Questions and rumbles about solid state amplifiers

[PRR]

> a topology that isn't from a hifi cookbook, or power IC datasheet. Less loop gain, less (or no) global feedback.

Most audio amps run Class B. It is very hard to hide the crossover distortion without ample NFB.

The JLH runs Class A. Nothing changes very fast, the way a Class B stage *switches* from top to bottom device every half-cycle.

The JLH can easily be re-rigged for little or no NFB (or for current-feedback). Though I'll wait to rack my brain until tca has one working at real power output and clean ("too" clean).

Tubes *are* an answer, for the silly reason that they appear happy when run HOT so are usually run much nearer Class A than typical transistor amps. When run into Class AB they don't typically go into crossover distortion until fairly high levels, and the effects appear over a broad range of level without the sand-on-cone tone of hard-switched transistors.

The JLH driver is heavily single-ended so even-order distortion dominates. While I don't think this is essential in the power amp (Bassman 5F6a driver and output stage is heavily odd-order), it couldn't hurt. (Tube Fenders get 1%-5% even-order in the first stage and/or a tone-driver stage. Late Ampeg V40 had a Fender-like and a dead-clean input, even I could hear more musicality in the crude Fender-inspired input.)

[tca]

So here goes my little JHL guitar power amp.



So far so good. It does get hot and it still needs some tweaking to make it sound more like a guitar style amp. It overdrives nicely (up to a point) and it gets loud. I'm powering it with a 18V laptop power source.

One thing that I've notice with discrete (class ab and class a) designs is that they do not complaint when using a L-Pad attenuator, something like this:

 

Chip amps don't accept all that well a L-Pad on the output, the ones I've tried start to heat up and then shuts down.

Cheers.

P.S.
I've not tested yet (just sim)  but it seems that by adding the first device (as the original JHL) similar power can be obtained with a smaller idle current.

[PRR]

That's not the cannonical JLH.

You have the top output (your T1) ham-strung as a constant current device.

Maybe not *constant*; but far from push-pull.

Maybe you have confused the boot-strap.

And adding the input stage can't change the power output significantly.

Wire it *just* like the original JLH. It's only a couple more parts, wired right.

And I don't think it will drive well without a preamp. Classic amplifier design is, like a concerto, in _3_ parts. Bring weak signal up to a decent level. Gain-pot. Bring decent level up to BIG level for speaker.

Don't know why chip-amps would hate an L-pad. They are fairly widely used in hi-fi remote speaker control.

[tca]

> You have the top output (your T1) ham-strung as a constant current device.
> Maybe not *constant*; but far from push-pull.
> Maybe you have confused the boot-strap.

You are absolutely correct, the schematic has an error, I've confused the boot-strap. Thanks.

It should be something like this (that is what I have on the breadboard, without Q1):


> Don't know why chip-amps would hate an L-pad. They are fairly widely used in hi-fi remote speaker control.
Now that you mention that, I guess the problem was that I've used the L-Pad when trying to overdrive the chip, and thus it shuts down not because of the L-Pad but due to the overdriving signal.

[tca]

Just corrected the schematic.

Cheers.

[tca]

Hi,
so the JLH version did sound ok... but not completely happy with it. It had some annoying fuzz/fizz/buzz when overdriven... that could be fixed, but if it is there why bother taking it out!  So I've turn to the most simple amp that I could think of, a mosfet and a light bulb at the drain (as PRR suggested). It sounded also ok, as I recall from my previous experiments. I've already seen some of the amps made by Nelson Pass (delite amp is the mosfet+bulb version) and specially the Zen amp that uses a active current source. Here is my hack on it (biased at 1A):



So, how does it sound? It is the most amazing amp I've build. It stable as a rock, no fuzz/fizz/buzz, it has an incredible warm sound, so amazing that I've plug it in to a hi-fi set of speakers (what a sacrilege!) and still is amazing! As R.G. point out earlier, there is no such thing as a specific guitar amplifier. It does get hot, much hotter than the JLH.

I found this topology in  "Design of VMOS Ciruits" book from the 80's (the same book that Jack mentions on his BS170 booster webpage) but Pass's patent for a very similar topology is dated from 1998. Go figure?!?

Cheers.

[R.G.]

it has an incredible warm sound, so amazing that I've plug it in to a hi-fi set of speakers (what a sacrilege!) and still is amazing!

That is the sound of a percent or so of second harmonic distortion. It's very pleasant, and part of the highly-sought-after tube sound. Self biased triodes, linear as they are, have a little of it. Single-ended tube power amps have a relatively lot of it.

Sorry I didn't think of it earlier. You'd probably enjoy Rod Elliot's pages on power amps. See http://sound.westhost.com/project36.htm for some info on a similar design, and see the rest of his articles for general reading in a similar vein.

[tca]

You'd probably enjoy Rod Elliot's pages on power amps. See http://sound.westhost.com/project36.htm for some info on a similar design, and see the rest of his articles for general reading in a similar vein.

Yes, thanks.

[tca]

I've been cruising some older entries on this forum and found this old topic: http://www.diystompboxes.com/smfforum/index.php?topic=51555 Very interesting reading! One of the most interesting comments  was done by Doug H, and I quote:

The thing I could never get around with 9v pedal circuits is the fact that tube pre's are usually running around 200-250v for their supply voltage. Then there is a tradeoff between gain and clipping headroom that you can't duplicate at 9v- unless you attenuate the guitar signal down to scale it for the circuit. So a 100mv guitar signal @ 250v scales down to a 3.6mv signal @ 9v, if you have the same gain characterstic with the JFET that you would with a tube.

Although the topic was about amp emulation with fets, the argument follows for any active device. Doug's comment confirms what I've tested in all the amps that I've done. If you attenuate the guitar signal down to scale it for the circuit (voltage power source) and keep the gain stages high enough the feel of playing with these settings is very expressive, dynamical and sounds real good. I've tested this idea with an opamp booster and common BJTs amplifiers, but there is a catch... the noise to signal ratio is very tiny and so noise is inevitable. Any thoughts on keeping the ratio  1/25 in a low voltage amplifier with a high signal to noise ratio?

Cheers.

[artifus]

hello mother nature?

[WaveshapeIllusions]

To be honest the only passable way would be to just raise the supply voltage. The noise level doesn't really go down too much past a certain point; -90dB is usually pretty good, which would still be noticeable with the guitar signal attenuated that much. I suppose a gating circuit might help a bit, as long as the noise is less than the guitar signal.

Of course all the different noise minimizing techniques apply too. A major one is keeping the impedances low. If you buffer the guitar before attenuating you can use smaller resistors, which will cut down noise. Good shielding helps of course. Hunt down Texas Instruments' Opamps Design guide. There's a full chapter on noise.

Another suggestion, I don't know if this will even work, would be a balanced circuit. Transformers would be best. Buffer and send to transformer. Take the signal and an inverted copy from each end of the secondary. Attenuate each by the same amount, though you could use the trafo to step down too. Run each copy through identical circuits. Sum at another trafo, maybe step up if necessary. If everything is matched well, perhaps the noise will cancel. Of course, noise is random so maybe it'll be worse too. I'm just guessing here though. No idea if it will actually work.

[PRR]

> perhaps the noise will cancel

Random hiss won't cancel. (As artifus says: You can not cheat Mother Nature.)

Power supply and other noise can cancel; there's usually other ways.

[KMG]

Some time ago I found very interesting MOSFETS IXTP01N100D.
http://ixapps.ixys.com/DataSheet/DS98809C%28IXTP-U-Y01N100D%29.pdf
1kV of the drain voltage, 25W power dissipation. That allows to use them with standard transformers for tube amplifiers.
I already received ten of them from arrow electronics.
Now I'm waiting for the ordered transformer http://www.classictone.net/40-18045.pdf to start testing.

[R.G.]

Also see at geofex: MOSFET Follies

And worry about the operating conditions. Tubes in general have "breakdown voltages" from plate to cathode that are well over 2x their maximum plate voltages. An output tube in a push-pull setup pulls down on its half of the primary with the center tap tied to the B+. The other half-primary, which is connected to the other output tube is pulled UP by transformer action. The tubes have to withstand nearly 2x B+ on each peak.

MOSFETs will pull the on-side down closer to ground, so the off-side will fly higher. 1000V sounds good, but is actually marginal.

MOSFETs have a characteristic somewhat like a pentode or power beam tube, OK, but are actually different beasts. An output transformer for a pair of MOSFETs is a different design than that for a couple of tubes.

And happy hunting.   

[KMG]

Also see at geofex: MOSFET Follies

And worry about the operating conditions. Tubes in general have "breakdown voltages" from plate to cathode that are well over 2x their maximum plate voltages. An output tube in a push-pull setup pulls down on its half of the primary with the center tap tied to the B+. The other half-primary, which is connected to the other output tube is pulled UP by transformer action. The tubes have to withstand nearly 2x B+ on each peak.

MOSFETs will pull the on-side down closer to ground, so the off-side will fly higher. 1000V sounds good, but is actually marginal.

MOSFETs have a characteristic somewhat like a pentode or power beam tube, OK, but are actually different beasts. An output transformer for a pair of MOSFETs is a different design than that for a couple of tubes.

And happy hunting.   

I know about these limitations, because I already made wide range of mosfet PA's with transformer output from 250mW to 90W.
http://milas.spb.ru/~kmg/fetpa_en.html
All of them sounds very close to tube PA's.
I look at these transistors because 2SK216 become EOL, so I need replacement for them.
http://am.renesas.com/req/support_products_list.do?layerId=555
For transformer, mentioned above PA requires about 300V supply with drive by IXTP01N100D. Twice voltage will be far enough from maximum limit.

[R.G.]

Good.

[tca]

And happy hunting.  

I really think that this "thing" called "tube sound" with SS devices, is a ill defined problem. There is not a unique solution, there is not a such thing as a "tube sound" stricto sensu, it is much more a physical feeling of a playful musical experience. I'm not saying that you can't get beautiful sounds by "copying" and building a 300V mosfet amp, it is not that at all. What I'm saying is that simple is best and also the best.

For instance, look at a "typical" tube amp:

1. Guitar in
2. gain stage -> tone control/tone shaping
3. repeat 2 or goto 4.
4. power amp
5. speaker

If you design a SS amp without the "tube sound restriction/bias" a lot of new sounds appear.

Cheers

[tca]

hello mother nature?

Mother nature works in mysterious ways! The best way to circumvent this is to have the power voltage source high enough, at least to get 20dB in the first gain stage: 20V?!.

[gritz]

I'm of the opinion that if something makes noises appropriate to what you're trying to do, does so repeatably and reliably, is justifiable in terms of expense and doesn't take Chuck Norris to carry it up three flights of stairs then I'm pretty pragmatic about what's inside.

[artifus]

Mother nature works in mysterious ways! The best way to circumvent this is to have the power voltage source high enough, at least to get 20dB in the first gain stage: 20V?!.

indeed she does. one of her odd ways is that most passive pick ups only put out .2 - 2 volts on average. we could raise that (say) 2v to 20v but what else do we raise with it?

i like this; lets consider this further:

For instance, look at a "typical" tube amp:
1. Guitar in
2. gain stage -> tone control/tone shaping
3. repeat 2 or goto 4.
4. power amp
5. speaker

this could be extended to any signal 'processor'

*gosh* that sounds a bit aggressive reading it back - i didn't intend it to. genuine request for open discussion.