Hi,
typical British style without Cdom (Self) Miller comp across VAS.
Note, in addition to the input filters and the isolating resistance on the output there are 4 other compensation schemes to make this amp amp work. Crimson and Sugden do similar.
Kean,
in the next version the R10 changed to a variable resistor and was "tuned" to best shape of squarewave. JLH claimed this gave best sound.
I do wish he had wired RV4 correctly, but a pnp for Q8 would sort it without having to change the PCB.
typical British style without Cdom (Self) Miller comp across VAS.
Note, in addition to the input filters and the isolating resistance on the output there are 4 other compensation schemes to make this amp amp work. Crimson and Sugden do similar.
Kean,
in the next version the R10 changed to a variable resistor and was "tuned" to best shape of squarewave. JLH claimed this gave best sound.
I do wish he had wired RV4 correctly, but a pnp for Q8 would sort it without having to change the PCB.
Last edited:
Take a look at this thread http://www.diyaudio.com/forums/solid-state/69527-jlh-80w-mosfet-power-amplifier-modifying.html
Hi Hugh, JLH 80W amp was a second solid state amp, after Sinclair Z30, I made and still in use boht of them. I followed all JLH writings I could find(ETI, Wireless World). I followed your work too after I read about it in TNT and on Elliot pages, and I am still curious how did you succseed to keep all essencial secrets of a circuit diagrams now in internet time. I know only that what I saw on Elliot pages before it was removed from it(and some more from this forum).
dado
Ah, KT, that's an artefact of gif conversion from the original protel .pcb format. Nothing more.....
Dadod,
JLH was interesting. He applied both science and careful listening to his craft, worked by day as a nuclear scientist at Sellarfield facility (Windscale).
I'm not sure how I kept the secrets myself..... I'm sure they are all there if you look around the net. The amp was conventional but based on careful tweaking and listening tests. It deliberately introduced H2, and while THD was good, it was not outstanding.
These days my work is far more secretive, I try to preserve a few tricks of my own.
Cheers,
Hugh
Dadod,
JLH was interesting. He applied both science and careful listening to his craft, worked by day as a nuclear scientist at Sellarfield facility (Windscale).
I'm not sure how I kept the secrets myself..... I'm sure they are all there if you look around the net. The amp was conventional but based on careful tweaking and listening tests. It deliberately introduced H2, and while THD was good, it was not outstanding.
These days my work is far more secretive, I try to preserve a few tricks of my own.
Cheers,
Hugh
http://www.diyaudio.com/forums/aksa/191053-swordfishy-aspen-fetzilla-power-amp-3.html#post2636502
NFB tapping point is very close to not working correctly.
Move a few mm to the left (towards the output inductor) and it is then clearly on the output trace.
NFB tapping point is very close to not working correctly.
Move a few mm to the left (towards the output inductor) and it is then clearly on the output trace.
I have run a few sinewave and squarewave tests from 30Hz to 1MHz.
Sinewave input from 1kHz all the way to 200kHz into open circuit, into 100n,1u,4u7, into 50r//100n, //1u, //4u7 no change in output to be seen. The whole lot repeated with the 47pF VAS miller comp disconnected, again no change in output.
Squarewave from 100Hz to 200kHz. Same range of 7 loads with 47pF VAS miller and without 47pF.
The typical inductor ringing with the load capacitance shows but damps out as usual.
Very slight difference in the leading edge waveshape as the 47pF is clipped in and out.
Everything is so benign that I think the 680ns input filter is doing it's job too well.
The rise time of the amp output is very slow. 200kHz squarewave input becomes a 200kHz sinewave (well almost) at the output.
I wish I could video the scope trace to show the sinewave changing back to a filtered squarewave as the test frequency is reduced. Never noticed that before, probably because I haven't been so carefree in applying such high frequency signals and touching the VAS compensation with my finger to bring it back into circuit.
Listening now with the VAS miller comp disconnected. I think I'll try that finger test again to hear if the comp in or out is audible.
My conclusion is that neither oscillation nor inadequate phase margin had anything to do with the runaway output offset. I have been working at that same ~300mA of reduced output bias.
Sinewave input from 1kHz all the way to 200kHz into open circuit, into 100n,1u,4u7, into 50r//100n, //1u, //4u7 no change in output to be seen. The whole lot repeated with the 47pF VAS miller comp disconnected, again no change in output.
Squarewave from 100Hz to 200kHz. Same range of 7 loads with 47pF VAS miller and without 47pF.
The typical inductor ringing with the load capacitance shows but damps out as usual.
Very slight difference in the leading edge waveshape as the 47pF is clipped in and out.
Everything is so benign that I think the 680ns input filter is doing it's job too well.
The rise time of the amp output is very slow. 200kHz squarewave input becomes a 200kHz sinewave (well almost) at the output.
I wish I could video the scope trace to show the sinewave changing back to a filtered squarewave as the test frequency is reduced. Never noticed that before, probably because I haven't been so carefree in applying such high frequency signals and touching the VAS compensation with my finger to bring it back into circuit.
Listening now with the VAS miller comp disconnected. I think I'll try that finger test again to hear if the comp in or out is audible.
My conclusion is that neither oscillation nor inadequate phase margin had anything to do with the runaway output offset. I have been working at that same ~300mA of reduced output bias.
Last edited:
For a HP amp, I would choose a SE output stage, with a CCS.
Next preference could be a quasi output stage, using decent mosfets.
Greater speed could be a liability, of course.
Problem is, you need to accommodate at the least 75R cans, 32R at a pinch. This requirement is not all that far removed from driving a speaker.
Cheers,
Hugh
Next preference could be a quasi output stage, using decent mosfets.
Greater speed could be a liability, of course.
Problem is, you need to accommodate at the least 75R cans, 32R at a pinch. This requirement is not all that far removed from driving a speaker.
Cheers,
Hugh
Headphones are low-impedance but don't need much drive as I understand. Mine are 60R and need at most 100mV.
However my headphone amp is class A at 100mA and will drive my sensitive speakers well enough.
I've been contemplating a preamp/headphone amp that uses a BC550C/560C EF output at 12mA class A, which I think would work fine for my headphones.
- keantoken
However my headphone amp is class A at 100mA and will drive my sensitive speakers well enough.
I've been contemplating a preamp/headphone amp that uses a BC550C/560C EF output at 12mA class A, which I think would work fine for my headphones.
- keantoken
you are kidding yourself !
That is just 83uW or about 80dB peak from a sensitivity of 91dB/mW
I would expect normal peaks to be at least 20dB above this and extreme peaks to be a further 10dB higher again.
That requires 1Vpk for normal peaks and 3Vpk for extreme transients.
Younger listeners (= eejits) will listen even higher !!!! But they don't know the difference between impending deafness and clipping and distorted sounds.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.