diyAudio (
-   Class D (
-   -   Properly sizing transformers for Class D amps (

RHosch 7th January 2005 05:37 PM

Properly sizing transformers for Class D amps
I've been doing some searching and haven't found any thorough explanations of how to properly match a transformer to a Class D amp. I'll admit up front... I'm quite confused on this one.

I've studied Slone's writings on sizing a transformer, and his calculations all seem to be topology invariant... unless he is making the unstated assumption that the amp is always Class B. It seems natural to think that the efficiency of the amplifier itself will have an impact on the required size of the transformer, but no such efficiency coefficient is to be found in his calculation.

So I did some less "scientific" research, and took a look at hundreds of posts here, elsewhere, and looked at commercial Class D offerings to see what transformer sizes were being used for various amplifier output ratings. It looks like it's all over the board. I see some here using 1kVA or 1.5kVA trafos for 400W outputs. I see in some commercial products what looks like 500VA~800VA trafos for 1000W+ output. Using Slone's method, 1000W looks like it would need some 1.5kVA, but I'm not sure if that applies only to Class B or to all topologies.

Could someone point me in the right direction? I'm sure the information is already on the Web somewhere, probably in multiple places, but my Googling skills seem to be lacking lately.

classd4sure 9th January 2005 04:15 AM


Slone's methods are for a no compromise class B topology, hence all the added losses, if you used that method for class d you'd have extreme overkill.

You can still use it as a reference though but keep in mind the differences in topology.

Assuming class d amps are ~90% efficient (close enough), if you want 100W at the load, then 100W+10%, the only voltage drop of worth is that of the rectifier diodes, so you can take that into account when working backwards to get your secondary requirement. Double that VA rating for a 2 channel amp off one transformer.

The OEM penny pinch method is to use ~70% of the calculated requirement based on the fact that music signals are not pure sine waves and therefore dont' demand the same kind of power.

Then you have everybody else who swears bigger is best.

The determining factor might be that using a higher VA rating might get you a transformer that has a tighter load regulation factor.

Hope that helps ya.

RHosch 9th January 2005 05:54 PM

Yeah... that does help. Thanks.

djk 7th February 2005 08:19 AM

I just spent the evening at the B&O ICE Power site.

They claim 79% efficent into 4 ohms and 82% at 8 ohms.

Those are believable numbers, others I have seen are not.

First timers' always forget the power factor for cap input supplies. A 1KVA transformer will only drive about 700W into caps around 10,000F or so, bigger caps only make the power factor worse (no, you can't 'make up' for a weenie transformer with big caps).

Duty cycle on rap bass and some lead guitar may be as bad as 50%, but driving a 45* load will make up for this. A wash.

100W + 100W at 8R would give X2 at 4R. Divide by 0.8 for efficency and you need 500W in. Divide by a 0.7 power factor and you need a 700VA transformer for your 100W stereo amp that will be 4 ohm stable into a real loudspeaker (a 45* load).

Less is less.

You can figure your design around a 25% duty cycle, I have done this before. But the bigger transformer sounds better.

I also burned up a 25% duty cycle transformer when bench testing an amplifier into clipping using FM radio as the source material (of course no one you ever met has driven an amplifier into clipping, right?).

Don't try and save too much money on the transformer, in the long run it doesn't pay.

Yves Smolders 7th February 2005 08:45 AM


Your post is very contradictory to the discussion in the UcD power supply thread.

According to calculations there, 1000VA/2*40V with 33.000uF per rail should be enough to drive UcD400's up to 3 ohms (>550W/channel?)

Do you believe these calculations wrong? What would you recommend?

There's also the "hard limit" - UcD400 V2 pumps 20A max.

Would love to hear your comments.

UrSv 7th February 2005 09:00 AM

Want to have and need to have are two different things.

Don't forget that duty cycle as we are not talking about powers running full power 100% of the time which will "never" happen. Also a transformer rated at 500 VA is rated 500 VA continuous with a certain temperature rise. It will happily provide 750 VA during shorter intervals and since it is not continous it will do this without getting warm.

The general rule IMHO, is thus to take as much as you want to pay for. Lars at LC Audio had very good results using 80 VA (or possibly 50 VA) with his 200 W ZAPpulse. I myself would be very happy with 500 VA for a stereo UcD400 driving 8 Ohm speaker.

djk 7th February 2005 10:13 AM

You obvously don't understand things like 'duty cycle', and 'power factor'.

"Don't forget that duty cycle as we are not talking about powers running full power 100% of the time which will "never" happen."

As the service manager in a hi-fi shop I found that many customers drove their amplifiers into clipping on a regular basis, if not all the time. What is the duty cycle of a square wave?

"Lars at LC Audio had very good results using 80 VA (or possibly 50 VA) with his 200 W ZAPpulse. "

Sure, driving 2W out at a duty cycle of 4%, so what?

If we are going to brag about doing stupid things...I know someone that mounted a 10HP lawnmower engine in his pickup truck. It got in excess of 100MPG and would go maybe 60MPH on flat, level ground with no wind.

I have about a ton of Yamaha, Sony, and Denon transformers that have the thermal fuse out in the primary, why I'm saving them I'm no longer sure.

Feel free to ignore my comments based on 30 years of experience in the hi-fi business.

To finish the story about the 25% duty cycle transformer that burned up during testing, I also provided a 50% duty cycle transformer I recommended they use in production, and a 100% duty cycle transformer just to show what a difference in sound it made.

While they cheated and used the 100% rated model at the trade show, to their credit they rejected the 25% model and sold the amplifier with the 50% model.

You guys are welcome to do what you want.

(are you all bean counters, or are some of you DIY hobby types?)

Yves Smolders 7th February 2005 10:51 AM


We are only asking you your opinion, and when confronted by other people's opinion you call them bean counters... I believe that is unasked for. It's a discussion and I want to see opinions from different people, also yours. Just enlight us with your 30 years of knowledge and present us with your recommended setup.

So again, I'd love to have your opinion for PS and transformer for this setup:

I want to be able to run 3 ohms on 2 UcD400's, fully loaded. Feel free to enter a duty cycle that you feel is "difficult".


classd4sure 7th February 2005 10:51 AM


Take a deep breath. Just because you were disagreed with is no reason to take it as a personal assault which you seem to have done.

As to the question of who understands things like duty cycle...


Duty cycle on rap bass and some lead guitar may be as bad as 50%, but driving a 45* load will make up for this. A wash.
You do realize that a 50% duty cycle =0W output? What the hell is a 45* load anyway, and what the hell is a 25, 50 or 100% duty cycle transformer?


Sure, driving 2W out at a duty cycle of 4%, so what?
You do realize a 4% duty cycle = near clipping? While a 25% duty cycle = about half power?

This isn't an attack really, I'm just a bit confused by your lingo, maybe you can help me out.

I had previously decided on an 800VA transformer for a stereo UCD400 driving 4 ohm loads, 1000VA would of course be nicer, hardly a necessity though. I think even an 800VA would handle a pure sine wave test with reasonable heating, for maybe an hour or so, not 24.


Yves Smolders 7th February 2005 10:57 AM


That would already exceed any real-life test I believe... except maybe continuous max-power near-clipping bass for multiple hours, like in discotheques.

If a sine wave at full output can be handled at 3 ohms for an hour, there is *no way* I'd ever run into problems with my speakers.


All times are GMT. The time now is 10:58 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2