multi-channel amplifier power supplies

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

This is my first post and I just want to say that this is a really great, informative forum. I've taken up this hobby just recently and have been devouring as much literature on power amplifier design as I could which includes many an hour going through messages on this site. Naturally I have a host of silly questions for you all. :)

My first is around power supply design when dealing with multi-channel (by that I mean 3+) amplifiers units. I've read a number of conflicting opinions on whether or not seperate transformers, or bridge rectifiers or even filter caps make a difference in the quality and stability of the amp as well as the other measurable characteristics like THD, slew, and especially noise.

I understand what the benefits are (namely lower cost) but what are the risks involved in trying to build a 7 channel amp (150W RMS @ 8Ohms/channel, for home theater) with, say a 1500 VA transformer, a 35A bridge rec (comments?) and just parallel the output rails to the 7 amp boards. Is it just a matter of doing the math correctly and ensuring there is enough filter capacitance to supply adequate current or are there other aspects to take into consideration? Any comments on circuit/speaker protection and/or the design of the power supply itself are appreciated.
 
Disabled Account
Joined 2003
MikeA said:
I understand what the benefits are (namely lower cost) but what are the risks involved in trying to build a 7 channel amp (150W RMS @ 8Ohms/channel, for home theater) with, say a 1500 VA transformer, a 35A bridge rec (comments?) and just parallel the output rails to the 7 amp boards.

I am actually not a fan of monoblock construction for multi-channel amps. I understand that you can improve inter-channel modulation (cross-talk via power supplies), but I think the built-in "redundancy" in a big power supply, vs. several small power supplies, can be quite benefitial in a multi-channel set-up.

For example, see you use a 200w transformer per channel. in a 5 channel set-up, the "juice" in other channels power supplies cannot help out a channel that is being overloaded. If you use a 1kw transformer, you wouldn't have that problem at all.

so there are pros and cons and for me I went with a big transformer (and beefier diodes / caps) for power sharing, simplicity, and ease of construction.
 
There are some interesting tricks you can do to make multichannel amps work well on a single supply*, but I wouldn't go there to start.

The big issue is going to be crosstalk- the supply has a finite source impedance (which itself can change with large changes in load current), and the voltage will be modulated by the current draw of each stage. If the amp circuit has great supply rejection, you'll possibly be able to get away with this. But if your goal is making something that measures as close to perfect as possible, to insure freedom from motorboating, and maximizing the output from a channel when the other ones are working hard, single channel supplies are more certain to get you there.

* One such is to pair the channels and have each member of the pair run with opposite polarity to the other. Tough with an odd number, though, and far from perfect.
 
safety issues?

Given the fact that the current will be drawn against a single transformer to power all 7 boards is there a greater risk of damaging the transformer as the result of having to increase the mains fuse in the PSU if the secondaries or bridge rec short?

I figure a 150W/8Ohm load would require about 40V rails or so (with normal losses) and draw about 3A of peak current. That means I would need a mains fuse of about 21A for 7 channels is that right? I understand the chances of fully taxing all 7 channels at 1 time would be remote but I want to take into account high amplitude transients which you see in movies all the time (explosions for example) which I would like the amp to produce faithfully without sag or droop. Would this power supply take too much abuse under these circumstances considering that a 1500 VA transformer is rated for about 18.5 amps with 40-0-40 secondaries? Do I need to go 2k or is that overkill?

Thanks for your input. You are really helping a newbie out here.
 
Hi Mike,

Not sure I follow your calculations. 150 W RMS means ~212 W peak. V= Sqrt(P x Impedance) = 41. You seem to have used 150 for P, giving a swing of 35 volts, and added 5 for losses in the output devices (reasonable). You should shoot for 46 volt rails to get 150 W RMS.

Rail voltage is not transformer voltage. Theoretically you'll get 1.414 x Secondary voltage. Diode losses and ripple eat up some of that, so count on 1.2-1.3x secondary voltage. A 35-0-35 transformer should get you close for class AB operation. Losses will be higher if you are building a class A amp.

Current draw =SQRT(P/R)= 5.1 A peak/ 3.6 A RMS per channel. 25.2 amps for 7 channels. => 1150 Watts RMS all channels driven. Now its time for balancing costs and performance objectives. With class AB amps, you can probably get away with a 1.5KVA or 2 KVA transformer. The DIYer rule of thumb is it should be rated 2-3x expected load.

There is no such thing as overkill in DIY. ;)
 
Humm being in Canada, you'd have to check the rating of your electricity lines in the house. Most of them are on a 15A breaker. Perhaps that a 21A fuse would not offer much protection. Anyway, for those big transients such as the one in movies, your amps will mostly rely on your filter capacitors, not your transformer. So instead of going for a bigger transformer, I'd go for more capacitance in the power supply lines.

I have built a 6 x 200W amp (Class AB) and it uses a 1.5kW toroidal transformer (the main fuse is a 12A) with about 72 000uF per rail and even at the loudest levels it never even glitches down. Given that my speakers are pretty efficient, I rarely need all that power but, with dummy loads, I've been able to put out more than 1kW using a sine wave at 10Hz, and the 12A fuse held ok.

I hope my 2¢ will help you.
Sébastien
 
BobEllis said:
...There is no such thing as overkill in DIY. ;)
Now that's true, but a professional amp designengineer I know, told me that because the energy content of music material is far lower than if you were playing sinewaves, you can get away with a far smaller PSU. In a typical consumer amplifier, the supply will be able to output around 1/8th of the current draw from the amp (output power/efficiency in %) continuously, while a professional amplifier for stage or PA-use typically has a PSU continuous output of 1/3 of the current draw. This would mean that a 7 ch. AB amp with a 60% efficiency could get by with a supply rated as little as 220 W and even for professional use the supply would still only be capable of supplying ~580W. Of course there is some performance-gains to be had by upgrading the supply, but even then the numbers are interesting. :)
If you wanted to play sinewaves at full output the supply for my example amp would have to supply 1750W with all channels driven (!!)


/U.
 
Re: safety issues?

MikeA said:
Given the fact that the current will be drawn against a single transformer to power all 7 boards is there a greater risk of damaging the transformer as the result of having to increase the mains fuse in the PSU if the secondaries or bridge rec short?
The real risk is toasting one of the amp channels because a single fuse on the input side has to be so large. I would fuse the mains appropriately for rated transformer primary current, and also fuse the DC side to each amplifier rail.
 
I consider that putting a fuse on each rail of every amplifier channel is definitely necessary in a multi channel amp. It happened to my quite often to blow a part in one channel and seing the fuses blow on the board, leaving every other channel intact. But what if only the output of the transformer had been fused? Well probably that the spike in current generated by the blown transistor would not have been enough to blow that fuse and that the spike could have killed parts in other channels.

Fuses are cheap, mosfets aren't. I would suggest not to skimp on fusing each channel.

Sébastien
 
Mike,

Greetings fellow Canadian, what amp are you building? I found this thread due to looking for exactly the same info, thanks. It has been very helpfull to me also, I am building a 7 channel ESP P101 amp for home theatre and driving it with my computer set up as a HTPC. But plan on using it for 2 channel audio also.

I have been looking for the same info as you and I am not trying to hijack your thread but just want to throw this out there to see what you and everyone else thinks, maybe it might help you too.

I am still in the research and development stage of my power supply but this is what I have so far, I plan on going with two transformers 1000VA 45-0-45 each. One rectifier per transformer and four 6800uf 75V caps per channel. I am connecting the left channels to one transformer and the rights to the other, to minimize crosstalk. This may all change as I find more info, as I am also completely new to this DIY, but aint it fun;)

What do you think??

Clint
 
Multichannel power

"Current draw =SQRT(P/R)= 5.1 A peak/ 3.6 A RMS per channel. 25.2 amps for 7 channels. => 1150 Watts RMS all channels driven. Now its time for balancing costs and performance objectives. With class AB amps, you can probably get away with a 1.5KVA or 2 KVA transformer. The DIYer rule of thumb is it should be rated 2-3x expected load."

Hello all, this is my first post. I am a bit of a home theatre loony, in good company here I see (with the best intentions;) )

I have spent some time pondering this issue as well, power for 6 or 7 channels at one time. As my system is powered by a newish (but gutless) Pioneer VSXD-811, I have thought about the same power supply issues with the same reasoning. All this is fine, but in reality, when during a movie (for example) do all channels operate at peak level?? Not often to almost never would be my suggestion, after watching most movies with bombs and explosions (a prerequisite for movies:).

May I suggest that using a stand-alone pre-amp (aka Digital Decoder - which is why I bought the Pioneer - to use as a pre-amp) as the source via DVD, then using 3 x stereo power amps as the actual supply? These would be set up as 2 channels front mains, 2 channels surround, 2 channels front/rear centre and maybe a 4th for bridging into a passive sub (although a discrete/bridged centre front channel would be my preference). The other issue is that of having power to match the real requirements, mine for example for power would be as follows: 2 x front mains 200 Watts per channel, 2 x surround 100 Watts per channel, centre front 150 or 200 Watts per channel and centre rear at 100 Watts.

It would not be hugely difficult to have similar tone and specs for the amps in different power ratings. Why have even 100 Watts if you don't have the speakers to use it?

I suppose this is really the same as an integrated 6/7 channel amp as you propose, but would be easier to manage in construction (I will be building my amps), separate power supplies and no issues (a lot less anyway) with interference / channel separation.

It comes down to quantities of materials in the end, with a multi channel amp being simply more space effective and cheaper to build than 3/4 two channel amps (but wouldn't look as awesome in your rack)

As my Pioneer is still under warranty, I won't be touching it, but as an interim project, I will replacing a few bits here and there to get some more grunt from this unit.

See yas later...
 
"Anyway, for those big transients such as the one in movies, your amps will mostly rely on your filter capacitors, not your transformer. "

Given that a 1F cap will provide 1A for 1 second, how much is he going to need for seven channels at 100W/4R? 50F?

Real lame and anemic sounding stuff is designed for 25% duty cycle.

Does this mean the transformer is only 25% as big?

No.

The reduction in transformer size is the square-root of the power ratio, in this case 4:1, or 25% duty cycle is 1/2 the size.

Tests by Crown on live music show duty cycles in the bass portion of multi-amp system to exceed 50% (58% actual).

Real power supplies have a power factor of about ~0.7

Real class AB amplifiers with low bias have efficencies of about ~70%

Real woofers look like a 45* load, not a resistor. This draws more power. I consider this a wash with the 58% duty cycle.

Bottom line:

200VA transformer for every 100W of rated output seems to be best in terms of cost vs sound. You can go bigger, but the returns dimminish above this point. Below this point the sound suffers.

100VA for every 100W of rated output into a resistor (not a speaker), is 25% duty cycle size. This seems to be the de-facto standard for most big name superstore electronics. Driving music into clipping on speakers for a long time will cause the transformer to fail.
 
djk said:
"Anyway, for those big transients such as the one in movies, your amps will mostly rely on your filter capacitors, not your transformer. "

Yes, but is the inference with sound reproduction on other types of material mean 6 or 7 channel music reproduction? I would have thought live music maybe, and SACD maybe but what else can use so many channels discretely? This is an honest question. Would an owner built multi channel amp for live music for example really be reinventing the wheel? Although I suppose powering the amps is the same and has the same problems.

Originally posted by djk 100VA for every 100W of rated output into a resistor (not a speaker), is 25% duty cycle size. This seems to be the de-facto standard for most big name superstore electronics. Driving music into clipping on speakers for a long time will cause the transformer to fail.

Bigname superstore electronics need suckers like me to buy their stuff. Reading threads like this makes suckers like me into "knowledge is power" people right? :) therefore being able to ask questions of salespeople that they cannot answer (which is a list getting longer by the minute now I am reading you guys info - thanks)

Ok, so why not adopt a margin for power supply "overkill"? Is this what you are suggesting? I (for one) am very interested in how this would be adopted and approached via cap size, transformer size/type etc, as I now would like to get my soldering iron warmed up! (or is that credit card warmed up - I always get them confused)

Is 100% duty cycle the aiming point, rather than 70%? the extra cost (for the transformer anyway) is marginal, I suppose the caps are going to burn a hole though. Remember, I am a newby at this, some maths and explanation would be good.
 
quote:
Originally posted by djk
"Anyway, for those big transients such as the one in movies, your amps will mostly rely on your filter capacitors, not your transformer. "

You misconstrued what I said, go re-read it. The quote was someone else saying something absurd. A filter cap has inadequate storage to make up for a small transformer.

The point of diminishing returns on the filter caps for a transformer in the recommended size range is 3,300µF for a single 8 ohm load for a 60hz line, 4,000µF for 50hz, ie: a stereo 4 ohm amplifier for 50hz needs a pair of 16,000µF caps.

Caps larger than about 30,000µF when used with a typical large transformer can cause a 35A bridge to fail. A multi-stage soft start may be needed.
 
Power Supply Issues

Hello everyone, thanks for contributing to this post.

From what I understand from reading Randy Slone's book, it is quite normal in the industry to underrate the power transformers in amplifiers, particularly multi channel, home theatre amplifiers where there is a large degree of dynamics in the material. However, one real benefit to using beefier transformers is the advantage that larger transformers tend to have lower power supply droop in the secondaries when fully loaded. That is, transformer ratings are usually rated unloaded. When fully loaded there will usually be droop in the secondary voltage of anywhere between 4-15%. I just finished looking at Plitron's website and the regulation ratios that they publish seem to indicate that lower power supply droop factors are tied to higher rated transformers.

I'm not aware of any other effects that underrating a transformer (say to 70% of fully rated output) may cause except for the possible overheating of the transformer under very stressful voltage requirements.

Is the general recommendation then to use a seperate rectifier and filter caps for each of the channels or just parallel the caps to the various amplifiers? From what I've read it sounds like there are some serious safety issues for boards from drawing from the same filter caps.
 
"That is, transformer ratings are usually rated unloaded. "

Transformers are rated at their full load, you add the regulation factor to get the no-load voltage.

TOROIDAL 30VA 60VA 100VA 160VA 250VA 330VA
Height (mm) 30 33 40 42 55 50
Diameter (mm) 70 87 88 108 120 130
Bolt M5 M6 M6 M6 M6 M8
Weight (g) 450 800 1150 1700 2600 2800
Regulation % 15 13 10 10 6 6

This brand of 330VA has 6% regulation, so a 40-0-40 would be 42.4-0-42.4 no load. With a rectifier and filter cap the no-load voltage will be about ±59.3V
 
  • Like
Reactions: 1 user
Re: Power Supply Issues

MikeA said:
Is the general recommendation then to use a seperate rectifier and filter caps for each of the channels or just parallel the caps to the various amplifiers? [/B]

I found this discussion very interesting and still have a question that nobody answered:

if there's any advantage of using separate diode bridges and caps for every channel?
At first it seems to be an advantage, because you separate the interaction of the amps during the discharge phase of the caps, but I do not knolw about any bad side effects it could appear.

cheers

Ric
 
Re: Re: Power Supply Issues

Ricren said:


I found this discussion very interesting and still have a question that nobody answered:

if there's any advantage of using separate diode bridges and caps for every channel?
At first it seems to be an advantage, because you separate the interaction of the amps during the discharge phase of the caps, but I do not knolw about any bad side effects it could appear.
Ric

The only advantage is a possible reduction in crosstalk between channels, if your amps are poorly designed.

The cost is much higher, as will be size and weight.

It's really just that simple.
 
Just out of personal experience, I used to drive 4 chip amps from a single 375 VA transformer with large caps (22,000 uF). It worked OK. Then I ran a Mini-Aleph clone prototype from the same supplies for testing, and of course this has a lot of bias current draw and lower PSRR. The modulation of the Aleph's output (used for mids in bi-amped system) by the chip amps' power demands (used for bass) was quite noticeable. No hum or motorboating, but a very strong coloration of the music. It actually sounded mellow - just not real :(

Bottom line, chips or other high PSRR topologies may not care so much, but for some applications it's better to have one transformer per channel.

Corollary, I looked at the prices for the smaller standard toroids at RS, and two 120 VA transformers barely cost 10% more than a single 225 VA unit. Two, say, 4700 uF caps cost less than one 10,000 uF cap. Bridges cost pennies. In other words, I found the only real world penalties in redundant supplies are space, and soldering time.

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