2 Designs for Class D Power Supply - Which One?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,

After reviewing documents found in this site, ESP (sound.westhost.com) and others, I have come up with a couple of versions. I need input in terms of:

(1) if there are any real differences between the two (other than the doubling of capacitance in version 2)

(2) if yes to above question, which one is better suited for:

(a) a stereo amplifier sharing the power supply, and
(b) a mono amplifier with its own power supply

The images (PSU Version 1, PSU Version 2) are attached below in separate replies.
 
PSU Version1:

C1 - 100nF
C2, C3 - same as C1 (100nF)
C4 - C11 - very low value (22nF)
C12 - C15 - 2,200 4,700 or 6,800 or 10,000uF capacitors
R1 - R4 - bleeder resistors 10W 6.8K (some high value)
Transformer - rated for use
 

Attachments

  • psu1.gif
    psu1.gif
    6.6 KB · Views: 1,069
bck2diy said:
Bump to see if this thread can be replied to... No responses in two days with 161 views!


Hi,

Sorry we're not quick enough, really you ought to have used those two days to try and come up with the answers yourself though.

1. Yes, big differences. #2 doesn't double capacitance, it creates two more rails for a second amplifier, it is simply paralleled center tap based supplies.

2. #1, and there's been alot written about it, tnt audio, zero-distortion.com both, along with the mention of it countless times on the forums.

Regards,
Chris
 
power supply question

I am a little surprised no one has responded. Version one will work fine with a minimum of line noise but not the best channel separation. Version 2 will give the best channel separation but has a lot of ripple current in the grounds and this probably will be put right on top of the supply voltages so will create audible artifacts, not recommended and no easy fix.
Roger
 
Thanks sx881663! That is exactly the kind of response I was looking for. I was trying to understand how audio quality (channel separation, noise, etc) is affected by the two designs. Now I have some additional avenues to explore (decrease ripple current effects in version 2 perhaps?, use version 1 to build dual mono amp, etc) ...

Chirs, no need to apologize ;-) This is my first time posting on this forum and I had noticed a message icon with a black dot on it against my thread (this coupled with the fact that my replies are moderated which I was aware) made me unsure if the thread was open to replies - couldn't find the meaning of the icon in faq and my email to the forum admin didn't result in a response so the attempt to bump. But it did result in what I was looking for! The sites you mention, I have been to but I didn't get the kind of answer that made me aware of the differences that sx881663 provided. I was not aware of zero-distortion.com - thanks for that though! I will go to that site and see if I can get more knowledge.

Back to the topic, my goal is to identify and understand power supply designs (for a mono amp case as well as a stereo amp one) and I hope to use this site to get some questions answered!
 
General stuff

The common transformer version and many variations have been used in commercial designs for years. The best implementations use a heavy copper plate as the common ground point. Even with this there is still the line noise problem. This is due to very high ripple currents charging the caps at 120 Hz. The copper plate will minimize the voltages generated but actually will increase the magnetic fields due to the high currents this low resistance path encourages. These magnetic fields will transformer couple into anything nearby. This is why we recommend all wiring be shielded or at least twisted pair and separated, this minimizes the interaction.
Of course the reason this configuration is popular is cost. It is not the best but the cheapest, both from parts count and labor to assemble in the factory. My experience has been that if the amp output shows any line residual byproducts this will directly shorten the sound stage and mask low level information. Not a good thing. Two transformers and preferably two chassis will always be more expensive in labor and money but will be worth it in the long run. A benefit is you will probably have more room to work and do experiments in. This becomes particularly important when it limits the choices of how you can arrange and wiring and the parts, even restricting what parts you can use.
Understand this is just my opinion and you know what they say about opinions! Also understand your mileage may vary (YMMV), rephrase that to will vary! I still much prefer the dual monoblock approach and think it is worth the extra time and cost.
As a new poster you will find this forum has some very sharp and knowledgeable people in it. Most all will be eager to help if you do your home work before asking. We are very reluctant to help people that are too lazy to do a little reading beforehand. I am not saying you didn’t but your post didn’t mention references as to any research you had done so leaving the question open. Also it is better if you sign your posts so it isn’t totally anonymous (can be a nickname or something you want to be referred too), it is nice to feel we are responding to an individual. After you have a few 100 posts done you don’t need to do it as everyone will know you by then.
Roger
 
Roger,

I am planning to go through a series of diy projects (simple to complex let's say).

1. Put together pre-packaged psu, amp, chassis, etc to create a simple, inexpensive, good sounding system: two of my current choices are the LM3875 kit by Brian/Peter and the Ucd180 ST (psu, amp) kit from hypex. The goal of this step is to create an amplifer that is easy to put together and will provide an insight into the components of the amplifier. The result will be a self-built stereo amplifier.

2. Change the power supply unit in the above kit to a self implemented one and study any improvements (measured and audible). By self implemented, I mean one whose component values are calculated to the specific amplifer power requirements (e.g, the capacitance values, rating, and the bleeder resistor values, etc) i.e., tayloring one of the designs available. In the process I also hope to try simplified versions of the design that I can build on (e.g., the capcitors which are designed to reduce diode noise will be absent in the initial build and will be introduced later).

3. From the knowledge gained from steps 1 and 2, put together two mono amplifiers each having an improved self designed power supply (snubberized - a word I discovered during my searches but one which I haven't fully understood the concept of), soft start, dc protection, etc.

At this stage I do not plan to attempt implementing amplifer design from scratch. I will only be using Class D amp pre-packaged amplifier kits. Nor is this a quest to find/design/implement the best sounding system (read: better than audiophile quality amplifiers available for sale out in the marketplace). I hope to get started somewhere in diy audio and this seems to be an achievable goal to me. In terms of cost, I can say that I expect to use inexpensive parts and simple design in (2) to keep costs low and go to a more expensive (better parts, complex design, etc) in (3).

Websites referenced so far:

1. http://www.zero-distortion.com/techno/powersupply/powersi.htm (ps design)
2. http://www.diyaudio.com (searched for "power supply design" and also looked at articles related to the topic under power supply, chip amp, and class d forums.)
3. http://sound.westhost.com/power-supplies.htm (linear ps design)
3. http://www.tnt-audio.com/clinica/ssps1_e.html (ps design)
4. google search for "amplifier power supply design"

What I have learnt so far (or what I think I have learnt!) ...

1. How to choose the VA rating of tranformer (i.e., transformer sizing)
2. Calculating correct value for primary fuse
3. Selecting diode ratings for rectifier bridge
4. Need for small valued capacitors to reduce noise in line input and across diodes (diode switching noise)
5. Calculating energy storage requirements for amplifier
6. Identifying capacitors for energy storage (rating, charge/discharge rates, etc)
7. Computing bleeder resistor values and their benefits (increasing capacitor life, removing dangerous volatages quickly)
8. Isolating supply ground from audio ground (I found an article that explained how to do this using diodes, a capacitor and resistor - I am reviewing this, which is the reason for the initial set of diagrams missing the associated circuit)

What I don't know:

1. Everything else!

I am certain I haven't answered all of the questions still (comes with the territory being a newbie!) I hope this provides some of the answers.

Going back to the start of thread, when I presented the two designs, I intuitively held the belief that version 1 is good for mono amp configuration and version 2 is good for stereo amplifier design. I did not know about the ripple current effects on audio distortion (I am sure this has been mentioned in the urls above - I will have to go back and read those again as well as do research on how to implement a ps design that provides good channel separation when used in stereo amplifier). Since my diy stages require building a stereo amplifer (with self implemented psu) before getting to the mono-amp setup, I will have to spend time on this topic further. I will be back with additional questions (and any new implementation ideas I have for a stereo amplifier) soon.

Thanks!
-bck2diy

p.s.: call me bck2diy - it has a special meaning to me :)
 
My recommendation would be to use a single transformer, but separate rectifier and capacitor system for each amp. This gains you the better regulation and smaller size and lower cost of one bigger transformer, but minimised crosstalk to levels below worth bothering about.

As for rectifier and capacitor designs, I prefer to use the dual rectifier approach. This requires separate secondary windings. So you will need a total of 4 bridge rectifiers for a 2 channel system.

I prefer this way because it gives you better quality ground and loads the transformer better so you can get that little bit more out of it.

On those images you put up, the first one has too many snubbing caps used IMO. The one on the AC side makes the most difference.
 
I must commend you on your level of thorough research. An example to be followed.

Secondly I think you have a pretty good plan there. It should let you learn alot along the way.

I pretty well agree with richie00boy, though I think I'd definately want the beefy transformer which will have better regulation in itself and will keep all those caps pumped up.

Bleeders also have the advantage of helping to maintain equal voltage off of "series" caps.

It seems like you've done more than enough to help you get started, everything else just comes one step at a time.

Oh, I like the fuse before the switch :)
 
the big drawback of having a stereo amp. with 2 transformers is the weight it will have and the size of the cabinet in comparison with a 1 transformer setup. The advantage of having a bit more power out of it is marginal. Did a lot of testing with different ways of making PSU's for stereo amps. and it just is a choice of either having a heavy and big stereo amp. versus a smaller 1 with a tiny bit less power. And the difference in quality, well that's personal and for me not enough/marginal. For me soundstage is far more important than absolute power. And the size/power of the transformer is not so important as with an A/B amp. Class D is soooo efficient.
 
Thanks richie00boy, classd4sure and Bert for your input. I have attached an updated version which follows the design advocated by richie00boy. Also, I have included the circuit for isolating supply ground from audio ground (ref: adire audio) which I plan to use (for single ended input setup).

When I have to move to step 3 (mono amp setup for each channel), I would replicate the blue box contents and rewire T1 - T4 (see attached image) as in version 1 to each of the blue boxes (i.e., two rectifier bridges per channel)

Richie00boy, I assume you're talking about the snubbing capacitors C4 - C11 (both in version 1 and the attached image) and C17 - C24 (in the attached image). I don't plan on putting them on initially. I plan to add them to see if there is any measurable or audible (mostly the latter) difference and remove them if none is found.

Classd4sure or Bert, so a single transformer does provide good channel separation (soundstaging) if the bridge circuit is done correctly? Do I correctly understand what you're saying?

If so, I think I have my final design and I can proceed to the next steps. This helps me get started!!

Thanks all for your help!!
-bck2diy
 

Attachments

  • psu3.gif
    psu3.gif
    16.1 KB · Views: 363
That looks good in principle. I would move C2 and C3 to be as close to the bridge though as it's only there where they will do any good. You know my thoughts on the individual diode caps.

It is unecessary to include two sets of bleeders, one resistor across each rail will suffice. Although personally I do not like bleeders as they just waste power and capacity, and are another thing to fail. The amp will drain the caps down perfectly well, and during testing when the PSU might be powered with no load all you need to do is bleed them manually by a resistor or light bulb.

All that's missing is a cap across the neutral switch. I would use Y-rated caps for these positions to ensure they fail safe (open) should the worst occur. I use 2.2nF or 4.7nF.

I assume you have an effective soft-start system for the transformer?
 
the schematic you've drawn here does not allow you to connect the grounds together. Each supply should have their separate windings. So you'll need a transformer with 4 separate windings like this. Otherwise problems. If you want to use 2 bridges per amp. use 2 transformers so you'll have 4 windings.
I also do not prefer caps. across the rectifier diodes or bleeders.
 
Oh, I like bleeders too for series-connected caps. Like the app note says they are a wise choice. But in audio power supplies where series caps are not used in the vast majority of cases, they are simply a waste of energy and potential for hazard.

A split rail audio PSU is not akin to connecting caps in series, especially the twin bridge approach like what has been advocated in this thread.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.