For those interested in where the meat is: Zo = 5R is very obviously an error. Clear for several things:
- The circuit diagram, anyone able to read a circuit can see that;
- The typ. OL gain drops 16dB when, in open loop, going from a 10K load to 600R load. Would fit reasonably with a 5k OL Zout.
I send off an error report to TI.
Jan
- The circuit diagram, anyone able to read a circuit can see that;
- The typ. OL gain drops 16dB when, in open loop, going from a 10K load to 600R load. Would fit reasonably with a 5k OL Zout.
I send off an error report to TI.
Jan
Yes. The comp cap from output provides increasing feedback with frequency, lowering Zout with frequency.
Such a configuration is often used for an opamp with unlimited cap load capability.
Jan
Such a configuration is often used for an opamp with unlimited cap load capability.
Jan
0201 is manageable by machine. At my office we have made designs that have between 100 and 200 0201 passives each and had no issues.
Agreed. 0201 is routine for modern pick&place machines. The only reason larger packages are used is to account for voltage and thermal de-rating.
No I found it much more interesting to use this technique to get very good results with very cheap and mediocre opamps.
Blunt truth is that the needed qualities of the master and slave op-amps does not put any of them into the mediocre range.
Last edited:
b_force probably meant to write "Now I would find ...".
Indeed there is potential. You can use $0.30 opamps eg in a x100 gain stage as per "Feedback Plots Define Op Amp AC Performance", Fig.9, thus optimising bandwidth and distortion vs cost. The 1/ß approach to compensation is a very clever idea.
For the G/P and Syn08 composites which are dominantly Aol-controlled type of composite, the issue is the required bandwidth of the slave opamp. If it isn't large enough the net benefit at 20kHz isn't that great.
That appnote is an all-time favorite of mine, btw. Notably the section "PHASE ONLY MATTERS AT THE INTERCEPT" is hard to swallow at first but then becomes a real eye-opener.
Indeed there is potential. You can use $0.30 opamps eg in a x100 gain stage as per "Feedback Plots Define Op Amp AC Performance", Fig.9, thus optimising bandwidth and distortion vs cost. The 1/ß approach to compensation is a very clever idea.
For the G/P and Syn08 composites which are dominantly Aol-controlled type of composite, the issue is the required bandwidth of the slave opamp. If it isn't large enough the net benefit at 20kHz isn't that great.
That appnote is an all-time favorite of mine, btw. Notably the section "PHASE ONLY MATTERS AT THE INTERCEPT" is hard to swallow at first but then becomes a real eye-opener.
Is Samuel's super composite op amp good for anything in audio besides audio analysers?
How about for an instrumentation amplifier?
Or for a LOMC phono?
tim
I don't know anything about SS/Op amps et al
How about for an instrumentation amplifier?
Or for a LOMC phono?
tim
I don't know anything about SS/Op amps et al
Headphone driver would be a very good use for this. Stability is a concern (load capacitance), though, and must be addressed.
Phono input with a single stage (like shown in the LT115 datasheet) could be another application. Again, stability conditions have to be checked.
Phono input with a single stage (like shown in the LT115 datasheet) could be another application. Again, stability conditions have to be checked.
I have most articles and a couple of books by Gerald Graeme. A man who should be read more, but for some reason never got the attention like others in the field.
His writings often offer insight rather than facts and figures.
Jan
I think, we should be discuss cheaper solution of this composite op-amp.
Example: Input stage ca be a good op-amp and output stage can be discrete op-amp. Because discrete op-amp difficult to achieve a good DC precision or very low noise, but easily achieve high slew rate and high output current capability. Target distortion 120 ~ 130 dB at 20kHz should be enough.
Example: Input stage ca be a good op-amp and output stage can be discrete op-amp. Because discrete op-amp difficult to achieve a good DC precision or very low noise, but easily achieve high slew rate and high output current capability. Target distortion 120 ~ 130 dB at 20kHz should be enough.
Last edited:
That's one of the physical books I have in my library. Thanks for the PDF, I'll put it in my personal cloud ;-)
Jan
Jan
That's correct, unfortunately it's not possible here on the DIY audio forum to edit your post after a while.
Did you already tried this with some very jelly bean components?
I was more thinking about the $10 cent range 4558 series/TL072 etc
Last edited:
So you have been making yourself useful. Again. Great project Jan. Now I need even better ears.
This technique really comes into its own when you have the bandwidth to squeeze in all that extra loop gain, which sadly precludes a 4558. At least at modest gain values. You might be able to better leverage for an active phono preamp, but it's not a problem I've studied.
The control opamp gets all the glory but the challenging part is getting the slave opamp to work well, and it really needs to be *fast*. I don't know of any older jellybean high current opamps with 20+ MHz GBW. Otherwise you have pull almost all the gain out of the global loop to keep things stable.
Even before that there was the applications literature from Teledyne Philbrick in the 60's with a similar approach, and some of that might even be lost.