB design and analog electronics.
Mr. Mooly, don't worry about the original components, I only buy semiconductors from Mouser and RS, and to a lesser extent Farnell and Digikey. 80% of my semiconductor purchases were always made at Mouser. There is no problem with original parts.
I had a design made of PCB, I will modify it to change Q3/Q5 from the original diagram to TO-126.
Mr. Mooly, don't worry about the original components, I only buy semiconductors from Mouser and RS, and to a lesser extent Farnell and Digikey. 80% of my semiconductor purchases were always made at Mouser. There is no problem with original parts.
I had a design made of PCB, I will modify it to change Q3/Q5 from the original diagram to TO-126.
I've already modified it...
The heatsinks chosen:
@brian92fs, I hope that with these heatsinks and these transistors the dissipation will be sufficient, if you want you can do a dissipation calculation for me with these new components.
@brian92fs, I hope that with these heatsinks and these transistors the dissipation will be sufficient, if you want you can do a dissipation calculation for me with these new components.
I also have a PCB designed for the power supply and speaker protection.
You're good. A small heatsink on TO-126 devices shouldn't have any problem.
How will that heatsink mount? The resistors appear to be interfering. Will it be mounted high above them?
Hi Mooly, As I think about this more, isn't Q4 more important than Q5? With the beta-enhanced VAS, I thought the specs of the second transistor (Q5) aren't as important. Isn't it Q4 that has the larger influence?
Yes, that's correct. The heatsinks are small and light, and are mounted above the resistors. They do not interfere with the assembly.