Does anyone care to ask an amplifier designer a technical question? My door is open.

Thank you for your kind offer, Roger.

I build solid state amplifiers, and like to match output transistors. At first I matched for HFE, then VBE at constant current similar to operating spec. Only later did I realize how much this parameter drifts over the first hour of warmup.

However, my latest amps are Class A, and I suspect that a more realistic match is obtained by culling outliers by HFE, then match from VBE using the bias at constant potential and sufficient to generate the operating current. Finally, instead of using matched emitter resistors, I use emitter resistors tailored to the output devices, so that each emitter resistor sees the same potential drop.

Your thoughts? Any advice appreciated.

I was measuring HFE with a handheld, not warm, not at operating current, and only using the result as a first cut to cull outliers.

Emitter resistors are 0R1 +/- 0R02, so not much.

4 x PNP and 4 x NPN.

No VBE multiplier, but then I leave my amps on most of the time, and they stabilize below 50C.

Thanks very much for taking an interest, Roger.

In case you are wondering, the 25V rails are to accommodate Quad ESL's. With 75:1 step up transformers (Plitron), the amps can only punch the panels to 3750V, exactly half the level which engages the protection circuit. So I can bypass the protection circuit with impunity, and get a remarably refined ESL sound.

20 C/122F is nicely cool, only 25 C above ambient in a standard room. You either have big heat sinks or not so much idle current.

Do you mean resistors from 0.1 ohms to 0.02 ohms? I like the American system of resistors. That is a large range. Do your account for this and make the idle currents equal in each transistor?

HFE for a power transistor at low current is rather meaningless. Why not do all of it at operating current?

A VBE multiplier is a very good thing and easy to add. How much does your total idle current vary from cold to hot and how long does it take to get there? What is the idle dissipation?



Big heat sinks: 250 in2 , plus similar sized black anodized aluminum end pieces, per mono block.

I tried to indicate 0.08 ohm to 0.12 ohm. Mostly tighter around 0.1.

(Blush) Because it’s easier (blush)

Idle current diminishes from 1600mA to 1200 mA, in the course of one hour. Rails are +25V and -25V, so dissipation 80W reducing to 60W.

Thanks for the nod ... I guess the takeaway is, "Don't be so lazy. Match HFE at operating temperature and current." Thanks!

Thank you, Ralph!

Roger, you are being very generous with your time and your expertise. Thank you!

You have given me a container load of practical information which only comes from practical experience. Much appreciated! As requested, here are some details of my system.

My system is 98% analogue. Phono is integrated with the preamp. Twinned MATs reduce the noise floor. All caps in the signal path are air gap or vacuum, including RIAA. Battery power, filtered.

Speakers are 5 (soon to be 6) Quad 2905's. These have been modified for the Plitron step-up, which I will revisit - but that protection circuit is not at all transparent! It desperately wants attention, if not removal. The small disgusting ceramic caps in the phase ladder are being replaced with air gap (teflon is a big improvement on ceramic, but coloured). The larger disgusting ceramic caps are replaced with a chain of RTE styrene F&F caps. Air gap caps are made in the usual way.

Amplifier power comes from a big variac driving an isolation transformer, driving power transformers. Rectifiers are Ixys, feeding LCLCLC filters. Theoretically at least, this brings ripple down to the level of intrinsic noise.

Amp topology is reminiscent of the newer Brystons, but not close to identical. Transistors are AD, Toshiba, Sanken. All resistors are nude Vishay or DIY nichrome.

Theory as well as experiment have led me to suspect that much of what we hear is dielectric-sourced distortion. Accordingly, my caps are air-gap or vacuum wherever possible, styrene when not. Wiring is insulated and protected with unshrunk shrink-wrap, except at the ends to prevent corrosion. Just my poor observations in part repayment for your generosity.

Your expertise is highly valued, so anything you have to say is welcome. Thank you!

Jumped on the bandwagon? No - Shanghaied. Didn't want to be convinced, but the two-alternative forced choice paradigm left me no choice. Like when your optometrist says, "Better or worse? Better of worse?". Easy to design a single-blind experiment. Did so with resistors and caps. Results were clear with some components always preferred: air-gap and vacuum caps, nude Vishay resistors.

A justly famous person's apparent preference, a species of Caddock resistor, had large minimum orders, and so was not part of the test. His other apparent preference, teflon, always scored highly.

Agreed, not much AC - but conversely, not much distortion via dielectric absorption is required to degrade a small signal. 

Confirming idle current falling monotonically with time. Can't post the schematic because it's not mine and it's proprietary. Sorry. Hope you understand, even though it's poor payment for your generous help.

Each Quad has its own mono block. Two Quads facing ahead, and two Quads at one radian angles, facing in. A bit like one of Walker's experiments. Think of two sides of a hexagon on the left, mirror image on the right. And two at the back of the room, attenuated, but important.

Agreed that polypropylene is best for the money. Improvements on polyp are incremental and absolutely not worth it - unless you are a hobbyist who wants to see how far he can go, just because. 

What I see in Plitron (theoretically) is two things: first, the two plates of the ESL are driven by the same core, so they can be precisely balanced; and second, they offer 75:1 step-up instead of 250:1. Since I committed to 75:1 with 25V rails, going back to 250:1 would involve re-engineering the protection circuits.

Rather than that, I would try to build high voltage amps to drive the Quads directly. Wish me luck - and long life!

What I hear (subjectively) is that the Plitron 2905's are much brighter than the remaining stock 2905, without sounding harsh. Piano on vinyl sounds quite similar to the instrument upstairs on the modified units. The unmodified unit sounds mellow but imprecise compared to the Plitronned. In fairness I must add that this was accomplished with the help of nichrome wire and a small inductor to equalize primary impedance over the frequency band, as per the Vandersteen data sheets.

My DC offset used to be 50mV or less. What it is now is a good question, one that you have prompted me to ask. Thank you.

Thank you for your interest, Roger. I take your comments in the helpful spirit in which they were given.

You have motivated me to look again at the stock Quads, and think about driving them with my current amps. These amps have proven themselves to be very stable over the years, and don't really need protection circuits in the speakers. I think. So I just may continue to do without. To be clear, I am not using the Quads' protection circuits.

Near to SF is relative. Emotionally nearer than geographically. I did post-graduate work at UC Irvine, but then moved back home. I am now 1000 miles away. But what's that to a man with a thumb, as we said in the 60's.

I would be interested in a DD amplifier, but I am somewhat frightened of the high potentials. The one thing I know about super-KV work is that it's qualitatively different from sub-KV work. Perhaps you could point me to your website?

Thanks for all your help, Roger. You really have opened up a stack of stuff for me to do. Much, much appreciated.

Hate to be pedantic, but I think that the 45uF figure should be 4.5uF, which is less of a problem: (75**2)  *  (800  *  10**-12)  = 4.5  *  10**-6. Your other concerns with the Vanderveen (got it right this time?) transformer remain an area of intense interest (and research!) to me. Thanks again!