my take on the szekeres

[miket6000]

Not a silly question at all. That's exactly what I did with my SOMA2 (A glorified DC coupled szekeres with gain), both CCS and source follower user IRF620s. In all honesty I don't think there's a huge gain in both being the same part, as the FETs are being used in different capacities. So long as your follower keeps following and your CCS remains constant, the circuit will keep working. If getting a more constant CCS requires using a different FET from the follower, go for it.

Mike.

[Jen-B]

That's good to know, thanks for the info Mike.

(I searched for SOMA2 but couldn't find it. If you ever decide to post it you'll have at least one reader!)

Jen

[miket6000]

I'll throw it up for you this evening.

In the mean time I believe I owe AndrewT an apology, I believe his extra 0.7 derating factor may have been added in order to compensate for the pulsed current nature of cap charging rather than a repeat of the RMS/Peak correction.

0.7 is a rather arbitrary factor which is better determined through simulation. Even with constant output of 1.6A the peaky nature of the bulk caps charging means the RMS current drawn from the transformer will be somewhat larger.

Sorry for the hasty reply and confusion caused.

Mike.

[geekysuavo]

Originally Posted by miket6000

... I believe his extra 0.7 derating factor may have been added in order to compensate for the pulsed current nature of cap charging rather than a repeat of the RMS/Peak correction ...

in fact, simulations show Irms=4.6A and Ipeak=18A.

Originally Posted by Jen-B

... If you ever decide to post it you'll have at least one reader!

make that two. mike, how did you stabilize the output offset of your amp with temperature? many have complained of high dc offsets with the plain-jane dc-coupled szekeres...

Originally Posted by Jen-B

Just thinking out loud...

actually a very good question. as a matter of fact, i asked it myself when designing this amp initially. simulated distortion figures were better with an IRF640 in the ccs fet position instead of an IRF610.

~ brad.

[miket6000]

I posted the schematic last night. You can find it here:

http://www.diyaudio.com/forums/headp...-soma-2-a.html

DC offset was corrected by moving the output inside the feedback loop used for setting the gain.

I limited the RMS current in my power supply with a small resistor between the bridge rectifier and the bulk caps, you'll be surprised how little you need to really reduce the peaks. You'll have to take into account the voltage drop across it though. I use an 18V transformer to supply a 15VDC load.

[geekysuavo]

thanks mike for the link. interestingly enough, the reason i decided not to pursue an M^3-like design was its push-pull nature, but your design fixes that by combining se class-a with the buffered opamp topology... it seems i won't be satisfied until i've built one and listened thoroughly...

also, i've been simulating with lateral mosfets in place of the irf610 in the szekeres design, and the results are quite interesting. for one, the frequency response (both ends, esp. hf) is extended. obviously gain is increased due to the lower Vgs(th) values. but most interesting, the harmonic distortion 'character' looks at first glace to be more pleasing, with the same amount of 2nd harmonic and greatly decreased 3rd and up harmonics...

i've been simulating with the 2sk216, but its 500mA Id(max) isn't cutting the mustard (or muster... w/e, lol), so i'm going to look at the 2sk1058 in the sim next.

the obvious result of all this is a delay in pcb design until i can simulate and breadboard each design and find a clear victor.

~ brad.

[geekysuavo]

here are the quantitative results of my sims so far. the only problem with the 2sk216 circuit is a 0.032% THD, even though it's almost completely 2nd harmonic. (all further harmonics are at least -110dB)

~ brad.

[geekysuavo]

ok, i've simulated the 2sk1058 in the circuit and fiddled with optimum component values...

interestingly, the 2sk1058 performs much worse in frequency and phase response than any of the other mosfets. it does halve thd and present a nice distortion character, but hf response is ugly...

the next step is to look into (simulate) each of these mosfets (irf610, 2sk216, 2sk1058) in the SOMA2 circuit and see how they stack up there.

~ brad.

[geekysuavo]

so everybody knows, i haven't stopped working on this!

thanks to graduation, and now grad school, i've had little time (and money) to pursue this any further. but i have been able to simulate some designs which closely resemble the SOMA2. after looking closely at a few opamp/fet combinations, i've landed on the ad843 and 2sk216 as an optimal combination. i'm currently revising my schematics for this "v2" design.

the reason i'm going this direction is pretty basic: the trend in my latest szekeres amps was higher current -> lower thd, but it was getting a tad ridiculous. the v2 design ensures lower thd (thanks to the SOMA idea of moving the output FET into a GNFB loop) at lower dissipated power.

this does bring up an interesting question. these lateral fets seem to be made from unobtanium... any ideas where i can find them at a reasonable price? (borbely seems to be one)

i'll have a pdf up soon, hopefully.
~ brad.

[geekysuavo]

ok, here are the simulation results for frequency response, phase response, and power supply line rejection for the new version.

~ brad.