Power supplies on the Leach amp

Can somebody give me a little refresher on power supply design ?

The amp requires +/- 56-59V rails. Currently, I have the Victoria Magnetics 80VCT 6A toroid connected to the BR, with 1 Panasonic TUP 80V 12,000uF on the + and - rails for a total of 24,000uF.

The speakers I am really considering are the Martin Logan Aerius i's or maybe a larger model if i've got the cash at the time. As you may know these speakers exhibit a 4 ohm load at low frequencies which drops below a 2 ohm load at higher freq's. I do not intend on bi amping these.

My amplifier will not current limit until just below 2 ohms, but I am a bit worried that the power supply isn't sufficient for a true doubling from 8-4-2 ohms. (120-240-480wpc.)

I have thought about just buying another of the same toroid and making a true dual monoblock design. Will this be sufficient? It seems to me that if I wanted 480wpc @ 2 ohms I would be looking at xformers in the 900-1100VA range for each channel? Is this unreasonable? Do I really need to double my power to a 2 ohm load? Leach suggested using a toroid with two secondaries wired in parallel. Is this reasonable? Expensive?

On to filter capacitors. I have seen several people mention the've got 50,000uF or more per channel of their amp. Are there any downsides to running a lot of filter capacitance other than cost and the gigantic current rush at turn on? What is considered overkill?

Is there going to be much of an improvement if I use better quality filter caps? I know mine are about as cheap as they come.

Lastly, how much does the voltage drop from the toroid through the BR and caps ? all my audio book says is 'account 10-15%') does it get any more scientific?

I went into this project blindly and am now trying to fix some of my mistakes. ceramic caps! haha.

A bit off subject, but with electrostatic speakers, is the unusual impedance curve there because of lower high frequency sensitivity? Or is it an inherent aspect of electrostats and I need to be concerned that my highs are 3-6dB louder than the rest of the spectrum?

Thanks a lot!



2001-06-01 4:53 pm
For a brief background on power supplies check out the appropriate article(s) at http://www.sound.au.com.
I don't think you need to worry about having to pump out 480 watts into your speakers because the impedance drops at higher frequencies; only part of the power is going there.
I'd say go ahead with the speakers, then add another power supply later if you think you need it.
I'm not an expert but I think more capacitance would be a good idea for your system. It may be the best value for your dollar at this point.
Hanging around this discussion group is a great way to become informed and get a lot of references on audio topics.
Current limiting isn't the end of the world. As long as you're not asking the amp to play too loudly you won't even limit. The power supply will deliver current until it can't--you'll be okay up to some arbitrary wattage level, then the amp will run out of steam. It's not the kind of thing that will dog you at low volume levels.
Keep in mind that current limiting can come from the design of the circuit, too. Take Nelson's Zen design, for instance. It's biased at something like 2 amps, which represents the theoretical max that it could use to drive a speaker. No matter how large a power supply you put on a Zen, the only way to get more current out of it will be to raise the bias current.
On the other hand, a 2 ohm load is getting pretty fierce, particularly if it's largely capacitive. Electrostats are nothing but big capacitors, in the same sense that dynamic drivers are inductors, and ribbons are resistors. That impedance is complicated by the step-up transformer and by any additional circuitry that the manufacturer might choose to add. If you use the standard formula:
R= resistance (actually, in this case it should technically be Z for impedance)
PI= 3.14159
F= frequency
C= capacitance (note that that's in Farads)
you'll be able to find out what the Martin Logan's panel impedance is at any given frequency--at least the capacitive component. I doubt the company publishes the C of their panels, but you might be able to get it out of them if you called. Now, M-L will have done a few things to try to even out the impedance curve, but since anything they do along those lines will also effect the sound of the speaker, their options are limited. So you end up with a 2 ohm load with a large capacitive component.
Note that some amps act pretty squirrelly under those conditions. How the Leach behaves, I do not know. you might try asking M-L whether they've ever had stability problems with that amp. They'll have a list of amps that they've run into trouble with over the years; it's not the kind of thing you'd expect to see widely publicized. N.B.: I'm not criticizing M-L, it's the nature of electrostats to be a difficult load. It's part of the tradeoff for what they do well--few things in life come without a downside. (Do try to hear the Sound Lab electrostats before you make up your mind on the M-L. Some of the best sound I ever heard was placed in my lap via a pair of M-1s a few years back...)
Buying a transfomer with roughly double the VA you expect to use is a good rule of thumb. Ask a transformer to deliver more current and it will (to a point)...but it will get hot. And heat equals eventual reliability problems. No, it's not cheating to use a fan to circulate air around your transformer, but fans are noisy, and transformers aren't built to benefit much from forced air cooling. It's your choice.
As far as caps go, you've pretty much covered it. Cost and inrush current are the factors to watch out for. I suppose you could add physical storage space, as caps tend to be bulky. How much is too much? There's no hard and fast answer to that. I once had something like a quarter-Farad tagged onto one of my old Thresholds. I left the inboard caps alone and used a soft-start circuit for the outboard caps. Sounded wonderful. Cost-effective? For me, yes. The caps were free--I pulled them out of some old mainframe computer gear that was going to be thrown away. If I'd had to pay for them, it might have been a different story. To me, 12,000 uF per rail sounds a bit scant for a 120W amp. 50,000 sounds more like it. Most high end amps will be along the lines of 100,000 uF or more (total for both rails). Cap quality does make a difference, but hits hard in the wallet. Consider also bypassing the electrolytics with at least a few uF of film caps. (Even more bucks. Don't let me catch you robbing any banks...)
10-15% voltage drop across the bridge and caps sounds pretty excessive. Unless there are bleeder resistors across the caps, I'd say no more than two or three volts--it's just the junction drop across the diodes in the bridge and a smidgen for DC losses through the wiring harness. If you're dropping 10% through an electrolytic, it's time to replace the part.

Thanks for the replies.

I suppose the best thing for me to do at this point is add quite a bit more filter capacitance. It doesn't take much power to make any speaker loud in an apartment :eek: Maybe I'll concentrate on the toroid when I move. I guess 60-80,000uF sounds reasonable given my current financial status. Would a budget of $150-200 be reasonable? What are some quality brands and where can I get them? A lot of the people who were building their amps at the same time as me used mallory caps, which were about twice the size of mine for the same rating (and 3x the money). For now I have plenty of room in my chassis. If/when I decide to get another transformer I may have to use an outboard ps chassis.

As a side question, has anybody ever seen a 15A 115V AC switch that has a built-in blue LED pilot light ? I have been looking all over for something like this because I really do not want to drill two holes in my beautiful chassis.

Thanks again for the input.