UCD180 questions

[Bruno Putzeys]

Originally posted by UrSv
I know it's controversial and there are many views and beliefs on the subject of transformers for Class D amps but IMHO I think it's clear that if you don't play them flat out then a proportionally smaller transformer will *suffice*.

In consumer electronics the general rule of thumb is to use a transformer with a VA rating half of the amplifier rating. Fully-loaded output voltage will be roughly 80% of unloaded voltage, so if a supply is designed to have an unloaded output voltage of 90%*Vmax (to allow for Vmax at 10% mains over-voltage), fully loaded with nominal mains lands somewhere at 70% of Vmax.

The worst test normally applied to an amplifier is first to let it run at 33% of rated output for an hour, then to have it deliver full rated power for another 5 minutes. After this test, a transformer rated as per the above will be hot (70-80 ºC), but well below its rated maximum temperature

Now you know this, you share the full extent of power amplifier expertise most CE manufacturer's "power amp gurus" have. They guard this knowledge closely.

[JohnW]

Hi Bruno,

From your explanation, I’m I to understand that the Rise-Edge ringing seen on the output of an HBridge is not seen by the MOSFET Die itself? Surly the “off FET” in the bridge must see the full voltage (including the ringing) as measured between HBridge PSU rail and the output? The “Die” after all is tied between two inductance paths – S & D.

I’ve seen 60V MOSFET’s blow with as little as 10V across them, due ringing spike – I always put this failure down to the ringing spikes exceeding the rated D/S voltage – not my design I would like to add but a great fly-back converter!

I could be wrong?

John

[Bruno Putzeys]

The die sees most of that voltage. However, if the energy stored in the parasitics is low, it can be safely absorbed in avalanche mode (think of the FET as acting like a zener). I've seen that very graphically in a 500W/2ohm FB device I made years ago using 55V fets. As power was increased, the overshoots increased too, but they were neatly stopped at 60V. Power could be increased way beyond that without problems.

I presume that the extreme case you were confronted with had a lot of inductance sitting there. Could you elaborate?

Cheers,

Bruno

[JohnW]

The designer – or rather the PCB layout engineer made the classic mistake of insuring the HF decoupling cap was near to the Positive connection of the HBridge – but with about 1 Inch of poor (thin) PCB track, and then though a couple of vias to add - return to the lower FET ground pin! This when combined with 5nS Gate drive edges and wide open deadtime – meant the poor guy could not increase the PSU rails beyond say 10V before the “60V” Fets blow!!!

Get that Dead-time and decoupling right!

Yes – I’ve also seen the “Ringing” – “clamped” (you know with a bad design as the modulation index increases) - but is this safe and advisable to operate continuously like this?

For me avalanche, means driving a high power fast Laser with say 100 – 200 pulses (if you where lucky) before the pulser died!

John

[anomalous]

> In consumer electronics the general rule of thumb
> is to use a transformer with a VA rating half of the
> amplifier rating.

which rating? the UcD180 is rated at 180W into 4ohms, but only 105W into 8ohms.

does this mean that 2 modules driving 8ohms, would only require a smaller transformer (say, 150VA) than 2 modules driving 4ohms (which would presumably require 200VA)


>The maximum power supply voltage for the UCD180
>is specified at + and - 50V. So a transformer that gives
> 35-0-35 would be close to the edge, giving about 48V
> DC after rectification. I personally would take a bit
> more margin and go for say a 30-0-30 transformer,
> that would give about 41V power supply.

Would the modules' performance vary with a smaller supply voltage? (eg 41V vs 48V)


also, is it possible to use two smaller transformers of the same type instead of one larger one? eg. 2x300VA instead of 1x600VA? if so, how would you connect them? in parallel? before or after rectification?

i have a pair of 330VA 45-0-45 toroids, and would rather use them than buy something new, if i can. one alternative is to get UcD400's instead of 180's. Gertjan suggested these would deliver about 62V after rectification. would that be too much for the UcD400's?

cheers for all your help!

ben

[classd4sure]

Usually it would be into an 8 ohm load, that's why should you attempt to bridge them, or halve the load, you might get only 1/4 the power increase you expected to.

You can tell this by looking at the amp ratings for both an 8ohm and 4ohm load.

I would think it can also contribute to the nasty clipping issues all CE amps seem to have.. you can't turn them up the power just isn't there.

I remember with my pioneer amp, if a series of fast bass hits were to occur at its higher power level, the first one would be far louder then all those which followed, reproducible with any song.

That's when you start looking into making your own amplifiers so you can make the decisions not to cut such corners in order to save a dime, which is why manufacturers under rate their power supplies in the first place.

[UrSv]

Originally posted by Bruno Putzeys


In consumer electronics the general rule of thumb is to use a transformer with a VA rating half of the amplifier rating. Fully-loaded output voltage will be roughly 80% of unloaded voltage, so if a supply is designed to have an unloaded output voltage of 90%*Vmax (to allow for Vmax at 10% mains over-voltage), fully loaded with nominal mains lands somewhere at 70% of Vmax.

The worst test normally applied to an amplifier is first to let it run at 33% of rated output for an hour, then to have it deliver full rated power for another 5 minutes. After this test, a transformer rated as per the above will be hot (70-80 ºC), but well below its rated maximum temperature

Now you know this, you share the full extent of power amplifier expertise most CE manufacturer's "power amp gurus" have. They guard this knowledge closely.

I'm aware of this and my remark was actually in relation/reference to the opinions normally expressed here which usually are quite different than the rule of thumb you posted. I think rosbacke would be quite happy with the UcD400s running off of his existing supply which was the point I was trying to make. The commercial rule of thumb has been discussed before but most people here like something not quite so commercial as you might have noticed...

Thanks for your great input in this thread and others.

[phase_accurate]

The commercial rule of thumb has been discussed before but most people here like something not quite so commercial as you might have noticed...

I have purchased two 500 VA transformers with 2x32 Volts for supplying 3 UcDs each. From the average power requirements this would not have been necessary. But I did so because of the better "load regulation". The UcD only has a PSRR of about 36 dB, so I think my active crossover solution will profit from this.

Apart from that, amps with generously dimensioned PSUs usually have a little more authority.
OTOH switching amps usually excel in this respect from the beginning .....

Regards

Charles

[Bruno Putzeys]

Originally posted by JohnW
Yes – I’ve also seen the “Ringing” – “clamped” (you know with a bad design as the modulation index increases) - but is this safe and advisable to operate continuously like this?

Not only bad designs have overshoot... Good designs delivering 30A also produce overshoot. Think of it: 10...20nH circuit inductance, 30A current... No way you can stop that from overshooting, apart from using these fancy top-cooled SMD fets (e.g. IR DirectFET). Ringing is a different affair. If you have it (ie when using through-hole packages) snubbers do magic.

I've been talking to MOSFET designers about operating FETs in avalanche like this. Having properly analysed the whole thing we agreed there's nothing to worry about.

Cheers,

Bruno

[JohnW]

Hi Bruno,

Thanks for the info. This then leads to the next question - do the FET’s have to be "Avalanche rated" – or will "normal" FET’s be OK? The reason I ask is I get conflicting advise from the MOSFET designer guys – they seem to be concerned about the edge termination on the Die…

30A…. thankfully I design only for consumer applications – leadless SMD fine…

Cheers,

John