Yes, I did google this.
A useful result I found e.g. on the ESP pages.
Also Horowitz/Hill, The Art of Electronics (Third Edition) cover the subject in chapter 2.4.3.
None of these references speak about how to split the collector resistors and all of them split them evenly. You don't. Why?
Bipolar Transistor Cookbook — Part 7 | Nuts & Volts Magazine
Figure 13 has split values.
Can I try an explanation? In the nomenclature of #180 you want R11 to be large, b/c more R means more constant current at signal frequency and thus more OLG. R11+R12 sets the collector current for Q4 and R12 is - in AC terms - in parallel with the speaker. If you go low on R12 you eat into the speaker. Also note the 1W rating on R12. Some output power goes there.
Figure 13 has split values.
Can I try an explanation? In the nomenclature of #180 you want R11 to be large, b/c more R means more constant current at signal frequency and thus more OLG. R11+R12 sets the collector current for Q4 and R12 is - in AC terms - in parallel with the speaker. If you go low on R12 you eat into the speaker. Also note the 1W rating on R12. Some output power goes there.
Last edited:
The sum of two resistors determines collector current. At the middle point of that voltage divider we connect the capacitor to the amp's output in order to provide voltage swing above the rail (it's a positive feedback - AC voltages add). The higher the value of the "lower" resistor the higher the swing (voltage is proprtional to resistance) but we can't go too high since the "rail" (ie power supply) is a short circuit AC wise and we have to keep some distance from it or our previous effort will be ruined. So, as in all things, the trick is finding a good balance.
I have a prototype of the schematic in post #180 (does it have a name?) on the bench and boy, this thing is fast.
I built it on protoboard and there are a few cm of wire to reach the MOSFETs on the heat sink I use for testing. In this configuration the circuit oscillates somewhere btn 10 to 25 MHz.
I introduced 100pF btn base and collector of Q4 and changed both gate stoppers for the latfets to 470 Ohm to make the circuit stable.
Next I want to do some testing. Square waves already look razor sharp.
Are there better ways to stabilize the circuit? What are the critical elements in the PCB layout?
I built it on protoboard and there are a few cm of wire to reach the MOSFETs on the heat sink I use for testing. In this configuration the circuit oscillates somewhere btn 10 to 25 MHz.
I introduced 100pF btn base and collector of Q4 and changed both gate stoppers for the latfets to 470 Ohm to make the circuit stable.
Next I want to do some testing. Square waves already look razor sharp.
Are there better ways to stabilize the circuit? What are the critical elements in the PCB layout?
That circuit is tested and stable. Clumsy prototypes often create problems.
Make reliable connections (PS feedback loops, input, grounding) - no long wires, alligator clips and similar, ground your heatsinks....
100pf on Q4's base and collector is way too much - 10pf or so is more like it.
Try to slightly increase C4 or introduce a couple of pf over R10. Keep gate stoppers (R3, R6, R13, R14) very close to the gates.
Make reliable connections (PS feedback loops, input, grounding) - no long wires, alligator clips and similar, ground your heatsinks....
100pf on Q4's base and collector is way too much - 10pf or so is more like it.
Try to slightly increase C4 or introduce a couple of pf over R10. Keep gate stoppers (R3, R6, R13, R14) very close to the gates.
It's evening and I'm still working...
Amazing.
I tightened the layout of the board, eliminated the leads to the mosfet, brought C8 very close, but still had the oscillations if I used less than 100pf btn Q4 base and collector.
I followed Anatoly's advice and changed C4 to connect to Q4 base instead of the output behind C8 and my circuit went stable. I still use the same 10pf. I could even bring down the mosfet gate stoppers to their orginal 120 Ohm (was 470) and it is still stable.
Like this my amp shows 10MHz ringing on both rising and falling edges of a square wave. I changed R13, the n-FET gate stopper to 220 Ohm to dampen the ringing. It still there but only mildly.
I can see that Anatoly's change kills a 20MHz peak in the frequency response if I simulate it. I understand how C4 works in the origial schematic. Why is it working differently in the changed setup?
I tightened the layout of the board, eliminated the leads to the mosfet, brought C8 very close, but still had the oscillations if I used less than 100pf btn Q4 base and collector.
I followed Anatoly's advice and changed C4 to connect to Q4 base instead of the output behind C8 and my circuit went stable. I still use the same 10pf. I could even bring down the mosfet gate stoppers to their orginal 120 Ohm (was 470) and it is still stable.
Like this my amp shows 10MHz ringing on both rising and falling edges of a square wave. I changed R13, the n-FET gate stopper to 220 Ohm to dampen the ringing. It still there but only mildly.
I can see that Anatoly's change kills a 20MHz peak in the frequency response if I simulate it. I understand how C4 works in the origial schematic. Why is it working differently in the changed setup?