Discrete Opamp Open Design

[bcarso]

Originally Posted by linuxguru

There were several until Hollywood released 'Sideways' and screwed my happiness. Now the total number of US-imported cases of French Pinot Noir is 100x the official numbers of *all* production of Pinot Noir in France.
Do the math, most of the stuff is likely to be Syrah-Merlot rebadged as Pinot Noir - most sheeple can't tell the difference. Real Pinot Noir has a distinctive steely edge in the finish - a bit like licking antique ironware.

The 2-buck *real* Pinot Noir, but grown in Coastal CA or OR, was Nathanson Creek, available at Safeway. That's one of several inexpensive gems from the US that I miss in Bangalore; Ghirardelli chocolates are another.

But Sideways did make many fine Merlots more affordable No effect on top-notch Pomerol or Saint-Emilion however.

[linuxguru]

Originally Posted by bcarso

No effect on top-notch Pomerol or Saint-Emilion however.

Yup, the great Bordeauxs were never sold as Merlots or even listed their varietals on the label. Not that it would have mattered, a great vintage is a great vintage, regardless of the varietals. 'Sideways' probably killed the lesser California Merlots, and made some better ones affordable.

[Kindhornman]

Glad you all love your red wines, me personally I get sick from any red wine, sulfate intolerance on my part. Guess that means I will never purchase any of those expensive bottles, just give me a nice white or blush and at least it won't make me sick.

[myhrrhleine]

Originally Posted by kgrlee

Sonitus, will you oblige us with your LTspice file so us wannabe SPICE gurus can have a play? Also your Rush-Rush version?

On 2nd thoughts, 210mA peak may be a bit too stringent. c2240/a970 & bc556/546 are only 100mA. 4401/3 are OK but I was trying to get a series of tests for the contenders which didn't need too many special devices in special places. Special i/ps are unavoidable.

Got a link or sketch of Cordell's output stage accessible to beach bums?

Low loads aren't a problem if you are prepared to use medium power devices like the original 990 or even just 4401/3.

Scott, did you get your breadboard up & running?

My contention about offsets is that you may need tweak pots to avoid Volts of offset rather than mVs even with loadsa device matching. But this is just from SPICE so the breadboard man has the last (??) say.

Bob is a regular contributor here.
He might even be reading along in this thread.
But his amp is at http://www.cordellaudio.com/papers/MOSFET_Power_Amp.pdf
Hope this helps

[scott wurcer]

Back to circuits. Just fired up the output stage only, I want to measure its open loop THD, etc. That's from the two transitors whose bases connect to the VAS to the output. No matching, 4401/4403's right out of the bag 1% RN55's and no bias trim for now. I built my board with no thermal considerations yet either.

On +-15 fired up at 27mA and crept up to 28mA this will come down with trim. The offset was only -0.0021V. I will be doing the measurements over the next few days. I will be looking for the openloop distortion with and without a large source resistor.

[Joshua_G]

Originally Posted by scott wurcer

Back to circuits. Just fired up the output stage only, I want to measure its open loop THD, etc. That's from the two transitors whose bases connect to the VAS to the output. No matching, 4401/4403's right out of the bag 1% RN55's and no bias trim for now. I built my board with no thermal considerations yet either.

On +-15 fired up at 27mA and crept up to 28mA this will come down with trim. The offset was only -0.0021V. I will be doing the measurements over the next few days. I will be looking for the openloop distortion with and without a large source resistor.

Hi Scott,
Thumb up.

[fas42]

Also, for kgrlee and anyone else, just to say I'm still working on the "more accurate" spice model of BF862; the curves for drain current vs. both gate-source and drain-source, typical case look pretty reasonable; requires a subcircuit definition to work at the moment.

And, plugging it into SW-OPA on a first attempt causes barely a ripple: yes, the quiescent currents were thrown out of kilter, but after adjusting the appropriate R's to close enough, the 2nd and 3rd harmonics for 20KHz, 10Vp-p, 100x gain had changed by less than 1/2 db ...

Frank

[RNMarsh]

Originally Posted by scott wurcer

Back to circuits. Just fired up the output stage only, I want to measure its open loop THD, etc. That's from the two transitors whose bases connect to the VAS to the output. No matching, 4401/4403's right out of the bag 1% RN55's and no bias trim for now. I built my board with no thermal considerations yet either.

On +-15 fired up at 27mA and crept up to 28mA this will come down with trim. The offset was only -0.0021V. I will be doing the measurements over the next few days. I will be looking for the openloop distortion with and without a large source resistor.

what is your choice for test equipment that you will use? Thx-RNM

[kgrlee]

Originally Posted by scott wurcer

The offset was only -0.0021V.

After initial excitement on your advice to twiddle the 20Rs in the output stage, I found that this didn't help things at all.

Twiddling the 250R for offset as you recommend is definitely less twitchy than twiddling the 30R source resistor. But still more twitchy than twiddling the current mirror in a 990 type circuit.

I think it will be wise for anyone doing PCBs to make provisions for a twiddle pot on the 250. Don't think we can match FETs accurately enough.

Frank, can you post your models?

So far I've been impressed by LTspice. It seems to at least qualitatively behave as I expect real life even if I don't take the numbers as Gospel. But I don't believe the input stage at all as swapping between all the models I have for BF862, J305, J111, 2sk170 doesn't seem to make any difference on these amps.

myhrrhleine, thanks for the link to Bob's stuff.

sonitus, thanks for your SPICE files

Brad, Scott & co., I don't see how yus guys can keep good wine for so long. Surely you can find an excuse to broach a bottle.

lee sipping his meths in Cooktown ...

[fas42]

Originally Posted by kgrlee

Frank, can you post your models?

So far I've been impressed by LTspice. It seems to at least qualitatively behave as I expect real life even if I don't take the numbers as Gospel. But I don't believe the input stage at all as swapping between all the models I have for BF862, J305, J111, 2sk170 doesn't seem to make any difference on these amps.

Can do, but be warned this has only had a rough first check - no guarantees in other words!

Note that this is a subcircuit - even though the icon is the same you need to replace the JFETs with this identically shaped item; give us a yell if problems ...

The results of modelling are only as good as the accuracy of the models; the more you add all the little parasitics and what have you the more real info you get from the exercise.

Cheers,
Frank