ICEpower amps

[muralman1]

This was about 3 years ago. It was the 500A which requires TWO power supplies - an unregulated one for the current stage, and a regulated one for the input buffer.

I have amps made with the 500A. I can't improve the sound anymore than I already have. I already told you, your CD player is a no no when it comes to bringing the best out of the 500A. Only NOS players will do. Even the NOS players need to be tweaked to bring out their best. Even the transport choice is critical.

Cables choices to use with ICE amps are few in number. No SCs with but a scant dielectric can be used without injecting a lot of gravely noise. Solid wire and braided wire will disappoint. Thin flat cable will amaze.

The power supply design must be first rate. The ones I use have 4 very large Capacitors feeding a bank of much smaller capacitors. Besides that is circuitry that adds up to perfection.

I haven't mentioned preamps. The abysmally low impedance of ICE amps necessitate a tube pre, or for best performance, the solid state Fire preamp. Other solid state preamps are a poor match.

I worked six years to find out facts I just enumerated. It was hit and miss. With this system I can demonstrate the lies behind cable sale ploys. I can show all DAC chips should be avoided except for NOS. I have been demonstrating these truths to skeptics. Anyone within 200 miles around come challenge me here in Sacramento.

 

[Bjorn]

Basically the lab I use for design - carpet on concrete floor, standard US-type dry-wall, just enough acoustic treatment by ASC (bass traps, diffusion panels and absorption panels), acoustic tiles in the ceiling (backed by fibre).

Thanks for answering. When one tests electronic is vital to have good acoustics and a decent frequency response imo. Otherwise, one really doesn't know what one is hearing.

 

[muralman1]

This just came in from my good friend, Henry, who has the same speakers, amps, preamp, ICs, and my SCs. I sent Henry my source duo, and this is what he had to say:

"I have been listening all morning and YES!!! It is incredible. It is unreal how good the Flatfish and the Audio note sounds. It has the right amount of ambience, air, staging. Bass is beyond description. Very tuneful and meaty with right amount of pitch definition. Vocal is NIRVANA to listen to. Nothing is exaggerated. No wonder you have been screaming how good your system sounds, unfortunately mostly to unbelievers. Oh well, Too bad they missed out big time. Top end is so exxxxxquisite...."

 

[coolnose]

Hello to everybody,

I am new to this forum and joined especially for this thread, but will have a look around to check out some more subjects.

I am currently starting to gather parts for building a non-regulated/linear Power supply for 2 x 500A modules.

I understand that muralman1 has some experience in this field, I already have noted the big lines: big reservoir caps and small, i.e. low ESR and fast behind that...

Now my question, what kinf of total capacitance do you use ? Is it a CRC or even CRC supply ? I am new to power supplys and don't realy know what to look out for concerning the specs...

I'd love to hear your comments.

Cheers,

Max

 

[amirm]

Now my question, what kinf of total capacitance do you use ? Is it a CRC or even CRC supply ? I am new to power supplys and don't realy know what to look out for concerning the specs...

Designing power supplies for class D amps can be very tricky. Unlike normal amplifiers, power supply variations in class D can result in increased distortion as it directly changes output waveform. Feedback reduces this effect but still, this can be tricky business. Here is a good albeit, somewhat technical article on it: http://www.eetimes.com/design/audio...g-Class-D-amplifier-power-supply-requirements

As to CRC, the series resistor will be useful in reducing power supply ripple. But unfortunately, it will also dissipate huge amount of power and reduce efficiency.

You are also dealing with pretty high currents here. Lack of understanding can lead to catastrophic failures including risk of fire and burns. So I would opt to buy a pre-built power supply and not attempt to design one from scratch if you are not absolutely comfortable with design of high-current electronic circuits.

 

[muralman1]

Designing power supplies for class D amps can be very tricky. Unlike normal amplifiers, power supply variations in class D can result in increased distortion as it directly changes output waveform. Feedback reduces this effect but still, this can be tricky business. Here is a good albeit, somewhat technical article on it: http://www.eetimes.com/design/audio...g-Class-D-amplifier-power-supply-requirements

As to CRC, the series resistor will be useful in reducing power supply ripple. But unfortunately, it will also dissipate huge amount of power and reduce efficiency.

You are also dealing with pretty high currents here. Lack of understanding can lead to catastrophic failures including risk of fire and burns. So I would opt to buy a pre-built power supply and not attempt to design one from scratch if you are not absolutely comfortable with design of high-current electronic circuits.

I would agree with what Amir says. The power supplies of the H20 Have storage caps plus two component filled circuit boards. I don't know what any of it does, though. One thing about ripple, The four huge caps second job is to quell ripple according the the builder. There is not a hint of audible distortion if you get it just right. If you do that, you have to be very patient in system matching.

 

[fas42]

Is it a CRC or even CRC supply ? I am new to power supplys and don't realy know what to look out for concerning the specs...

Unfortunately, to get the best bang for your buck in terms of components you probably need to do some modelling in a Spice type program. A CRC circuit will work, because only a relatively small value of resistor will do a great deal of good in reducing ripple, and you can live with the power wasted in that resistor. I've done quite a bit of playing in this area with Spice, and you can get some very good results. The biggest problem is that you have to know everything about the voltages and what the maximum currents will be, to get the right figures for a particular circuit to be supplied.

Frank

 

[mep]

I haven't mentioned preamps. The abysmally low impedance of ICE amps necessitate a tube pre, or for best performance, the solid state Fire preamp. Other solid state preamps are a poor match.

Don't you have this backwards? Tube preamps like to see a high input impedance, typically 100K. SS preamps work better with lower input impedance.

 

[coolnose]

Well,

thanks alot for all your answers.

I have already simulated a PSU using a 500VA 2x55V toroid for 2 ICE 500A units. Per channel: One Full Bridge using 4 x 30A 100V schottkys, 1 x 10000uF 100V cap + 1 x 4,7 Ohm 25 Watt precision resistor + 6800uF 100V cap. ESR of the caps are around 0,02 Ohm, output voltage around 78 V and something like 4mV ripple.

Just wondering if that ripple would do fine for feeding the ICE.

Thanks also for all your warnings, I actually did quite some PSU mostly for class A and just wanted not to ommit anything when designing them for class D.

One should maybe not say that nowadays, but I don't care too much about efficiency and heat... as far as I ahve taken them into account.

Cheers,

Max

 

[DonH50]

What's the output power?

For reference 100 W into 8 ohms is 28.28 Vrms, 80 Vpp if I did the math right. That means 4 mVpp is -86 dB relative to 100 W; I suspect that's low enough. The real issue is the ripple over frequency. Class D amps in particular generate a lot of HF energy and you'd rather not have that modulating the output, but you need to supply it to keep the switching edges clean. Those big electrolytics will not do a good job at supplying short-term (HF) transients;that low ESR is at LF. Probe the output of your supply and measure the impedance over frequency, assuming you can model the self-resonant frequency of the caps. I am betting you are going to want to add several smaller values of electrolytics and film capacitors (polypropylene, teflon, whatever) to provide wideband decoupling.

 

[muralman1]

Don't you have this backwards? Tube preamps like to see a high input impedance, typically 100K. SS preamps work better with lower input impedance.

That would make sense, but actual practice proved the opposite. It became a major concern for my amp builder. That is why he made his own. By the way, the H2O Fire is named that because it gets hot. There is one for sale at audiogon for a super deal.

 

[muralman1]

What's the output power?

For reference 100 W into 8 ohms is 28.28 Vrms, 80 Vpp if I did the math right. That means 4 mVpp is -86 dB relative to 100 W; I suspect that's low enough. The real issue is the ripple over frequency. Class D amps in particular generate a lot of HF energy and you'd rather not have that modulating the output, but you need to supply it to keep the switching edges clean. Those big electrolytics will not do a good job at supplying short-term (HF) transients;that low ESR is at LF. Probe the output of your supply and measure the impedance over frequency, assuming you can model the self-resonant frequency of the caps. I am betting you are going to want to add several smaller values of electrolytics and film capacitors (polypropylene, teflon, whatever) to provide wideband decoupling.

Well, that goes right over my head. I can say for HF travel, all four big caps are bridged with .1µ caps. I chose Soncicap Platinum caps for that purpose.

 

[DonH50]

Well, it was directed at the guy designing the supply, but it sounds like you done good (no pun intended!) You need big, slow caps for lots of long-term storage, and little ones to provide energy for those fast transients. Is that under the hairline? (A comment that no longer applies to me, alas... )

 

[muralman1]

Well, it was directed at the guy designing the supply, but it sounds like you done good (no pun intended!) You need big, slow caps for lots of long-term storage, and little ones to provide energy for those fast transients. Is that under the hairline? (A comment that no longer applies to me, alas... )

LOL, the bypass caps are there to slip the highs over the big cans. Besides those, there is a bank of medium small caps that provide quick punch lines.

 

[coolnose]

Well thanks to all for your answers on the PSU question. I think I will actually go for 2 big cans par channel with small bypasses each and on a bank of smaller sizer caps (lets say 5 x 4700uF or similar) for a lower ESR. All together in a CRCRC combination...that should do fine.

Would you mind telling us what your exact PSU config is, muralman1, as you seem to be very satisfied with it ?

EDIT: DonH50, I just read your post again... the 500A is actually something like 500 Watts @ 4 ohms, so I guess the 4mV will do fine... but as stated above, I'll do some more anyway..

Cheers,

Max

 

[muralman1]

Well thanks to all for your answers on the PSU question. I think I will actually go for 2 big cans par channel with small bypasses each and on a bank of smaller sizer caps (lets say 5 x 4700uF or similar) for a lower ESR. All together in a CRCRC combination...that should do fine.

Would you mind telling us what your exact PSU config is, muralman1, as you seem to be very satisfied with it ?

That would be close to right for a stereo amp. Mine are monos. I didn't build this amp. I am not sure I can know the PSU configuration. Here is a top view of the H2O.

http://pic8.audiogon.com/i/c/f/1299339194.jpg

 

[DonH50]

@coolnose: The real answer is the noise/distortion at listening levels. If that is only 10 W average, the ripple is only -76 dB, etc. assuming it is relatively constant at 4 mV (doubtful). You really need to measure the ripple current under load and over frequency to optimize your design. Given it's Class D, I'd be more worried about getting enough small, HF capacitors in place (say a bank of 0.1 uF caps) to control the HF switching noise from the amps than having a bigger bank of intermediate caps, but of course it all depends upon the switching BW and self-resonant frequencies of the caps. Broadband decoupling gets tricky...

 

[fas42]

Couple of easy rules for decoupling with caps to prevent resonance problems: have the values very close together and/or a considerable distance apart. To pull some figures out of the air: either less than 3 times or so difference in magnitude, or, more than 50 times difference. It's having value differences in the range between these two which can or will cause trouble ...

Also, keep leads as short as you possibly can. The smaller the cap the more critical for the component to be soldered exactly where it's needed. A wire run to the cap adds inductance, and makes it worse than useless!

Frank

 

[DonH50]

Having values far apart has always caused problems for me as it leaves a hi-Z "hole" in the middle... I generally prefer no more than a decade (one order of magnitude, 10 times) distance in values, but it depends upon the self-resonance of the capacitors. I have used some really nice broadband capacitors (Fsr > 45 GHz) but they are very expensive and small in value...

"Three times distance in magnitude" = a factor of 1000; did you simply mean a factor of three (and 50)?

I agree with keeping the leads as short as possible. Of course, years ago I told somebody that, and they clipped the leads to snubs then soldered 6" wires to them. "But the leads were short!" Ouch.

YMMV - Don

 

[fas42]

Three times, I meant a factor of 3, so, say, 1nF, then 0.33nF, and 0.1nF; yes, I shouldn't have used the word magnitude! To my mind, using a factor of 10 is getting pretty dangerous in terms of causing an impedance spike to occur. If you play around with a Spice program you can get an excellent idea of what goes on with the interaction of cap's when you include the parasitic resistance and inductance.

Yes, I have seen a modded CD player with high quality caps bypassing electrolytics, with the bodies beautifully suspended in space with lovingly formed patterns of wire to look just right; just the thing to cause a very nasty high frequency resonance and blow your tweeters!

Frank