Hello Richard,
I can't to touch the source.
I looked at caps you suggested. I need 6.8-10uF cap, to use with the 10KOhm input. PPMFX is claimed as ultimate series. The dissipation factor is:
< 0.05% 1.0 - 10.0 uf @ 1kHz @ 25 degress C
tol: 10%
Counterpart from Mundorf - Mcap supreme:
Loss factor: tan ∂ = 0.0002@1 kHz; 0.0001@10 kHz
tol: 2%
Jantzen Superior Z-Cap MKP looks similar to MCap from Mundorf.
Tolerance value is non critical here, but Mcap far better in terms of looses.. Or I missed something ?
Now the DC servo. I tried to simulate you amp circuit with DC servo added, and noticed kind of improvement in THD floor. The amp benefits from low THD if output DC offset is so small as possible. DC servo doing here perfect job.
I can't to touch the source.
I looked at caps you suggested. I need 6.8-10uF cap, to use with the 10KOhm input. PPMFX is claimed as ultimate series. The dissipation factor is:
< 0.05% 1.0 - 10.0 uf @ 1kHz @ 25 degress C
tol: 10%
Counterpart from Mundorf - Mcap supreme:
Loss factor: tan ∂ = 0.0002@1 kHz; 0.0001@10 kHz
tol: 2%
Jantzen Superior Z-Cap MKP looks similar to MCap from Mundorf.
Tolerance value is non critical here, but Mcap far better in terms of looses.. Or I missed something ?
Now the DC servo. I tried to simulate you amp circuit with DC servo added, and noticed kind of improvement in THD floor. The amp benefits from low THD if output DC offset is so small as possible. DC servo doing here perfect job.
Last edited:
Of course, simulation isn't precise, especially while software optimized mostly for MCU. But even such simulation can to show a general behaviour, and to give some start point.
I tried to measure HPA using Creative Zx soundcard, and it show about 0.009-0.005% (-81 -88dB) THD. The soundcard + Arta are not the ethalon of the precision also.
I waiting for 'DIY Audio Analyzer with AK5397 and AK4490' or that Onix ADC became available.
I tried to measure HPA using Creative Zx soundcard, and it show about 0.009-0.005% (-81 -88dB) THD. The soundcard + Arta are not the ethalon of the precision also.
I waiting for 'DIY Audio Analyzer with AK5397 and AK4490' or that Onix ADC became available.
You probably wont be able to measure the distortion with anything but the best instrumentation and care in layout and shielding when testing; If you use the transistors I used. But the prposed DIY analyzer should get you easily to -100dB accurately (.001%).
Tell me more about the sound you have now, then with cap and then with servo.
Thx - Richard
Tell me more about the sound you have now, then with cap and then with servo.
Thx - Richard
Wow Wow Wow... I need some time 
I'm not living in either USA EU or Japan.
There is no Hi Fi electronic components stores here. So capacitors I should to order from you country. Then PCB for DC servo I need to order from East, because PCB produced in my country will cost just more than the whole amplifier.
The PPMFX caps I found here Michael Percy Audio Ordering Information 8uF X 200V costs $16.25 , for PCB I'll contact a factory in Hong Kong.
The HPA sounds now very good to my ears, very accurate and natural. There is tight and deep bass, detailed, airy, and smooth treble. The treble are similar to treble produced by my hybrid HPA - just very pleasant, and absolutely non fatiguing.
I'm not living in either USA EU or Japan.There is no Hi Fi electronic components stores here. So capacitors I should to order from you country. Then PCB for DC servo I need to order from East, because PCB produced in my country will cost just more than the whole amplifier.
The PPMFX caps I found here Michael Percy Audio Ordering Information 8uF X 200V costs $16.25 , for PCB I'll contact a factory in Hong Kong.
The HPA sounds now very good to my ears, very accurate and natural. There is tight and deep bass, detailed, airy, and smooth treble. The treble are similar to treble produced by my hybrid HPA - just very pleasant, and absolutely non fatiguing.
Well, I assembled a servo at kind of breadboard. First of all it defeated that voltage offset.
While amp with input shorted to ground was drifting +-3...5mV, with source connected and volume on max it was about 20mV, now, it 0-0.4mV regardless to what's going on the input.
The servo works better if it ground connected to the star ground point just between the regulators.
With 100K resistors on input and 50K at output of the servo, the LF roll off begins somewhere about 2Hz -0.1dB, 0.8Hz -0.2dB.
I didn't noticed either noticeable changes in the sound. Probably hi end became more transparent. But it can be a my subjective bias.
While amp with input shorted to ground was drifting +-3...5mV, with source connected and volume on max it was about 20mV, now, it 0-0.4mV regardless to what's going on the input.
The servo works better if it ground connected to the star ground point just between the regulators.
With 100K resistors on input and 50K at output of the servo, the LF roll off begins somewhere about 2Hz -0.1dB, 0.8Hz -0.2dB.
I didn't noticed either noticeable changes in the sound. Probably hi end became more transparent. But it can be a my subjective bias.
yes. I suspect the same p/n for transistors were not used. The ones I used were also operated at their zero TC bias point and does not drift.
When you force an unbalanced amp to balance by servo you often get higher distortion.... balanced for DC isnt always the same as balanced for AC.
However, I would not bother with servo with the small offset being talked about.
A misuse of a dc-servo has been to zero out an unmatched/unbalanced stage.... I developed the servo concept primarily for reducing dc drift. With closely matched devices and the ones selected and operated as I did.... offset and balance is not an issue and distortion (as measured) is extremely low.
THx-RNMarsh
When you force an unbalanced amp to balance by servo you often get higher distortion.... balanced for DC isnt always the same as balanced for AC.
However, I would not bother with servo with the small offset being talked about.
A misuse of a dc-servo has been to zero out an unmatched/unbalanced stage.... I developed the servo concept primarily for reducing dc drift. With closely matched devices and the ones selected and operated as I did.... offset and balance is not an issue and distortion (as measured) is extremely low.
THx-RNMarsh
Last edited:
the small signal transistors are the same. The power transistors are 2SB649AC 2SD669AC.
Do I have to change these to MJE210G 200G ? The problem is that obtaining MJE2XXG is subject to import from US only, and farnell asks $0.64 per transistor if I buying a bunch of 25+ units.
And I found strange behaviour: while voltage across CE of Q1 and Q2 are equal at one channel, the second channel shows 0.5V imbalance, and the output of that channel is always 2.9-3.2mV(even with the servo connected) while another channel 0.1-0.4mV ...
Looks like something was not matched carefully enough ?
Another question - do I have to use separate radiator for each channel ?
I'm just trying to get my amp running properly. I'm not looking for the way to pervert , or to "improve" excellent circuit of Dr Marsh.