3 phase power supply for audio !

Forgot the grid neutral in the previous post (connects to the midpoint of the SMPS' inputs). The zero sequence choke is optional. Thanks.

However, please note that the functionality of this circuit is at the expense of input current THD and power factor, in view of the OP's choice to disregard the grid-side power quality.

Untitled-1.jpg
 
Last edited:
My view was that...of mains frequency transformer/rectifier/filter equipment, and then....a smps block of equipment, had any practical benefit.

Ok, that's what I understood you to mean and I am in accord that there is little benefit, so it's not an option I proposed.
What I proposed was no transformer, just a 3 phase rectifier to provide the input to a SMPS.
More or less a typical off-line SMPS except the 3 phase rectifier means there is no need for "bulk" capacitors.
That eliminates one of the more expensive and sometimes troublesome components.

3ph input smps 'rectifiers' are commercial and of the power capability that you appear to be after, and are likely able to provide a quality circa 100Vdc output

The telecom units I have seen are all 48 V maximum.
You mentioned 100 V supplies earlier too, do you have a link to the sort of unit you have in mind?


Best wishes
David
 
Last edited:
Besides, what do think about a pair ... for all the 25 amplifiers...?

My plan is to have the main amplifiers as Class H modules with switch mode buck converters to act as rail trackers.
The smaller amplifiers are just conventional Class H with the switch mode complexity removed.
The smallest amplifiers can be just Class B+, the same core amplifier but with even the Class H complexity removed.
This results in an efficient amplifier system with nice modularity and no more complexity than is needed to meet the efficiency requirements.
(there won't be quite 25, maybe 16)


...the method...does not reach 560V like the 3-phase (6 diode) rectifier...

Yes, the 560 V (or ~600 V in Australia) is a bit of a worry.
Ideally I would like the low EMI and noise of a Zero Volt Switched supply.
But the resonant LLC circuit tends to increase the maximum potential stress on the transistors even more so it's a trade-off.
However, as I said to "trobbins", I don't think there is sufficient benefit to justify an input transformer ahead of a SMPS.
I notice some of the 3 phase SMPS style welders and plasma cutters do more or less what I propose.
They tend to run lower volts and more amps than I want but sufficiently similar to provide a few hints.
Thanks for the ideas, did you read the Cuk article I link-posted?

Best wishes
David
 
Last edited:
Dave Zan said:
My plan is to have the main amplifiers as Class H modules with switch mode buck converters to act as rail trackers.

I am not very familiar with class H, but I guess these are amplifiers that vary their supply voltages according to input signal. If I'm guessing correctly, you're right, they won't be needing any "special" regulation, as their supplies follow the envelope of the audio signal.

Dave Zan said:
there won't be quite 25, maybe 16..

Yes Sir, I get your idea. Actually, even I was planning something like a digital 16-ch decoder / processor and amplifiers, with 3-way LCR, 2 subs and 4 regular surrounds, i.e. 15 channels. I even worked out a lot of things, but haven't done any PCBs yet, but will post in the digital source/line-level forum when there's enough progress.

Dave Zan said:
the ~600 V ...is a bit of a worry...

Not if you choose to split the 600V into 2x300V and feed two 300V SMPSs (post #80, 81). With the zero-sequence choke in place, the resulting ripple is reduced further, even when small capacitors (<1000uF) are used.

Dave Zan said:
I don't think there is sufficient benefit to justify an input transformer ahead of a SMPS.

I did notice that point, which was why I went on to make posts #80 and #81. However, there could be control / performance benefits to having "non-isolated" converters after transformers.

Dave Zan said:
I notice some of the 3 phase SMPS style welders and plasma cutters do more or less what I propose. They tend to run lower volts and more amps than I want but sufficiently similar to provide a few hints.

That's great, but do make sure that they're are not the "constant current" type, which is very popular in resistance welding.

Dave Zan said:
did you read the Cuk article I link-posted?

I remember having read that article a few years ago when I was looking for a bridge-less AC-DC, low-voltage, step-up converter for a work-related requirement.

The advantage of the method is that it provides bridgelessness, PFC, isolation, step-down and regulation in a single stage, but you would have to design, implement, test, iterate and characterise the converter (plus HF transformer) on your own, as it appears to be a novel method that is not likely to be available as a commercial product.
 
Direct rectified 3ph Oz mains would need something like post #81 to get bulk DCV down to a manageable level, but still requires any subsequent smps to be rated for mains tolerance and for safety isolation - have you identified anything suitable?

RTP, Enatel, Eltek are still manufacturing I believe. Powerbox is a commercial outlet, and has Enatel. I've also used a small Rectronic rack system, but the major manufacturers likely only supply to telco industry players, so Powerbox may be a reasonable indication of what is available to the public unless you come across 2nd hand telco stuff (eg. from a telco service/installation provider).
 
Last edited:
I am not very familiar with class H...as their supplies follow the envelope of the audio

Basically, yes, use of the term is not always consistent.

...16-ch decoder / processor and amplifiers...i.e. 15 channels. I even worked out a lot...
Can you provide a preview?;)
Decoded with what?
Only AVRs seem to have all the Dolby/DTS licenses for surround sound but the DACs and audio circuitry are usually mediocre at best (Check AudioScienceReview for the measurements if you haven't already)
And license restrictions mean the manufactures don't make available the PCM data streams.
It is possible to hack the IIS data before the DACs and export as S/PDIF or AES.
But then one losses volume control, would need a multi channel DAC with master volume, doable but not cheap.

Not if you choose to split the 600V into 2x300V and feed two 300V SMPSs (post #80, 81)

Yes, thanks, some version of that looks quite feasible.

but do make sure that they're are not the "constant current" type, which is very popular in resistance...

Will do, but I think the basic circuit is still more or less the same, just the control loop varies.

...on your own, as it appears to be a novel method that is not likely to be available as a commercial product.

Yes, looks like a lot of work, I would much prefer to buy a finished product.

Best wishes
David
 
... but still requires any subsequent smps to be rated for mains tolerance and for safety isolation - have you identified any...

I must admit I have looked at Ebay and Alibaba :eek: and am a little conflicted.
A SMPS is basically pretty simple - if you use a control IC anyway, because the IC maker has done most of the hard work and can split the development costs over millions of units.
So it shouldn't be too expensive, even with quality components.
There is, of course, the need for proper safety certification and tests and I realise that "name" brand products will cost more.
But the price premium seems more than I can rationalise.
So I am tempted to try an Ebay/Alibaba SMPS, check it myself, use it conservatively, in a nice earthed metal box so if it fails the worst it can do is trip a breaker or earth leak safety switch.

...Powerbox is a commercial outlet, and has Enatel. I've also used a small Rectronic rack system, but the major manufacturers likely only supply to telco industry players, so Powerbox may be a reasonable indication of what is available to the public unless you come across 2nd hand telco stuff ...

Thanks for the information

Best wishes
David
 
Last edited:
Dave Zan said:
Can you provide a preview? Only AVRs seem to have all the Dolby/DTS licenses for surround sound..

There exist chips that already have the firmware pre-loaded, such as the one discussed in this thread (from post#52). Fortunately, I happened to procure a few pieces of the above-mentioned chip (a few years ago) and still have them with me. Besides, I also have plans to allow external I2S streams directly into the processing section, to allow external decoders if desired.

Currently, I do not plan to include the newer HD formats such as E-AC3/True-HD and DTS-HD HR/MA or MPEG-AAC, as DTS being backward compatible, would still play 6 channels at 1536 kbps. The only downside is that the back surrounds need to be generated from side ones by way of a matrix (5.1 to 7.1 expansion). I also have a Yamaha 7.1 AVR capable of all the HD formats, which I can hack into, just in case.

For the processing (EQ/crossover etc.) part, I plan to use two STA309As (8ch, 24-bit, 192kHz processing), along with two 8-ch PCM1680s for the D/A conversion. Volume control should not be problem, as it's available on the decoder, processor and the DACs. There is also the option of having digitally-controlled stepped attenuation just before the amplifiers.

Dave Zan said:
SMPS is basically pretty simple...the IC maker has done most of the hard work....safety certification and tests.....will cost more...So I am tempted to try an Ebay/Alibaba SMPS.

Yes, certifications (CE/IEC/EN) raise the prices, so an Ebay/Alibaba SMPS appears to be the way to go. They're reasonably safe for regular use, as 300-350V is just low-tension by power industry standards. Besides, a lot of computers across the world are operating from China SMPSes as we speak. Along with the ELCB/MCB, the metal enclosure also keeps the user safe from flying shards of transistor casing, in case a violent device failure should occur.
 
Last edited:
You are doing this for a 800W load, and to be on the safe side you can take the load to be 6 amps, 2 per phase.

6 * 230 = 1380 Watts

Who is going to make such a small transformer?


Better buy a single phase SMPS, be happy.
There are many reputed makers who meet the strict standards of your country, though they may be based in the Far East.


850 watt computer SMPS in Aussie is about $160....

Another thing, in car alternators a 3 phase rectifier is common, you can use those.
The theory is fine but the practical aspect is to be addressed also.
 
My city is a hub for transformer manufacturing.

Power distribution transformers, not the low voltage ones used in audio stuff, these have to work continuously in harsh conditions for years.
Many big and small makers, suppliers too of the parts, such as housings, laminations, insulation materials and so on.
The other center for transformers in India is Hyderabad.
I have friends who make and repair transformers, and also those who make IGBT based voltage controllers with very fast response.
The minimum size for a series produced three phase transformer. 11kV primary, is 25 kVA, and there are pole mounted single phase 10 kVA units. Secondary are 415 and 230 in three and single phase, more or less the standard you have.

design - Why don't three-phase transformers use toroidal cores? - Electrical Engineering Stack Exchange

Making a three phase toroid is difficult, and a specially designed three phase unit like the one you are thinking about will be expensive.
That is why I asked you the kW rating.
 
Last edited:
There are a few transformer manufacturers in Australia that would typically quote and make one-off 3ph 3 limb transformers, but the OP should know what mechanical and electrical ratings to request, including primary winding format. Most manufacturers would have a designer that can work backwards if provided with just an application design such as identifying the secondary side rectifier and filter and load characteristics.
 
Ask the price.
I would stick to a ready made SMPS, the design is professional, and replacements are available.
If you find they are good, and want more, then too it is easy.


I have a single phase 5 kVA voltage regulator at my factory, we had high voltages at night as the grid was not loaded.
The OP is asking for a 1.5 kVA class three phase supply, which is a very small unit, designing and making for such a small load is more an academic exercise than a practical thing to do.
Window AC compressors run at about 1700 Watts, with more than 30 Amp start current and about 10 Amps for a normal 1.5 ton unit.
Those are common, so are split units with inverters and so on.
Just compare their ratings, they are all single phase, till about 2.5 tons you will not even get a A/c or refrigeration compressor that uses three phase supply.
The OP wants to power a home entertainment system with 3 phase...
 
Last edited:
There are likely many industrial and commercial situations where an auxiliary 3ph LV transformer is required with a low power level (eg. < a few kVA) - very practical. In many situations it is because there is no local neutral, and bringing in a neutral is too costly to provide 1ph powered auxiliaries. One local manufacturer I know happily makes 3ph LV input down to 100VA.
 
Good...
That is what my point was, how practical this whole discussion was in terms of extra effort.

And compare the price for 3 single phase transformers with a single 3 phase unit.


3 units single phase 500 VA, and one unit 1500 VA, prices for both single and three phase for the bigger one.
 
NareshBrd said:
That is what my point was, how practical this whole discussion was in terms of extra effort.

There may not be a lot of extra effort in wiring up a 3-phase rectifier / commercial unit when compared to doing the same for single-phase, especially if a strongly motivated person feels it's beneficial in some way.

To be honest, I use 3-phase power wherever possible but unfortunately, I do not have a 3-phase connection at home. And if I do upgrade someday, I would be running not only 3-phase but also PFC.