Thanks, 80dB of feedback and 96dB Aol a real tube power op-amp. You can hear in on youtube too!
hI,
Not every product is reviewed in all publications, not all publications are posted on line, I spoke of products in general. And maybe you need to buy the occasional print magazine.
Many reviews are re-published on AMR's website as well, so you could look there.
I generally no longer read reviews of my stuff (unless someone makes me to). I can only retain attention so long and rave reviews and awards get boring after first dozen or so, which I had pretty much complete before we even started AMR...
Ciao T
Not every product is reviewed in all publications, not all publications are posted on line, I spoke of products in general. And maybe you need to buy the occasional print magazine.
Many reviews are re-published on AMR's website as well, so you could look there.
I generally no longer read reviews of my stuff (unless someone makes me to). I can only retain attention so long and rave reviews and awards get boring after first dozen or so, which I had pretty much complete before we even started AMR...
Ciao T
Hi,
Only at 1KHz.
At 15Hz and 30KHz we loose over 20dB NFB. This is largely because the high OL Gain is achieved using positive feedback from the output stage and is thus subject to a number of limitations.
Also, by the time we have enough gain for the Amp to be useful we have only only a bit over 60dB NFB at 1KHz and a bit over 40dB NFB at 15Hz/30KHz...
The number of compensation networks shows the problems of this approach.
I tried a similar principle in my modifications for Shanling SP-80's. It is a very decent tube Amp that ejit insited to cripple by throwing in a Op-Amp/BB Electronic Volume Control "Preamp" in. The best mod is to simply remove this board...
I did some extreme mods on these Amp's. In those I tried raising Feedback Factor by using Positive Feedback. I got as far as > 70W from a pair of EL34's and very low distortion at 1KHz.
But it sounded better when I took the positive feedback out thus reducing the NFB at least 20dB and widening the open loop bandwidth.
The final version used a feedback loop from the output anodes to the grids of the drivers and local feedback to the output stage cathodes only, output transformer and input stage stayed out of the loop.
It sounded substantially better that way, of course THD was now up quite a bit, but not unreasonably high, it retained around 30dB of negative feedback which was not varying significantly from 10Hz to 40KHz.
Actually, THD at normal levels (not near output stage clipping) was now mostly down to the input stage, which I had degenerated a lot, but I needed SOME gain in the Amp, so there where limits...
Ciao T
Only at 1KHz.
At 15Hz and 30KHz we loose over 20dB NFB. This is largely because the high OL Gain is achieved using positive feedback from the output stage and is thus subject to a number of limitations.
Also, by the time we have enough gain for the Amp to be useful we have only only a bit over 60dB NFB at 1KHz and a bit over 40dB NFB at 15Hz/30KHz...
The number of compensation networks shows the problems of this approach.
I tried a similar principle in my modifications for Shanling SP-80's. It is a very decent tube Amp that ejit insited to cripple by throwing in a Op-Amp/BB Electronic Volume Control "Preamp" in. The best mod is to simply remove this board...
I did some extreme mods on these Amp's. In those I tried raising Feedback Factor by using Positive Feedback. I got as far as > 70W from a pair of EL34's and very low distortion at 1KHz.
But it sounded better when I took the positive feedback out thus reducing the NFB at least 20dB and widening the open loop bandwidth.
The final version used a feedback loop from the output anodes to the grids of the drivers and local feedback to the output stage cathodes only, output transformer and input stage stayed out of the loop.
It sounded substantially better that way, of course THD was now up quite a bit, but not unreasonably high, it retained around 30dB of negative feedback which was not varying significantly from 10Hz to 40KHz.
Actually, THD at normal levels (not near output stage clipping) was now mostly down to the input stage, which I had degenerated a lot, but I needed SOME gain in the Amp, so there where limits...
Ciao T
I compare that time to modern days, John. And my point was, for audio different topologies are needed than opamps.
I like this point, about multiple feedbacks. And 10 MHz unity gain.
Last edited:
Hi,
Sure. However the driver stage in the KH is pentode based, not triode as Mac...
If all the gains in the output stage where truly unity the effective impedance of the driver stage load would become infinity, leaving us only with pentode's anode impedance and Gm to determine gain.
Using quickly the 6U8 Datasheet numbers this is 5mA/V and 200K, which will give a gain of 1000, compared to 190 for a direct 47K Anode load.
If the output stage gain drops to 0.7 (-3dB Points) the driver stage gain drops to under 400, so overall we loose theoretically 11dB open loop gain at the -3dB points of the combination of the output tubes and transformer.
Any further and our artificially inflated gain rapidly disappears, at the -6dB Point we already had 16dB loop gain evaporate.
By comparison the Triode drivers in the Mac Amps at best approach their Mu for gain, so more stages are used but loop gain remains much more constant with frequency...
Ciao T
Sure. However the driver stage in the KH is pentode based, not triode as Mac...
If all the gains in the output stage where truly unity the effective impedance of the driver stage load would become infinity, leaving us only with pentode's anode impedance and Gm to determine gain.
Using quickly the 6U8 Datasheet numbers this is 5mA/V and 200K, which will give a gain of 1000, compared to 190 for a direct 47K Anode load.
If the output stage gain drops to 0.7 (-3dB Points) the driver stage gain drops to under 400, so overall we loose theoretically 11dB open loop gain at the -3dB points of the combination of the output tubes and transformer.
Any further and our artificially inflated gain rapidly disappears, at the -6dB Point we already had 16dB loop gain evaporate.
By comparison the Triode drivers in the Mac Amps at best approach their Mu for gain, so more stages are used but loop gain remains much more constant with frequency...
Ciao T
I appreciate including Burwen's tube power amp, here. Perhaps it should be a DIY project. I see plenty of hi end techniques used.
I made something along the same line for Sound Technology almost 30 years ago. I lost the original schematic 20 years ago in the Oakland Hills firestorm, but I would like to find what they finally used in their later series beyond the ST1710, that they made in the early '80's. I used 2 pole feedback, mostly fet, circuit closer to Levinson JC-2 phono stage, multiple cascode, and a measured distortion into 600 ohms of something like 0.001% at 100KHz and below. They hired me, because they could not find any IC that could do it at the time. IN THIS CASE, for instrumentation, high feedback is mandatory. The same as it was for Dick Burwen and his op amp designs.
They only real difference between Dick Burwen and me was that he continued to believe in op amps as the best solution for audio, and I changed to more complementary topologies in order to further reduce the open loop distortion and ultimately, with less emphasis on using high negative feedback as a cure-all, due to my A-B experiences with the Grateful Dead, comparing standard IC's with indifferent coupling caps, selected IC's (of the same type) with quality caps, and class A comp differential jfet transconductance-amps with high open loop bandwidth.
IF selected IC's with quality caps, designed and built by a partnership between Dick Burwen and Mark Levinson, before I came on the scene, worked subjectively perfectly, it would have been instantly purchased by the Grateful Dead, after the 1973 the Watkins Glen gig. The mixer was beautiful, functional, reliable, and well within their budget.
What actually happened is that Mark Levinson found that my JC-1 phono preamp was quieter than Dick Burwen's effort, and sounded good too. Then, Mark Levinson tried experimenting with the comp diff jfet amps (with 100V/us), replacing the Burwen modules with the GD modules (same pin out) and decided to make the Levinson JC-2 with the GD modules, asking me to add a quieter phono stage. Later, I was asked to design modules to replace every analog circuit originally built by Dick Burwen for the LMP-2 and the LNC-2, etc.
Of course, it could have been our imagination, but TAS got ahold of both designs, the LNP-2 designed by Dick Burwen, and the JC-2 designed by Burwen, Mark and me, and the JC-2 won. The rest is history.
I made something along the same line for Sound Technology almost 30 years ago. I lost the original schematic 20 years ago in the Oakland Hills firestorm, but I would like to find what they finally used in their later series beyond the ST1710, that they made in the early '80's. I used 2 pole feedback, mostly fet, circuit closer to Levinson JC-2 phono stage, multiple cascode, and a measured distortion into 600 ohms of something like 0.001% at 100KHz and below. They hired me, because they could not find any IC that could do it at the time. IN THIS CASE, for instrumentation, high feedback is mandatory. The same as it was for Dick Burwen and his op amp designs.
They only real difference between Dick Burwen and me was that he continued to believe in op amps as the best solution for audio, and I changed to more complementary topologies in order to further reduce the open loop distortion and ultimately, with less emphasis on using high negative feedback as a cure-all, due to my A-B experiences with the Grateful Dead, comparing standard IC's with indifferent coupling caps, selected IC's (of the same type) with quality caps, and class A comp differential jfet transconductance-amps with high open loop bandwidth.
IF selected IC's with quality caps, designed and built by a partnership between Dick Burwen and Mark Levinson, before I came on the scene, worked subjectively perfectly, it would have been instantly purchased by the Grateful Dead, after the 1973 the Watkins Glen gig. The mixer was beautiful, functional, reliable, and well within their budget.
What actually happened is that Mark Levinson found that my JC-1 phono preamp was quieter than Dick Burwen's effort, and sounded good too. Then, Mark Levinson tried experimenting with the comp diff jfet amps (with 100V/us), replacing the Burwen modules with the GD modules (same pin out) and decided to make the Levinson JC-2 with the GD modules, asking me to add a quieter phono stage. Later, I was asked to design modules to replace every analog circuit originally built by Dick Burwen for the LMP-2 and the LNC-2, etc.
Of course, it could have been our imagination, but TAS got ahold of both designs, the LNP-2 designed by Dick Burwen, and the JC-2 designed by Burwen, Mark and me, and the JC-2 won. The rest is history.
Thorsten, how did you get this?
"Noise (A-weighted): below –145dBV (<0.056µV) equivalent input noise."
Abbingdon Music Research PH-77 Phono Equaliser Specifications | Stereophile.com
Compared to Atkinson's measurements
"For example, the A-weighted S/N ratios for the right channel in MM mode were 82dB with 48dB of gain and 68.7dB with 30dB of gain; for the right channel in MC mode, they were 79.1dB with 72dB of gain and 72.4dB with 54dB of gain. "
Abbingdon Music Research PH-77 Phono Equaliser Measurements | Stereophile.com
Now you might probably understand why I asked for trusted measurements rather than verbal proclamations.
"Noise (A-weighted): below –145dBV (<0.056µV) equivalent input noise."
Abbingdon Music Research PH-77 Phono Equaliser Specifications | Stereophile.com
Compared to Atkinson's measurements
"For example, the A-weighted S/N ratios for the right channel in MM mode were 82dB with 48dB of gain and 68.7dB with 30dB of gain; for the right channel in MC mode, they were 79.1dB with 72dB of gain and 72.4dB with 54dB of gain. "
Abbingdon Music Research PH-77 Phono Equaliser Measurements | Stereophile.com
Now you might probably understand why I asked for trusted measurements rather than verbal proclamations.
dB referenced to what? There's a 60dB (1000x) difference in the reported numbers.
The issue is noise referenced to input vs noise referenced to output.
- Status
- Not open for further replies.