Jam Jar: 100W class A monoblocks

Having had good success building FirstWatt F3, F5 and J2 clones, it's time to move on to something with enough balls to drive my main system. My Maggies are pretty inefficient, and they're in a 20' x 30' room with 15' ceilings. The F5 sounds excellent, but only manages background levels.

The Maggies are nominally 4 ohms. My Cronus Magnum II (100 wpc) drives them pretty well, but doesn't sound as good as the F5, and lacks a bit of dynamics at higher volumes. This suggests I'd need a boat-load of watts to go singled-ended, and it's not cold enough for that even in Ireland. So the output stage needs to be push-pull, class A, with a boat-load of class AB headroom.

I figure it's also time to go off-piste a bit and do my own design. I haven't tried lateral MOSFETs yet, so I've designed both a lateral and vertical output stage which will be housed on a single board. (Moving the power connections will determine which is driven.)

I've also designed 3 different front ends: a KISS front end derived from the Pass BA-1 to drive the vertical output stage, a CFA front end vaguely similar to the First Watt F5 to drive the lateral output stage, and a dual-LTP front end derived from a late '70s Hitachi app-note (which was also the basis of the Pioneer M-22) which can drive either output stage.

The name originated in the topology being derived from the HPA-1, but it evolved from there and ended up being different. I kept the name because Jam gave me a lot of help on it (as did Pico and Mark and others over on diyAudio).

 

100W...Class A...you have my attention!

 

A pair of bridge rectifiers for the output stage; the front-ends are powered from separate transformers/bridges/regulators.

(There's also a big-ass thermistor at the bottom for inrush control.)

 

And you even regulate the power supply? Be still, my heart.

I've built a similar front end to your KISS variation (with the cascode BJT). Any reason why you wouldn't bypass their bias supply? I've had some luck connecting a cap from the base to the FET source, sort of like a Hawksford cascode, to feed any BJT base current error back to a common mode node.

 

I don't like capacitance without base stoppers, so I tend to use both or neither. Since the goal of the KISS front-end was simple-as-possible, I went with neither.

Now one could argue that I should have gone with both on the Hitachi version, which is more sophisticated. (And it would provide a nice comparison.) But in the end I still like simple unless I know I need something more complex.

The Hawksford cascode-like idea sounds neat, but is probably over my pay-grade at present.

 

Have you considered using a LT4320 based rectifier?
https://www.diyaudio.com/forums/group-buys/336572-lt4320-based-active-rectifier-66.html#post6093176

Although it strays from your keep it simple approach.
My next likely project is a class a amp with a 4320 rectifier.

 

No, that's something I haven't stuck my head around yet.

I did have an opamp running the regulators earlier, but decided if the point was to design my own amp then I should design as much of it as possible. So the opamps got changed out for discrete comparators. There was a TL431 in the bias spreader that got removed for the same reason. (Well, and because it's much-maligned. )

Not saying any of this makes sense. Just that that's what happened.

 

Nice!

Are you familiar with the F5 turbo? Also the Aleph X?

http://www.firstwatt.com/pdf/art_f5_turbo.pdf

https://sites.google.com/bucknell.edu/ericshometheater/erics-home-theater/aleph-x-clone-amps

 

Yep, familiar with both. I find my DIY F5 a little /too/ revealing at times, preferring the sonic signature of my DIY J2. JamJar is in some ways an attempt to marry the F5Turbo and J2.

I really like the design of the Aleph X, but my preamp (a DIY HPA-1) is single-ended, so X'ing in general didn't seem like an optimal solution (although it certainly eases all the voltage issues). But it's something I'd like to look into in the future....

 

I think Aleph X is a Super Symmetry design which does SE -> BAL conversion?

regardless, this is a cool project

 

Interesting. So if you just ground the negative input, the feedback side of the LTP drives the other half of the 'X'? I'll have to sim it; I don't yet have a deep enough understanding to grok it just by looking at it.

 

Here's the LTSpice of the KISS front end with a vertical output stage:



And the LTSpice of the Hitachi front end with laterals:

 

I haven't gotten very far on this, but I did get one of the output stage PSUs built out: