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:

[syn08]

Hi!

I am glad to greet you on the forum of Russian audio community.
Your the Ultimate audio amp had attract a lot of attention because of its perfect characteristics. (At least quantitative and I hope qualitative too. )

Some questions arose simultaneously.
Main question of course is, how is possible to adjust the amplifier without your huge and expensive set of the measurement devices? Is it possible to do using an oscilloscope and audio freq. generator?

Second is, what was the design of the amp practically? Because for such ultimate low distortion the design plays crucial role. Some our experts confused about possibility to reach so high properties with the proposed design (at least how it looks on the photos).

third
After test of the amp I understood that you was not too happy about used MOSFET in the EC OPS. What may you propose now for trying at least: a BJT or other type MOSFET?


fourth I red the paper of prof. Howksford about error correction in output stage and have one question. Is this error correction something to do with the feed-forward error correction? Or, it is a pure feedback solution.

1. It all depends on the level of performance you are looking for. For the ultimate performance, I am afraid a spectrum analyzer (could be a professional sound card (as the Lynx TWO) plus some freeware software) and a high end THD meter are mandatory. Certainly, only a audio generator and a scope are not enough.

2. I*m not sure I understand the question. Is it about the electric design (schematics), or about the implementation design (PCB, case, etc...).

3. I am not happy with using lateral MOSFETs, at least the 2SK1048/2SJ162 I can get here for a reasonable price. Reason is low transconductance and not enough current capability to drive 4 ohm speakers. The next thing I am planning to try are the 2SK1530/2SJ201 vertical MOSFETs. Using bipolars in the output stage is possible (I have experimented with as well, using a single pair of 2SC5200/2SA1943). The problem with bipolars is that for the ultimate performance, everything except for the output devices should run in Class A. This means you need a pretty high bias current in the drivers and certainly on board heatsinks won*t be enough to reasonably cool them down. This is the reason why I can*t recommend the current implementation (with bipolars) for more than 50W output power. Certainly, by modifying the PCB layout, you can go up.

4. Hawksford error correction is ultimately a negative feedback implementation. The difference to other implementation is that the open loop gain is created by a separate positive feedback loop. Minimizing the THD is essentially balancing the positive and negative feedback loops. Feedforward error correction is a completely different story.

[митяич]

тема интересная, жаль стала англоязычная
может уважаемый Мескалито возьмется переводить посты?
просьба к европейскому рысю, если он вдруг заглянет сюда, не указывать на учебник английского

[Meskalito]

1. It all depends on the level of performance you are looking for. For the ultimate performance, I am afraid a spectrum analyzer (could be a professional sound card (as the Lynx TWO) plus some freeware software) and a high end THD meter are mandatory. Certainly, only a audio generator and a scope are not enough.

2. I*m not sure I understand the question. Is it about the electric design (schematics), or about the implementation design (PCB, case, etc...).

3. I am not happy with using lateral MOSFETs, at least the 2SK1048/2SJ162 I can get here for a reasonable price. Reason is low transconductance and not enough current capability to drive 4 ohm speakers. The next thing I am planning to try are the 2SK1530/2SJ201 vertical MOSFETs.

4. Using bipolars in the output stage is possible (I have experimented with as well, using a single pair of 2SC5200/2SA1943). The problem with bipolars is that for the ultimate performance, everything except for the output devices should run in Class A. This means you need a pretty high bias current in the drivers and certainly on board heatsinks won*t be enough to reasonably cool them down. This is the reason why I can*t recommend the current implementation (with bipolars) for more than 50W output power. Certainly, by modifying the PCB layout, you can go up.

1. Clear. I suppose it is most critical point in this story...

2. Question was about the implementation design (PCB, case and so on) Could you give more details about your design. (not immediately of course but in future)

3. Clear.

4. Does it mean that BJTs not work in AB mode with Howksford error correction?
Or, you speaking about all transistors except the output one, and it need a high power dissipation that out of reasonable design of the amp.

Добавлено через 2 минуты

тема интересная, жаль стала англоязычная
может уважаемый Мескалито возьмется переводить посты?
просьба к европейскому рысю, если он вдруг заглянет сюда, не указывать на учебник английского

Обязательно переведу, если до меня никто не сделает это...
Сейчас вынужден покинуть мир виртуальный и окунуться в реальный. В буквальном смысле - пошёл в бассейн, поплаваю. ))

[syn08]

1. Clear. I suppose it is most critical point in this story...

2. Question was about the implementation design (PCB, case and so on) Could you give more details about your design. (not immediately of course but in future)

3. Clear.

4. Does it mean that BJTs not work in AB mode with Howksford error correction?
Or, you speaking about all transistors except the output one, and it need a high power dissipation that out of reasonable design of the amp.

Добавлено через 2 минуты


Обязательно переведу, если до меня никто не сделает это...
Сейчас вынужден покинуть мир виртуальный и окунуться в реальный. В буквальном смысле - пошёл в бассейн, поплаваю. ))

2. Go ahead and ask. I am also ready to receive any amount of constructive criticism

4. I*m talking about all transistors except the outputs. Yes, the thermal management has to change for handling high power with bipolars.

Thanks,

[Анатолий_В]

А почему этот ВК слабее, у него ведь меньши собственные искажения?

Я имел в виду то, что плата в Вашем Зуевике имеет петли меньшего размера.

[MAXIM_A]

Я имел в виду то, что плата в Вашем Зуевике имеет петли меньшего размера.

Да, пожалуй, надо попробывать ещё вариантик. Уж очень заманчиво небольшими средствами улушить исходный вариант Зуева.

[Nick]

1. It all depends on the level of performance you are looking for. For the ultimate performance, I am afraid a spectrum analyzer (could be a professional sound card (as the Lynx TWO) plus some freeware software) and a high end THD meter are mandatory. Certainly, only a audio generator and a scope are not enough.

2. I*m not sure I understand the question. Is it about the electric design (schematics), or about the implementation design (PCB, case, etc...).

3. I am not happy with using lateral MOSFETs, at least the 2SK1048/2SJ162 I can get here for a reasonable price. Reason is low transconductance and not enough current capability to drive 4 ohm speakers. The next thing I am planning to try are the 2SK1530/2SJ201 vertical MOSFETs. Using bipolars in the output stage is possible (I have experimented with as well, using a single pair of 2SC5200/2SA1943). The problem with bipolars is that for the ultimate performance, everything except for the output devices should run in Class A. This means you need a pretty high bias current in the drivers and certainly on board heatsinks won*t be enough to reasonably cool them down. This is the reason why I can*t recommend the current implementation (with bipolars) for more than 50W output power. Certainly, by modifying the PCB layout, you can go up.

4. Hawksford error correction is ultimately a negative feedback implementation. The difference to other implementation is that the open loop gain is created by a separate positive feedback loop. Minimizing the THD is essentially balancing the positive and negative feedback loops. Feedforward error correction is a completely different story.

Hello syn08, and welcome to our russian*s diy audio board.
Here is the translation of your post:
1. Всё зависит от уровня характеристик как их вы хотите. Для наилучших характеристик нужно иметь анализатор спектра (можно очень хорошую звуковую карту (Lynx TWO) и бесплатный софт) и хай-энд измеритель нелинейных искажений. Несомненно, только аудио генератор и осциллограф не достаточно.
2. Я не уверен что понял ваш вопрос. Это про схемотехнику или имплементацию (платы, корпус..)
3. Мне не нравятся латеральные мосфеты, по крайней мере 2SK1048/2SJ162, которые я тут могу купить по нормальной цене. Причина это низкая крутизна и недостаточный ток чтобы драйвить 4-х омные колонки. Я планирую попробовать 2SK1530/2SJ201 вертикальные MOSFET-ы. Биболяры в на выходе тоже возможно поставить (я эксперементировал с ними тоже, с парой 2SC5200/2SA1943).
Проблема с биполярами в том, что для максимального качества, нужно чтобы всё, кроме выходных, работало в классе А. Это значит что нужно чтобы предвыходные имели очень большой ток, и радиаторов на плате для этого не хватит. Это и есть причина, почему я немогу рекомендовать их для мощностей более 50 ватт. Конечно, с модифицированной платой, можно сделать помощнее.
4. Коррекция Хаксфорда в конечном счёте отрицательная обратная связь. Разница от других имплементаций лишь в том что усиление без ООС создано отдельной положительной обратной связью. Уменьшение же искажений происходит посуществу при балансе положительной и отрицательной обратных связей. Коррекция вперёд это совсем другая история.

Добавлено через 18 минут

The problem with bipolars is that for the ultimate performance, everything except for the output devices should run in Class A.

With bipolars on the output, theirs bases currents make the pre-outputs run less linear any way. So I’m afraid that in order to achieve the same performance ( which you have with mosfet*s ) everything but output devices should run not in Class A but in deep Class A. But this is for low frequencies. Hawksford actually indirectly mentioned that in his paper. Did you try darlingtons ?

Добавлено через 1 час 2 минуты
syn08,
Because now I have some experience with utra low distortion amp (https://forum.vegalab.ru/showthread....3106&page=122), I have a few questions for you:
1. Why don*t you use huge shunt capacitors (>100uF) on the output stage PCB?
2. Did you experience any harmonic distortions caused by magnetic fields from output stage or power supply (I see the C31 is very close to the output stage) ?
3. What kind of material do you use for back panel and what kind of binding posts?
4. Why don*t you use separate power supply for the input stage, and part of output, which runs in Class A ?
Thank you, Nick

[syn08]

Добавлено через 18 минут
With bipolars on the output, theirs bases currents make the pre-outputs run less linear any way. So I’m afraid that in order to achieve the same performance ( which you have with mosfet*s ) everything but output devices should run not in Class A but in deep Class A. But this is for low frequencies. Hawksford actually indirectly mentioned that in his paper. Did you try darlingtons ?

---> You are right about the current required for the drivers. Unfortunately no, I haven*t tried Darlingtons. Another issue that impacts the use of bipolars (that I haven*t really looked for an optimal solution) is the base stored charge evacuation. It*s simple in a standard EF output stage, I*m not so sure is so simple in the Hawksford EC circuit.

Добавлено через 1 час 2 минуты ,
Because now I have some experience with utra low distortion amp (https://forum.vegalab.ru/showthread....3106&page=122), I have a few questions for you:
1. Why don*t you use shunt capacitors (>100uF) on output stage PCB?
2. Did you experience any harmonic distortions caused by magnetic fields from output stage or power supply?
3. What kind of material do you use for back panel and what kind of binding posts?

---> I*ll look at that thread - hopefully I*ll be able to get a grasp.

1. I*m not sure I understand - there are 100uF caps in parallel with film caps on each output stage board and on each supply rail. Why not more than 100uF? Simply because I have found 100uF is all that*s required to reach the target performance (<100ppm open loop).

2. Oh, yes, tons of issues! Lots of construction details (as mentioned on the web site) are targeted to avoiding 2nd harmonic distortions induced by the magnetic fields. Here are some: a) power wiring is vertical, perpendicular to signal wiring, which is horizontal. That*s the reason for the case boards layout, the output stage and front end boards are installed on top of the power supply boards. b) power wiring is kept short, always twisted with the matching return ground, c) tweaking the power wiring positions is the last adjustment that needs to be done before closing the case lid. That*s one of the reasons why you need a good THD meter. You can easily get >100ppm THD20 if the power wiring positions are not optimal. d) the output wiring is yet again physically away from the power wiring and as much as possible perpendicular to the input and feedback wiring (this is the reason for the odd position of the output yellow wires, as seen on the web site) e) all power ground connections are star*red at the power supply boards f) Separate grounds, floating outputs, etc... all described on the web site.

3. Everything is aluminum 1/8" thick. I have to admit I was never able to measure any impact of a 1/4" or thicker aluminum panel and the Maxwell guy that was invoked here is supporting this lack of impact. The output binding posts are dual: a set of high current goldened copper (about 1/2" in dia) and a set of Neutrik SPEAKON connectors.

Thank you, Nick[/QUOTE]

Thanks,

[Nick]

syn08, I*m sorry, but it seems I edited my post with the questions when you were answering. But you*ve answered on all my questions, except the last one.
Thank you.

Добавлено через 7 минут

3. Everything is aluminum 1/8" thick. I have to admit I was never able to measure any impact of a 1/4" or thicker aluminum panel...

Yes, I belive you can have a problem only with the ferromagnetic material (iron containing, steel).

[syn08]

4. Why don*t you use separate power supply for the input stage, and part of output, which runs in Class A ?

The original design was with separate power supplies. However, I was unable to source proper affordable (price wise) toroids. On the other side, preliminary measurements showed that using a common power supply is possible (performance wise), so this was the final implementation.

The output MOSFETs are NOT running in Class A, but only the drivers and the error correction stage.

Thanks,

[Nick]

---> I*ll look at that thread - hopefully I*ll be able to get a grasp.

I can translate it for you (some part of it, not the whole thread of course ). I was going to post my amp on diyaudio, but have a little time.

Добавлено через 2 минуты

The output MOSFETs are NOT running in Class A, but only the drivers and the error correction stage.

Yes, I know, I mean only the part of the output stage: drivers, which run in Calss A, and the error correction stage. Sorry if it was not clear.

Добавлено через 17 минут

2. Oh, yes, tons of issues! Lots of construction details (as mentioned on the web site) are targeted to avoiding 2nd harmonic distortions induced by the magnetic fields. Here are some: a) power wiring is vertical, perpendicular to signal wiring, which is horizontal. That*s the reason for the case boards layout, the output stage and front end boards are installed on top of the power supply boards...

I believe you can eleminate a lot of problems, by placing the power supply capacitors, directly on the output stage PCB. It should dramatically reduce the magnetic fields from the Class AB output stage (from power line, which has a rectified signal).

Добавлено через 5 минут
Like here http://radikal.ru/F/i049.radikal.ru/...a85de.jpg.html

Добавлено через 9 минут

Simply because I have found 100uF is all that*s required to reach the target performance (<100ppm open loop).

The problem with these capacitors you can see only with closed loop. It happens because high frequencies currents flow in the output stage power lines (especially with the small values of these capacitors) and it induces relatively a small amount of distortion (through the inductive coupling of the power lines and the input) which will not be reduced by feedback.

[32768]

Если он так уж невелик, тогда небыло бы и смысла ставить подстроечную ёмкость. Это очень сильно зависит от усилителя и уровня гармоник генератора, тут надо считать.

смысл только один - чем уже динамический диапазон, тем проще и точнее можно измерять. Если после дополнительного усиления сигнал основной гармоники не мешает, то емкость не нужна.

Фотографии собранного в корпус PGP - на алюминий не похоже

[MAXIM_A]

Фотографии собранного в корпус PGP - на алюминий не похоже

Корпус - оцинкованная сталь, а откуда такие фото?
Не пойму что за плата в центре? Толи пред, или защита.

[dekko]

Корпус - оцинкованная сталь

а на одном канале нет вентилятора? это как?

[MAXIM_A]

а на одном канале нет вентилятора? это как?

Ты его просто не видишь, увеличь и посмотри внимательно...

[32768]

Корпус - оцинкованная сталь, а откуда такие фото?

Откуда: -оттуда!(С)
Очень похоже на сталь.

[32768]

еще фото

[Nick]

еще фото

Честно скажу, мне очень не нравится местоположение большого жёлтенького конденсатора, это входной кондёр, и он очень рядом с ВК который без больших локальных шунтов.

Добавлено через 6 минут
Hello, syn08
Guys wondering: are you sure that it is an aluminum case ?

[syn08]

Честно скажу, мне очень не нравится местоположение большого жёлтенького конденсатора, это входной кондёр, и он очень рядом с ВК который без больших локальных шунтов.

Добавлено через 6 минут
Hello, syn08
Guys wondering: are you sure that it is aluminum case ?

Well, I am positively sure This is where I purchased the case from:

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

Case: RM-14213
Internal chassis (defining the front compartment with the toroids): C-14243
Bottom lid: TBC-14252
Top lid (perforated): TBC-14262

Thanks,

[Nick]

The translation:
Корпус алюминиевый, можно пойти по ссылке и посмотреть.

Well, I am positively sure This is where I purchased the case from:

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

Case: RM-14213
Internal chassis (defining the front compartment with the toroids): C-14243
Bottom lid: TBC-14252
Top lid (perforated): TBC-14262

Thanks,