Ultimate sounded Class D for diy without custom made IC's like Tripath TA2020

[tiefbassuebertr]

I note, that the most diy projects use only integrated solutions like follow pics:
http://www.6moons.com/audioreviews/kingrex/hero_tripath.jpg
http://www.russianelectronics.ru/files/14993/TPA3122D2-1.jpg
http://i62.servimg.com/u/f62/13/72/79/14/tpa10.jpg
http://www.irf.com/technical-info/images/tc-irs2092.jpg
High Performance Class-D Speaker / Headphone Amplifier Series | Audio ICs | ICs | ROHM CO., LTD.

but unfortunately no discrete ( I would better say "quasi discrete") solutions with Op-Amps and comparators I can find here with explicit analysing;

What is therefore the reason?
Is there no "Nelson Pass" or "John Curl" of Class-D designs (easy design, very high switching rate arround 10 MHz and good to understand circuit)?

Follow topology is more for beginners but not for top class requirements:
http://www.diyaudio.com/forums/class-d/159157-selfoscillating-class-d-amplifier.html

The topologies about this threads are good suited base for start own developement projects because the sound quality reach a certainly quality level:
http://www.diyaudio.com/forums/class-d/164460-newclassd-better-class-d-amp.html
(PDF pics about post #23)
http://www.diyaudio.com/forums/class-d/83244-lc-audio-zappulse-800xe.html
http://www.diyaudio.com/forums/class-d/158129-tc-electronics-switch-amp.html
http://www.audiopower.it/index.php?option=com_content&view=article&id=56&Itemid=53
and
L C Audio Technology / ZapPulse 800XE
but no detailled schematic diagram and circuit description is to be discover.

The only discrete schematic, that is to find here, is my own created from ICEPOWER, unfortunately not finished:
http://www.diyaudio.com/forums/clas...ic-about-rev-engin-not-complete-finished.html

 

[theAnonymous1]

Does the use of a mosfet driver fall under "quasi discrete" or "integrated solution" by your definition?

If it's close enough to "quasi discrete" for you, then this thread is a good read........

http://www.diyaudio.com/forums/class-d/116590-ucd-like-topology-amp.html

 

[Ouroboros]

If you get a good understanding of the UCD circuit topology and how the amplifier works, you'll find it is easy to design one using a high-speed comparator for the front end, followed by a suitable MOSFET gate driver chip to drive the output FETs. There's enough information on this site, and on other websites for you to get a thorough understanding of the circuit operaration.
Don't even think about trying to switch at 10MHz. With a post-filter NFB architecture like UCD, anything over 400kHz doesn't offer any advantages and has the disadvantage in that the quiescent current starts increasing.

 

[tiefbassuebertr]

Is Mr. Bruno Putzey the "Nelson Pass" of Class D concepts?

"universal Class-D" was a good advice - I will check Mr. Bruno Putzeys papers and UcD as keyword in the next time
some examples here:

http://www.nxp.com/documents/user_manual/UM10155.pdf
http://seniordesign.engr.uidaho.edu/2008_2009/audiophile/UcD_WhitePaper.pdf
....::: HIFIAkademie.de :::....
PSW Recording Forums: Bruno Putzeys

How many other such principles are there for discrete Class D concepts (I am complete new in Class D and don't know all the colloquial terms for the good known various class D topologies - sorry).

 

[Ouroboros]

People had been looking at ways to provide a stable post-filter feedback class-d amplifier for some time. When Bruno's paper appeared a few years ago, many people (including me) thought: "Oh yes, that's the way to do it, why on earth didn't I think of that".
Like so many good ideas, the solution was very simple, and very elegant in engineering terms. It just hadn't been obvious before Bruno thought of it.

Simple class-D amps split between open-loop designs and closed loop designs.
The closed-loop designs split between pre-filter NFB and post-filter NFB and most are self-oscillating. (This is for DIY designs. Many commercial IC designs use pre-filter NFB with a clocked architecture).
UCD-style designs dominate the post-filter types, but there are a number of pre-filter self-oscillating NFB topologies. Have a look at the IRAUDAMP designs from International Rectifier to see one way of doing this, but there are both hysteresis and phase shift oscillator techniques in use.
It's a fascinating subject! I suggest you get a SPICE package and learn how to use it if you really want to start investigating it in detail.

 

[tiefbassuebertr]

Most popular Class-D Modules without special IC's

1) Audiopower "DPA-600"
Audio Power, Digital Audio Equipment
Digital Amplifiers

2) L C Audio Technology "ZapPulse 800XE"
L C Audio Technology / ZapPulse 800XE
http://www.diyaudio.com/forums/class-d/83244-lc-audio-zappulse-800xe.html
http://www.diyaudio.com/forums/class-d/158129-tc-electronics-switch-amp.html

3) DEXA Technologies ApS (newClassD) Lars Clausen
NewClassD Front Page
http://www.diyaudio.com/forums/class-d/80526-new-module-lars-clausen.html

4) Sanyo ICEPOWER
ICEpower — Our History
ICEPower Amplifiers with Bang & Olufsen
http://www.icepower.bang-olufsen.com/files/solutions/icepower250a.pdf
http://www.semiconductor-sanyo.com/pamph_pdf_e/VOL184.pdf
Sanyo Class D High Frequency
http://www.diyaudio.com/forums/clas...ic-about-rev-engin-not-complete-finished.html

5) Bruno Putzey's Universal Class D (UcD)
http://www.hypex.nl/docs/UcD180HG_datasheet.pdf
http://www.hypex.nl/docs/UcD180400.pdf
www.nxp.com/documents/user_manual/UM10155.pdf (UcD version 1.00 demonstrator board, schematic)
http://www.diyaudio.com/forums/class-d/145493-um10155-nxp.html
·¢ÏÖÃÀͼ£º·ÖÁ¢DÀ๦·Å£¬ - ά¿âµç×ÓÊг¡Íø (many schemas)
....::: HIFIAkademie.de :::....
PSW Recording Forums: Bruno Putzeys (Designer) - Dave Hecht (Master Tech) => Op amp crossover distortion (Op-Amps for UCD)
Believe the Hypex, class-D DIY amplifier - UcD, universal class-D
http://www.diyaudio.com/forums/class-d/116590-ucd-like-topology-amp.html (schematic by post #172)
YouTube - IRS20954

Comparable test are here:
2008 - Class D Amplifier comparasion test

Schematics still wanted from all except the universal Class-D about number 5

 

[balaboo]

I have also been searching for a LOONNNG time for a discrete class-D amplifier that can use standard through-hole components and at least "mid-fi". Apparently nobody in this forum wants to share a complete design, just droppings ...

 

[tiefbassuebertr]

I have also been searching for a LOONNNG time for a discrete class-D amplifier that can use standard through-hole components and at least "mid-fi". Apparently nobody in this forum wants to share a complete design, just droppings ...

Regarded discrete designs this basic articles are of interest:
Simple Self-Oscillating Class D Amplifier with Full Output Filter Control - Bruno Putzeys
http://hypex.nl/docs/ucd_aes.pdf
Evaluation of such an Class D amp "Self-Oscillating"
http://www.nxp.com/documents/user_manual/UM10155.pdf
IRF's discrete Solutions with Demo Board:
http://www.irf.com/product-info/audio/classdtutorial.pdf

The problem with this forum is the inaccurate choice of the headline in any cases. e. g.
http://www.diyaudio.com/forums/class-d/34718-success-class-d.html
the actually headline must be follow: "discrete self oscillating designs vs. other" instead "Success with Class D? "

So one must check most of all subjects by the file "Class D" (more than 3500 in the meantime). Perhaps one of the member can do this complete. I am shure, that there is much more to find concerning discrete class d design's.

I have found follow threads in this case:

discrete solution with IR2113/LM311/IRFP460
http://www.diyaudio.com/forums/class-d/166214-ucd-25-watts-1200-watts-using-2-mosfets.html
New self oscillating post filter feedback topologies
http://www.diyaudio.com/forums/clas...scillating-post-filter-feedback-topology.html
Easy discrete topology
http://www.diyaudio.com/forums/class-d/119341-interesting-schematic.html
What is Class D ?
Project Web -- Archive
schematics, link overview
http://www.diyaudio.com/forums/class-d/28696-help-class-d-class-d-class-d.html
self oscillating designs vs. other
http://www.diyaudio.com/forums/class-d/34718-success-class-d.html

 

[balaboo]

Thank you, I am familiar with those designs - the NXP design would be my firts-choice, but I was told that the schematic has errors; the IRF still requires you to use a dedicated controller chip.

 

[marce]

Can I ask why a discrete version would be better! It goes against a lot of design technigues required for Class d amps (effectively a modulated SMPS). A discrete design would be more spread out introducing parasitic inductances amongs other gremlins IMO.

 

[Pano]

Can I ask why a discrete version would be better! It goes against a lot of design technigues required for Class d amps.......

A valid question.

 

[balaboo]

Er, I don't know about you, but DIY with SMT is NOT an option for a 50-something with bad eyesight ! Heck, just buy a commercial unit if that floats your boat.

 

[marce]

Even the people that do SMT re-work where I work use visual aids to do their job.
Its not whether it floats my boat, it is good design practice. Keeping things close together and thus reducing stray inductance is a must with these sort of designs. Due to the large, fast switching currents, it is paramount to control and keep these loops as small as possible to minimise EMC problems, and 80% of EMC problems effect the circuitry that creates them. That is why there are so many integrated solutions for Class D (and SMPS's).
I think that a discrete design would be a step backwards for class D amps, due to the switching element of the design, we are not talking of benign analogue signals but nasty noisy switched DC. And through hole components by their nature are inductive due to long leads.

 

[sonnya]

Even the people that do SMT re-work where I work use visual aids to do their job.
Its not whether it floats my boat, it is good design practice. Keeping things close together and thus reducing stray inductance is a must with these sort of designs. Due to the large, fast switching currents, it is paramount to control and keep these loops as small as possible to minimise EMC problems, and 80% of EMC problems effect the circuitry that creates them. That is why there are so many integrated solutions for Class D (and SMPS's).
I think that a discrete design would be a step backwards for class D amps, due to the switching element of the design, we are not talking of benign analogue signals but nasty noisy switched DC. And through hole components by their nature are inductive due to long leads.

I would give marce a thumb for his comments. He is right about is it not for fun that TI places X7R in several microfarads on the backside as close as possible to the chip. It is mainly to reduce the current loop size and ringing. And in this case we only talks about switching frequency of 384KHz!! A design with higher switch frequence is even more complicated.
Many MOSFET high speed bridge driver only support pulses greater than 50 nsec.

if you need higher performance with a predictable minimum pulse width you have to go with deltasigma as ANALOG DEVICES does, and they have made some very impressive chips. Not on power level, more in performance...
They have switch frequence of 6MHz, and this will give you a signal after the filter with very little amount of switchnoise and very low distortion levels in the audio band.

Deltasigma is the way to go if you want a fixed carrier frequency instead of a freerunning. A free running carrier will be able to produce unwanted audible tones if inductors are to close.....

 

[tiefbassuebertr]

Even the people that do SMT re-work where I work use visual aids to do their job.
Its not whether it floats my boat, it is good design practice. Keeping things close together and thus reducing stray inductance is a must with these sort of designs. Due to the large, fast switching currents, it is paramount to control and keep these loops as small as possible to minimise EMC problems, and 80% of EMC problems effect the circuitry that creates them. That is why there are so many integrated solutions for Class D (and SMPS's).
I think that a discrete design would be a step backwards for class D amps, due to the switching element of the design, we are not talking of benign analogue signals but nasty noisy switched DC. And through hole components by their nature are inductive due to long leads.

I am not familiar with Class-D. And at first glance I would agree. But what about the discrete topologies mentioned by post #6 ? Are this discrete class-D devices really a step backward ?
and designs like
http://www.analog.com/static/imported-files/data_sheets/SSM2380.pdf (for driving) or
http://www.analog.com/en/audiovideo-products/audio-signal-processors/ad1994/products/product.html
are really to prefer for new projects?
Very hard to believe for me.

Here an article from Analog Devices:
http://www.analog.com/library/analogDialogue/archives/40-06/class_d.html

 

[Pano]

Er, I don't know about you, but DIY with SMT is NOT an option for a 50-something with bad eyesight !

I am the same age and was very afraid to get into SMD. But i was forced to. It's not that hard with the right glasses and a hot air soldering station. In fact, it can be quite fun! Never thought I'd say that. And not being afraid of it comes in mighty handy, let me tell you. Would I rather do thu hole? You betcha! But SMD is a fact of life.

 

[marce]

To make matters worse, as they shrink down devices (the new fashion for IC's is QFN packages), the requirement for getting heat out of circuits will mean more designs requiring 'solder mask defined pads' with full connections to copper area's. This will make home assembly even more onerous, as realy this type of topology requires some form of reflow oven.
Once I finish this busy period at work, I want to look at all the aspects of SMD design that are going to effect the DIY comunity as there are numerous one, but I belive they can all be resolved. There is not only assembly to consider, but also PCB design and the minimum number of layers etc. One of my biggest niggles is the fact that the DIY community realy neads to use FREE design software, which in most cases is only suitable for simple PTH designs.
In my day to day job weve been designing with SMD since the late 80's and is so common place, that my current component library only has approx 20% through hole components, and most of those are connectors.

 

[balaboo]

I do repair work on guitar/PA systems, and let me tell you that the new breed of powered (active) speakers with built-in ClassD amplifiers are completely UNREPAIRABLE - cf. Behringer B212D, various Peavey and GVX units :-(.
I also repair PCs, including laptops - sorry, no fixee motherboard issues, especially on the Apple products.
Call me old-fashioned, but a "throwaway" approach galls me; I still do most of my own car repairs except those that require 'special' tools.
The hot air tool works great for those SMT components that expose their solder connections, but fails completely with BGA, etc. devices.
And I have designed and built RF amplifiers using standard mil-spec 2-layer PCB with through-hole components - please don't tell me kr*p like a few Mhz requires SMT,etc.
The main reason is compactness, and yes, radiation loops - I get the impression that most EE's seem to have lost the art and skill of high-frequency design AND implementation.
C'est la guerre !!

 

[tiefbassuebertr]

I do repair work on guitar/PA systems, and let me tell you that the new breed of powered (active) speakers with built-in ClassD amplifiers are completely UNREPAIRABLE - cf. Behringer B212D, various Peavey and GVX units :-(.
I also repair PCs, including laptops - sorry, no fixee motherboard issues, especially on the Apple products.
Call me old-fashioned, but a "throwaway" approach galls me; I still do most of my own car repairs except those that require 'special' tools.
The hot air tool works great for those SMT components that expose their solder connections, but fails completely with BGA, etc. devices.
And I have designed and built RF amplifiers using standard mil-spec 2-layer PCB with through-hole components - please don't tell me kr*p like a few Mhz requires SMT,etc.
The main reason is compactness, and yes, radiation loops - I get the impression that most EE's seem to have lost the art and skill of high-frequency design AND implementation.
C'est la guerre !!

Exactly this I think, too. For this reason I hate IC's, at least custom made IC's. For older CD-Players and remote control integrated amplifiers I am not able, to read out the software for clone an operating micro controller unit. The same problem one will get, if some years old special made class-D driver ICs are to replace. If there are discrete designs (or quasi discrete with industry standart ICs like IRF2110) such problems don't occur.

 

[AP2]

Hi,
interesting conversation, I understand the needs of an enthusiast who wants to build something and now .... everything is smd!
The main cause of SMD is the ability to produce large quantities at low price.
obviously producers SMD components, raise prices, ..the usual game.
Now the discrete components will give you ,almost ... almost makes me want to make a good class d, completely discreet.😀 (no ic,no driver)🙂

Regards