Усилитель ULTIMATE

[MAXIM_A]

Господа-товарищи соорудили классный усилок, слабо попробывать повторить?
Я хочу, начну с оконечника.
The project scope was defined in the early 2007.
Key performance requirements include:
• Class AB topology, 100W into 8 ohm and 200W into 4 ohm
• Bandwidth of minimum 20Hz to 20KHz +/-0.1dB
• Under 1 ppm (80KHz bandwidth) THD (that is, 0.0001%)
at all frequencies between 20Hz and 20KHz and output
powers between 1W and 200W
• Under 1 ppm 19+20KHz CCIF IMD (that is, 0.0001%) at all
output powers between 1W and 200W
http://www.synaesthesia.ca/index.html

Печатная плата ВК:

[mixxxxxer]

http://www.budind.com/view.php?part=ch-rm

Offтопик:
Не, ну не лафа ли этим карапузикам? Тока подумал - и вот те рэковый корпус за копей.. тьфу, центы.. Мля, всем - работать, работать и работать.. Особенно китайцам (у них лучше всего это получается)..

[MAXIM_A]

Offтопик:
Не, ну не лафа ли этим карапузикам? Тока подумал - и вот те рэковый корпус за копей.. тьфу, центы.. Мля, всем - работать, работать и работать.. Особенно китайцам (у них лучше всего это получается)..

Как то странно - такой топовый усилок, а кондёры так себе: "Jamicon"?
А в остальном всё клёво.

[Анатолий_В]

А в остальном всё клёво.

Плохо продуманное расположение блоков. Необходимость вентиляторов. Если уж такой топ, то можно было бы и радиаторы взять приличные с вертикальными ребрами и не накрывать их корпусом. Или накрыть как в простеньких Акуфейсах и Денсенах.

[Nick]

Плохо продуманное расположение блоков. Необходимость вентиляторов. Если уж такой топ, то можно было бы и радиаторы взять приличные с вертикальными ребрами и не накрывать их корпусом. Или накрыть как в простеньких Акуфейсах и Денсенах.

Это было продиктовано финансовыми ограничениями на данный прототип.

[32768]

Это было продиктовано финансовыми ограничениями на данный прототип.

схема раскидана по корпусу, много длинных проводов но по измерениям не плохая - причина в схемотехнике. Выходной каскад с низкими искажениями. И есть каскады с высоким выходным сопротивлением (токовые).

[Nick]

схема раскидана по корпусу, много длинных проводов но по измерениям не плохая - причина в схемотехнике. Выходной каскад с низкими искажениями. И есть каскады с высоким выходным сопротивлением (токовые).

Да схема не плохая, но больно уж сложный УН. Думаю стаким ВК можно было сделать и попроще. А результат низких искажений это результат тщательной настройки положения проводов, что сводит повторяемость без приборов к 0, а с приборами к долгой и кропотливой настройке. По сути же это обычный усилитель с ООС.

Добавлено через 1 час 39 минут
syn08, I am just wondering, are you trying to read these russian*s posts in some kind of online translator ? I*ve seen it in translate.google.com, it is really funny

[syn08]

Да схема не плохая, но больно уж сложный УН. Думаю стаким ВК можно было сделать и попроще. А результат низких искажений это результат тщательной настройки положения проводов, что сводит повторяемость без приборов к 0, а с приборами к долгой и кропотливой настройке. По сути же это обычный усилитель с ООС.

Добавлено через 1 час 39 минут
syn08, I am just wondering, are you trying to read these russian*s posts in some kind of online translator ? I*ve seen it in translate.google.com, it is really funny

I do, ocasionally, in Google or Babelfish, and it*s indeed funny - but I usually get the idea. You guys are using some sort of colloquial language (or jargon) Automatic language translations from publications are much better!

To adress a few concerns expressed lately:

- I finally sourced (for an acceptable cost) and in the process of replacing the heatsinks with another version having vertical fins.

- There is a fan control circuit, the fans are starting only when the case temperature reaches 55C, which so far never happened while listening to (very loud) music. From this perspective, think of the fans more as a "safety net".

- Indeed, this amp is ultimately based on negative feedback (aren*t they all, one way or another?) However, it is not a regular global feedback amp. The key to the ultimate performance are:

a) Prof. Cherry*s nesting differentiating feedback loops (in a slightly modified form). This can be viewed as a method of frequency compensation that allows increasing the loop gain at high frequencies (while still keeping the amp unconditionally stable) well beyond what you could get by using a single or dual pole compensation schema.

b) The common mode control loop concept, which allows a unique combination of bias stability and high gain in the VAS.

c) Prof. Hawksford error correction concept. This can be viewed and analyzed as negative feedback, however there are a few properties that make this circuit special; it allows you to minimize the distortions at a certain (high) frequency, usually where the loop gain starts to drop. This allows to squeeze another few dB*s of high frequency loop gain and another (small, but important!) drop in the closed loop distortions.

d) The non-slewing design approach. The rise and fall times at the output are controlled by the time constants of the NDFL networks and the response of the Zobel network, and are not limited by a lack of drive current (from any stage)! This minimizes the TIM, which is in way to many amps (commercial and hobby) responsible for the poor sound.

Yes, this amp is indeed a monster in complexity and size. A 10ppm amp could be built probably with less than 10% of the parts we used in the PGP. As also stated on the web site, this amp was build as an engineering proof of concept. I am not claiming in any way, shape or form that the 1ppm performance brings anything really special or outstanding to the listening experience. If I would be in the (unfortunately very sad looking) audio industry today, I wouldn*t even think in trying to manufacture and sell the PGP amp.

My challenge: if anybody could build and measure (no, simulations only do not count!) an amp with the same performanance but significantly simpler, I would be the first to congratulate them!

Thanks,

[Nick]

This can be viewed as a method of frequency compensation that allows increasing the loop gain at high frequencies (while still keeping the amp unconditionally stable) well beyond what you could get by using a single or dual pole compensation schema.

syn08, what is the open loop gain (feedback depth) of your amp on 20kHz?

Добавлено через 3 минуты

My challenge: if anybody could build and measure (no, simulations only do not count!) an amp with the same performanance but significantly simpler, I would be the first to congratulate them!

I believe it is not really a problem with а deep feedback (even with 2 pole compensation) and appropriate implementation. I have about 70dB feedback depth on 20кГц and less than 0.0004% on 10kHz in real amp, and it is not a limit. And this is only with 2 opamps AD8065 and LM3886 The problems that I have now mostly caused by implementation (inductive coupling of power lines and input).

[sia_2]

My challenge: if anybody could build and measure (no, simulations only do not count!) an amp with the same performanance but significantly simpler, I would be the first to congratulate them!
Thanks,

I have such amplifier since 2000 - it is slightly modified my "Superlinear" Amp. Key features - very deep and wideband feedback and high-power output stage with low inherent distortion, plus carefully traced PCB. Open-loop distortion about 0.3% on heavy load, feedback at 25 KHz about 90 dB. Main circuits are clean and simply, but carefully optimized on speed/linearity aspects.
Early version of fhis amplifier was published 1999 in russian Radio magazine.
I easy can measure it now on AP2 (especially DIM100 test), old measurements were performed on AP1, which didn*t have enough resolution. Single-tone measurements at 20 KHz, -2 dB from soft clipping, give THD results about -117...-125 dB (depended on load condition - real/resistive load).

Here are topic about this amp - https://forum.vegalab.ru/showthread.php?t=3273 .
Small-signal step resonce photos placed below. (2.1V, 0.5 and 0.05uS per division, >250MHZ analog scope).

[Nick]

Also I know one desing with 130dB of feedback depth at 20kГц. But it uses feddworward technique as well.

Добавлено через 4 минуты

I easy can measure it now on AP2 (especially DIM100 test), old measurements were performed on AP1, which didn*t have enough resolution. Single-tone measurements at 20 KHz, -2 dB from clipping, give THD results about -117...-125 dB

Добавлено через 2 минуты
syn08, you’ve probably seen a PCB of this amp, I posted a link on previous page.

Добавлено через 4 минуты
Probably it would be interesting for you too - 130dB of feedback depth at 20kГц amp :https://forum.vegalab.ru/showthread....E0%E2%F0%E8%ED

[syn08]

syn08, what is the open loop gain (feedback depth) of your amp on 20kHz?

Добавлено через 3 минуты
I believe it is not really a problem with а deep feedback (even with 2 pole compensation) and appropriate implementation. I have about 70dB feedback depth on 20кГц and less than 0.0004% on 10kHz in real amp, and it is not a limit. And this is only with 2 opamps AD8065 and LM3886 The problems that I have now mostly caused by implementation (inductive coupling of power lines and input).

I don*t know for sure (I am not aware of any direct method of measuring the open loop gain and the phase margin) but an estimate is 75dB for the loop gain (let*s use the right terminology, this is the A*Beta product) of the front end and the output stage (with the error correction negative feedback loop open, the gain in the output stage is generated by the positive feedback loop).

[Nick]

So I believe there are many ways to build a ultra low distortion amp, but its implementation is more important than design, especially if you want to make it reproducible without THD meter.

[syn08]

I have such amplifier since 2000 - it is slightly modified my "Superlinear" Amp. Key features - very deep and wideband feedback and high-power output stage with low inherent distortion. Open-loop distortion about 0.3% on heavy load, feedback at 25 KHz about 90 dB. Main circuits are clean and simply, but carefully optimized on speed/linearity aspects.
Early version of fhis amplifier was published 1999 in russian Radio magazine.
I easy can measure it now on AP2 (especially DIM100 test), old measurements were performed on AP1, which didn*t have enough resolution. Single-tone measurements at 20 KHz, -2 dB from clipping, give THD results about -117...-125 dB (depended on load condition - real/resistive load).

Here are topic about this amp - https://forum.vegalab.ru/showthread.php?t=3273 .
Small-signal step resonce photo placed below. (2.1V|0.5uS per division, >250MHZ analog scope).

90dB at 25KHz? This is absolutely amazing. On top of my head, this makes for a gain-bandwidth product of about 750MHz!!! Are you sure this is the loop gain? How is this amp compensated? Is it Class AB? Could you share the schematics?

[Nick]

90dB at 25KHz? This is absolutely amazing. On top of my head, this makes for a gain-bandwidth product of about 750MHz!!! Are you sure this is the loop gain? How is this amp compensated? Is it Class AB?

I belive the small singnal amp bandwith is about 5-7MHz. The correction is not a single pole of course. Am I right, Sergey ?

[syn08]

Also I know one desing with 130dB of feedback depth at 20kГц. But it uses feddworward technique as well.

Добавлено через 4 минуты


Добавлено через 2 минуты
syn08, you’ve probably seen a PCB of this amp, I posted a link on previous page.

Добавлено через 4 минуты
Probably it would be interesting for you too - 130dB of feedback depth at 20kГц amp :https://forum.vegalab.ru/showthread....E0%E2%F0%E8%ED

Guys, I think we should step back for a second in throwing numbers and settle some definitions. How exactly do you define "130dB of feedback depth at 20kГц"?

Also, what exactly you mean by "feedforward"? Feed forward compensation?

[Nick]

syn08, I belive it would be interesting to you too: https://forum.vegalab.ru/attachment....9&d=1192665399

[sia_2]

90dB at 25KHz? This is absolutely amazing. On top of my head, this makes for a gain-bandwidth product of about 750MHz!!! Are you sure this is the loop gain? How is this amp compensated? Is it Class AB? Could you share the schematics?

It* s No problem. Class AB, but very power output stage. Full circuits and a lot of info about it you can find on link above. Design based on old conception (Widerhold 1978) - input opamp with very linear input stage + discrete symmetrical VAS + output stage. Correction complex, multizeros/multipoles. Unity loop gain frequency in range 6...10 MHz in different versions, phase margin appr. 40-50 deg. Spice/MC model also implemented.

See PCB photo. ~70% of area - Power supply and output stage.

All this amp - are nothing in comparation with amps for long-distance analog multichannel telephony transmission systems. IMD of amps for 10800-channel system, must be less than -110 dB @ 60 MHz at nominal level ! And such amps was mass-produced by many telecom companies many year ago, in early 1970-h.

[Nick]

How exactly do you define "130dB of feedback depth at 20kГц"?

For example, take a look at this graph, and imagine that unity gain is much higher than 1MHz (for instance - 10Mhz, it could be easily reached by feedforward correction) https://forum.vegalab.ru/attachment....3&d=1202981975 (feedback depth vs freq)

[syn08]

Also I know one desing with 130dB of feedback depth at 20kГц. But it uses feddworward technique as well.

Добавлено через 4 минуты


Добавлено через 2 минуты
syn08, you’ve probably seen a PCB of this amp, I posted a link on previous page.

Добавлено через 4 минуты
Probably it would be interesting for you too - 130dB of feedback depth at 20kГц amp :https://forum.vegalab.ru/showthread....E0%E2%F0%E8%ED

Ok, the schematic in LITAVRIN-2.pdf is a feed forward error correction amp, as in the Quad-405 (the only commercial amp I know that ever implemented this method). Feedforward error correction is NOT negative feedback and I*m afraid the comparation with a negative feedback amp is not apples to apples.

[Nick]

Feedforward error correction is NOT negative feedback and I*m afraid the comparation with a negative feedback amp is not apples to apples

He uses feedforward correction to extend bandwith of amp in order to make very deep feedback at 20kHz.