Re: Where to get a choke for a class-A single-ended MOSFET power stage?

"Morris Slutsky" <(E-Mail Removed)> wrote in message
news:9360f4f1-c838-41fc-bb63-(E-Mail Removed)...
> Thinking of a circuit like this. Simple and elegant-looking:
>
> http://img524.imageshack.us/img524/4038/simplefet.jpg
>
> Really like a very cheap version of a single-ended triode stage.
> Basic 'parafeed' arrangement except that no transformer is necessary.
> Resistors provide bias, negative feedback, and give the FET a bit of
> 'plate resistance' like a triode to lower output impedance.
>
> What's great about this cheap FET circuit is that it's cheaper than
> the tube equivalent, much, and might sound something like. No need
> for expensive output transformer. A cheap FET like a IRF510, compare
> to a 300B, yeah. No need for heater supply! And the choke . . . the
> choke doesn't need to be NEARLY as large. A parafeed triode amp, to
> get decent bass response, might need a choke in the tens or hundreds
> of Henrys range. Here, the current is much greater, the voltage is
> much lower, impedance is much much lower, so you can get by with a
> choke in the millihenrys. Should be easy to find one I think.
>
> But I'm not sure where. Any ideas where to get a 50-100 mH choke for
> this? Needs to take an amp or three of current without saturating.
> That's the hard part there I think. Lots of dual "common mode" 5-30
> mH chokes for sale but I imagine those don't take much DC current at
> all before saturating, they're meant for AC circuits aren't they?
>
> If I could find a choke this would be a nice cheap thing to build
> though. Easy enough.

Just got back on-line after being without power for a couple days due to
that nasty snowstorm that left over two feet of snow sticking to trees and
power lines...

Signal transformer makes a choke (CH-1) that is 100 mH at 1 amp, 1.5 ohms.
It weighs 2.3 lb and is about 3" cube.

I tried an LTSpice of the circuit and I could not come up with a
combination of resistors that resulted in any better than about 10%
efficiency. I don't see any advantage to this design except that the output
voltage exceeds the rail voltage when the MOSFET is off. Following is my
LTSpice ASCII file. I'm just curious if there is a real reason to use this
design.

Paul

--------------------------------------------------------------------

Version 4
SHEET 1 880 680
WIRE 192 -80 -80 -80
WIRE 192 -64 192 -80
WIRE 192 16 64 16
WIRE 192 32 192 16
WIRE 288 32 192 32
WIRE 384 32 352 32
WIRE 192 48 192 32
WIRE -80 80 -80 -80
WIRE 64 128 64 96
WIRE 144 128 64 128
WIRE 384 128 384 32
WIRE 64 224 64 208
WIRE 64 224 -16 224
WIRE 64 240 64 224
WIRE -80 320 -80 160
WIRE -16 320 -16 304
WIRE -16 320 -80 320
WIRE 64 320 -16 320
WIRE 192 320 192 144
WIRE 192 320 64 320
WIRE 304 320 192 320
WIRE 384 320 384 208
WIRE 384 320 304 320
WIRE 304 336 304 320
FLAG 304 336 0
SYMBOL nmos 144 48 R0
SYMATTR InstName M1
SYMATTR Value IRF510
SYMBOL res 48 0 R0
SYMATTR InstName R1
SYMATTR Value 400
SYMBOL res 48 112 R0
SYMATTR InstName R2
SYMATTR Value 100
SYMBOL res 48 224 R0
SYMATTR InstName R3
SYMATTR Value 10k
SYMBOL ind 176 -80 R0
SYMATTR InstName L1
SYMATTR Value 100m
SYMATTR SpiceLine Rser=.1
SYMBOL cap 352 16 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C1
SYMATTR Value 1000µ
SYMBOL res 368 112 R0
SYMATTR InstName R4
SYMATTR Value 8
SYMBOL voltage -80 64 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 24
SYMBOL voltage -16 208 R0
WINDOW 3 -14 133 Left 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V2
SYMATTR Value SINE(0 2 200 0 0 0 100)
TEXT -18 378 Left 0 !.tran .5 startup

"JJTj" <up yers.con> wrote in message
news:(E-Mail Removed)...
>
>>But a choke load is much more efficient than a resistor load.
>
> Yes, but more hum and magnetic interference.
>
>>And twice as efficient as an active constant current source.
>
> I'd bet you don't know how to design CCS(s) that
> interact with the circuit. Didn't Mr. Pass teach
> you anything? Maybe it was over your head. Was for
> me till I built a few. Once you see what happens
> with SE designs, and how current is used, you get the
> picture. I'm not talking single device CCS devices,
> which are little better then resistors, I'm talking
> controlled transistor/Mosfet devices that read the amp
> and adjust to the need. A choke just sits there doing
> the best it can. Fighting what the PS feeds it.
>
>>Only the choke load can actually give back the idle current
>>as necessary.
>
> See what I said above. A choke is a one trick pony, often
> done well, but a interactive CCS reads the circuit and does
> what is needed to keep the current flowing full tilt.
>
> Yeah, more parts. But better operation.

I was able to make an LTSpice simulation that gives about 37% maximum
efficiency with a 6 VDC power supply, 1.94 VAC input, and 3.13 VAC output
into an 8 ohm load, at 200 Hz. It appears to work from 20 Hz to 20 kHz. But
the biasing is very critical, and probably will be unstable with power
supply, temperature and device variations. I used an STD30NF06L MOSFET
which has a low turn-on voltage. R3 in the schematic is not needed, but the
V2 signal source should be low impedance.

I have some 100 mH chokes rated at 10 amps that I found at a Hamfest in a
pile of freebies. But they are probably 30 lb each.

I agree that this is not really a very practical design, and much better
alternatives exist.

Paul

--------------------------------------------------------------------------
Version 4
SHEET 1 880 680
WIRE 192 -80 -80 -80
WIRE 192 -64 192 -80
WIRE 192 16 64 16
WIRE 192 32 192 16
WIRE 288 32 192 32
WIRE 384 32 352 32
WIRE 192 48 192 32
WIRE -80 80 -80 -80
WIRE 64 128 64 96
WIRE 144 128 64 128
WIRE 384 128 384 32
WIRE 64 224 64 208
WIRE 64 224 -16 224
WIRE 64 240 64 224
WIRE -80 320 -80 160
WIRE -16 320 -16 304
WIRE -16 320 -80 320
WIRE 64 320 -16 320
WIRE 192 320 192 144
WIRE 192 320 64 320
WIRE 304 320 192 320
WIRE 384 320 384 208
WIRE 384 320 304 320
WIRE 304 336 304 320
FLAG 304 336 0
SYMBOL nmos 144 48 R0
SYMATTR InstName M1
SYMATTR Value STD30NF06L
SYMBOL res 48 0 R0
SYMATTR InstName R1
SYMATTR Value 6800
SYMBOL res 48 112 R0
SYMATTR InstName R2
SYMATTR Value 3900
SYMBOL res 48 224 R0
SYMATTR InstName R3
SYMATTR Value 10k
SYMBOL ind 176 -80 R0
SYMATTR InstName L1
SYMATTR Value 100m
SYMATTR SpiceLine Rser=1.5
SYMBOL cap 352 16 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C1
SYMATTR Value 2200µ
SYMBOL res 368 112 R0
SYMATTR InstName R4
SYMATTR Value 8
SYMBOL voltage -80 64 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 6
SYMBOL voltage -16 208 R0
WINDOW 3 -14 133 Left 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR Value SINE(0 2.8 200 0 0 0 10000)
SYMATTR InstName V2
TEXT -16 376 Left 0 !.tran .5 startup

Morris Slutsky wrote:
> On Feb 9, 12:05 pm, JJTj <up yers.con> wrote:
>
<snip>
>
> Maybe it was over my head, maybe it wasn't, you decide I guess. Sure
> I read his website already. And I guess I see his Aleph-type current
> sources as not really single ended amplifiers. In fact, I know
> someone with a Pass power amp and it sounds very very good. I could
> almost see switching from a choke design to an Aleph type thing. But
> that's really, really push-pull from my point of view.

Huh. No, I'd just say he's using the CCS as an "active anti-resistor."
It's just how the JFET is biased. I think JFETs switch current,
not voltage - like tubes do.

> Nothing wrong
> with that, I could have more than 1 transistor per channel - it'd be
> so hilariously cool to do it with 1, though, you can understand that.
>
> Efficiency is something that single ended designs often suck at. Push-
> pull stages - ideal class B is 78.5% efficient, ideal class A push-
> pull is 50% efficient, class AB lies in between. Ideal single ended
> class A with an ideal choke or transformer is also 50% efficient.
> With an ideal CCS load, 25%, and with a resistor load 12.5%.
>
> Pass has a few designs up on his DIY site that lie between these
> situations. He has, for example, resistor-loaded single ended amps
> (with horribly low efficiency). He also suggests use of incandescent
> lamps as loads, which have a strong temperature coefficient and could
> perhaps even be seen as something BETWEEN a resistor and a CCS load.

I saw that, and I am surprised they would not add horrible
nonlinear distortions. But then again, they've used filament
light bulbs as current limiters on PA speakers for years.

> He has CCS loaded single ended class A, the "Zen" amplifier.
>

That thing is awesome looking.

> And the Aleph is also sort of an in between thing, between CCS load
> and push-pull class A.
>
>> I'm not talking single device CCS devices,
>> which are little better then resistors, I'm talking
>> controlled transistor/Mosfet devices that read the amp
>> and adjust to the need.
>
> What the Aleph current source does, as far as I can tell, is sense the
> current into the load and adjust itself accordingly. It's responding
> to the bottom FET that actually is providing the transconductance
> which drives the load. The amount of current sensing/adjustment seems
> to be variable - from pure CCS on one side, to fully compensating load
> current swings on the other - at that point, what you've got is a
> pretty clever way to do push-pull without an explicit phase inverter
> stage.

Okay, then why does it look like it's just a bias element to me? Is
it actually "sensing", or it just literally adding a DC bias? A
CCS is like a voltage source with a Zener somewhere ( wherever
it makes the thing act as a CCS). It's just
essentially a regulator - but a current regulator. Voltage can vary;
current is constant.

> A lot like an SRPP or some push-pull pentode stages that don't
> require a phase inverter - works so long as you stay in class A. And
> so you can push it up to the same efficiency level as a choke loaded
> stage if you want to.
>
> I will consider such designs. I suppose chokes being expensive and
> heavy and all. But there's something kind of elegant and brute force
> about a single MOSFET just sitting there heating away.
>

Oh yeah. Cool as all get out. And I could be wrong, but it looks
like the thing swings *current* more than it swings voltage.

> Assuming I can keep a stable bias on the damned thing and get a decent
> choke in the first place, which might be pretty hard actually.

Heh. I don't understand how the choke helps much at all. Not in a power
path like that. I always think of chokes as filters, as big ferrite bead
arrangements.

--
Les Cargill