LM4780 Based amp with serious high end PSU?

[dviswa]

Hi Everybody,

I am looking for an LM4780 based amp with a seriously high end PSU. I have looked at quite a few threads here and have build a couple too. Not too interested in a lot of features, would prefer a minimalist approach except for the PSU.

Typical boards I see have a single bridge with maybe atmost 20,000 mFd. I have also seen some earlier treads that claimed just a few thousand microfarads with the right snubbers sounded better than higher amounts of capacitors.

I beg to differ. I am looking for something beefy, like the kind Papa Pass would approve. How about a CLC or better still CLCLC supply?

Has anybody seen one? Tried one? Please point me the right way.

Thanks,
Dinesh

[Damon Hill]

I experimentally tried actively regulated supplies, but it didn't seem worth the additional complexity. A resistive pi filter type supply and snubbers on the rectifiers is fancy enough for me, and much simpler to build. These amplifiers have very good PSRR.

[AndrewT]

Originally Posted by dviswa

I am looking for an LM4780 based amp with a ....... PSU. How about a CLC or better still CLCLC supply?

a standard rC filter works well.
An rCLC is a couple of steps up from there.
Try a 0r5, 4700uF, 3mH air core, 10mF to 20mF
The quality of sound comes from the last C.
The first C sees a lot of ripple and must be rated to survive this duty.

[dviswa]

Originally Posted by Damon Hill

I experimentally tried actively regulated supplies, but it didn't seem worth the additional complexity. A resistive pi filter type supply and snubbers on the rectifiers is fancy enough for me, and much simpler to build. These amplifiers have very good PSRR.

Hi Damon,

Wow, you tried active regulation? That is a little beyond my goals too. All I am going after is low ripple while still having decent transformer utilization giving us a good load regulation.

You are talking about CRC and I am talking about CLC, not much difference in complexity, I would say.

Originally Posted by AndrewT

a standard rC filter works well.
An rCLC is a couple of steps up from there.
Try a 0r5, 4700uF, 3mH air core, 10mF to 20mF
The quality of sound comes from the last C.
The first C sees a lot of ripple and must be rated to survive this duty.

Hi AndrewT,

With a 0r5, assuming the secondary is say 22-0-22, I guess you are looking at something < 40 amps of peak ripple Right? To take that kind of current, the PCB trace will have to be humongous or it will add to the 0r5. Should be fine I guess, at the first C, probably not OK for the last C, right?. Which Cap do you think would do well to take that kind of punishment. Given that the last C determines the sound quality, which make would be good there? What about ferrite or iron core inductor? Is it saturation or is it EMI the concern?

What do you think of pushing the idea a bit further say, 0r5 - 4700 mfd - 3mh - 10,000 mfd - 3mh - 20,000 mfd. Where do we start seeing diminishing returns. Nelson Pass used 15,000 mfd on driving merely 20ma through a B1 preamp. In Zen V5 his power supply is far more sophisticated and beefy.

The more general question is have you seen any designs out there that have focused on the quality of the PSU?

Have either of you built it? What would you say was the difference in sound?

[Ted205]

i've taken a similar philosophy with my bpa300.

i'm using two 22mf caps (44mf) per rail (dual mono) with snubber/bypass caps and snubbed diodes on my rectifiers.

I haven't fully tested it yet (still working on the rest of the amp). CRC is something i can fairly easily add at a later date to experiment.

Remember safety is important when working with these high caps and it all comes at a higher cost

[AndrewT]

Originally Posted by dviswa

What do you think of pushing the idea a bit further say, 0r5 - 4700 mfd - 3mh - 10,000 mfd - 3mh - 20,000 mfd. Where do we start seeing diminishing returns. Nelson Pass used 15,000 mfd on driving merely 20ma through a B1 preamp. In Zen V5 his power supply is far more sophisticated and beefy.

The more general question is have you seen any designs out there that have focused on the quality of the PSU?

either read and sift the truth from the lies or experiment.
There's no real shortcut.

[jcx]

comparing to Pass' DIY designs is a "category error" - his preferred simple SE circuits have lousy psrr and some designers are attracted to "retro" passive measures which piles on the over sizing required

line frequency power filter inductors are a relic from tube days, tubes run at small % of SS currents and big, expensive output transformers match the tube 100s of mA currents and 100s of V swing to 8 Ohm nominal loads - even then line frequency filter L at tube current levels are large and heavy

SS works directly at 8 Ohm dynamic driver impedance levels, requiring Amps of current from the supply - its a real waste of money to size line frequency filter inductors for these currents when using high feedback/high psrr chip amps

look at and understand the psrr # on the chip amp datasheet - you have ~90 dB rejection at 100-120 Hz ripple frequency

some project amps have ps/gnd routing errors that would limit the finished amp's ps noise well before the chip's 90 dB psrr #

today a "sophisticated" power supply appropriate to SS power amp for dynamic loudspeakers would be a power factor corrected switcher - they can profitably use inductive components because of the high switching frequency

[dviswa]

Ted205,
That's a nice build. I hope you will experiment and let us know what kind of difference you see with just CC and CRC. BTW, where did you buy those RIFAs.

AndrewT,
Thanks, will comply.

[Ted205]

the rifas came from Farnell / Electronic Component Distributors / Suppliers / Electronics, Electrical Parts, Electrical Components and Wholesale Electronics.

[dviswa]

JCX,

I am not sure where you are going with the "category error" thing. You do not seem to like inductors. Fine, consider the fact that in almost all the cases I have seen an inductor in a linear PSU, it was smaller and cheaper than the capacitors that surround it.

We are talking about quality PSU and how it affects quality of an amp. Before you dismiss it, you might want to consider this. What do you think is the difference between using a generic silicon diode that costs next to nothing and expensive soft recovery FRED diodes. Surely it is so small, measuring it is difficult and is not visible easily on a scope. Such a small difference when further attenuated 90 DB due to PSRR, it should vanish completely. Yet there are any number of anecdotal evidence here that they make a big difference in the emotion conveyed by chip amps. I would take the "specmanhip" of a product with a pinch of salt.

PSU improvements, from my own experiments, I know, I can clearly hear it, better still my kids can feel it and often describe it. To me that is real, since they do not suffer the placebo effect I might, because they have no investment / attachment that I have.

PI-filter (or CLC vs CRC vs C)
Pass said:
Either a CLC or CRC is still a lot better than C. As an example,
I have here a supply with 30,000 uF after the rectifier, then
.12 ohms, then another 30,000 uF. The difference on either side
of the resistor is about 10 dB worth of ripple, and that's a lot.