opamp common-mode distortion - effect of source resistance

With Rin=100k

At 10k, the thd is approx 5x of the value at 1k, but at 100Hz it is only marginally lower, so, "not likely".

But the models I used are the standard ones that came with the prog, so may not show everything, and the schem I used may not be the "ultimate" platform to test and hypothesize, so - I'll bounce that off to the expert jury.

I observed this behaviour when I was developing some SE gaincards for myself, so it's apparently a more universal phenomenon, but hey, what do I know?

For one, with resistor in LTP instead of a CCS (as PRR mentioned), this behaviour is going to be a little more pronounced, so "more color".

Hey, Sam, perhaps you could whip up a real "test jig" and dig in when you find some time? Gotta run now.

What you're seeing is mostly the effect from Q1 Ccb. It's presence makes the input capacitance of the opamp vary with common-mode voltage, which manifests itself as extra distortion in noninverting configurations with large source impedances. The effect is ~proportional to frequency. With this simple opamp there are apparently other mechanisms involved (probably as you noted related to the basic base current of Q1) which make this hold up less well.

Samuel

PS: This was well explained in Putzeys' "Master Class Slides", but apparently these have disappeard from the Hypex server by now.

I don't think that the "varicap" mechanics doesn't hold well, I merely observed that there's yet another mechanism(s) involved.

Another thing that I wanted to point out (but have no means to test) is that bjt's hFE gets modulated with variying Ic, meaning that when you have a "enough nonlinear" LTP in a, say, suboptimal DOA, there will be possibly yet another source of unwanted behaviour.

But hey, I'm primarily seeking for "mojo" in my builds, sooo ... my primary concern is with nonlinearities that are cheap to build and make you say d.o.p.a.m.i.n.e.

You can buy clean opamps fairly cheap these days, or even get free samples, jay..

tv said:

Well, THB, the effect IS there, but, compared to the non-inv config, it is orders of magnitude lower (both in absolute terms and relative terms). So, I would say, "not plausible". To me, this looks exactly like I described, and it's almost stupidly simplistic. Probably the same "mechanics" behind it that caused designers to add "linearizing" diodes to OTAS, but here predominantly manifested on the "positive" input BJT of the LTP, i.e. its base getting non-linearly current-hungry at positive swing.

Of course, I might have done something very wrong there..
Also, the accuracy of "models" in question..

Click to expand...

I didn't trust sim at all, for more than finding crude mistakes, or designing complex filters. 
=====
The use of Gilbert diodes to linearize OTAs is because the input LTP is operated open loop and is the dominant source of distortion in OTAs.

In a GP opamp with negative feedback the actual voltage across the LTP inputs is quite small for most situations, so the gross nonlinearity is less of an issue with very high open loop gain. For some (internal) topologies or tradeoffs where the dominant compensation pole is low and well within the audio passband, this input LTP voltage and associated nonlinearity will change with both frequency and level (worse at HF).

=====
While not my cup of tea,,, if you are actually trying to make distortion for effect, you could probably modify the input stage to exaggerate the current change in the input stage more than it's differential output changes.

I also did some speculation about using more than 2 devices in the LTP (for other reasons). Throwing an extra device on one side and not the other, and/or connecting the outputs elsewhere could alter the transfer function (and alters more).

JR

Yeah, but there's question on where the "mojo" ends and where "distortion" begins. Personally, I'm in the "hair&mojo" camp and not so in the distortion.

This to me means nicely distributed sequence of 2nd to 4th harm. Some may argue that this is merely an emulation of "tube" stuff, but hey..

Can be done, most importantly sits in (a digital) mix nicely, and is suprisingly simple to design and build, if you know precisely which mechanisms are to be exploited. Surprisingly low self-noise, also.

In a DOA, I think that merely exploiting the "blessing" of replacing the CCS in LTP to a simple resistor and slightly unbalancing the collector loading will get very close to the mojo land - and actually cheaper.

You can do that in a 4-transistor "opamp" - but TBH with a little cheating, because one is going to be a darlington.

Regarding sims, I got a feeling about which things work 100% and which are only so-so, personally I test for the most crucial decisions by passing the audio files thru the sim first. It gets surprisingly close to the real thing. But I'm not one of the 0,000000000x% thd audiophile brigade, obviously. Hey, LM4562 is my friend.

tv said:

This to me means nicely distributed sequence of 2nd to 4th harm. Some may argue that this is merely an emulation of "tube" stuff, but hey..

Can be done, most importantly sits in (a digital) mix nicely, and is suprisingly simple to design and build, if you know precisely which mechanisms are to be exploited. Surprisingly low self-noise, also.

Click to expand...

Oh Guru TV, we grovel at your feet 

Please enlighten us unworthy ones.  Can you post some examples?

I joined this forum and diyaudio cos I wanted to build something with the mojo of the JLH Class A amp but without the heat.  Along the way I became a pseudo LTspice guru and did a lot of work on a 1pp zillion OPA to rival Guru Wurcers designs. Still a long way from designing mojo  ;D

Oooo, Master Ricardo, you came to just the right place.

Some time ago, I actually posted some of my ramblings here.

Actually, I recall I breadborded yet another variant which I think was better behaved, but hey, I forgot details. In the meantime I developed a simpler thing that I use now.


http://www.groupdiy.com/index.php?topic=45107


If it blows in your face, you'll know you did something wrong ... so don't blame me, haha.