|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
In Reply to: Re: Isn't Slew Rate a bit like THD? posted by john curl on July 22, 2003 at 21:36:24:
Hi,>Where is your dominant pole?
1.000Mhz
>What about non-linear capacitance?Its being modeled. Not sure what you mean.
Follow Ups:
Mfc, I think that there is some confusion as to what TIM is. It can be related to harmonic distortion, BUT harmonic distortion can also be generated by a non-linear transfer function, including crossover distortion and voltage sensitive capacitance built into every bipolar transistor and FET. Graduate study of nonlinearity goes into what happens when you operate solid state devices above 100KHz or so. Under these conditions, the nonlinear capacitances between the base-emitter-collector, for example, are dominant sources of distortion. Spice was invented at UCB at the time I took courses on non-linear effects like this, by the same professors who taught it to me, longhand. The Spice model will need to incorporate these effects, in order to be accurate at higher frequencies. Before Spice, we used ECAP, developed by IBM. It would not simulate non-linear caps, and was therefore much more limited in assessing distortion in active circuits. I am not an expert in Spice modeling, but I am reasonably sure that it is generically accurate, at least.
Now back to TIM. TIM is an op-amp dominated kind of distortion. To get significant TIM, you have to have a dominant pole within a feedback loop that is significantly lower than the working bandwidth of the system. Therefore, if you have an open loop bandwidth (dominant pole) of 1MHz, then you won't have any TIM distortion below 1MHz. However, you can have plenty of harmonic distortion that is frequency sensitive, due to non-linear cap effects. Is this making sense to you? This is off the top, so there may be some missing pieces that I might need to fill in. Keep up on doing your measurements. I'm glad to see some folks measuring or modeling something around here ;-)
like this?
I was wondering when you might come around here...How's it going?
Good reference. I think Spice does a good job on
nonlinear capacitance, as long as the model params
are well captured.
hi mike,doin OK, trying to write a paper on a high feedback technique, so i
shouldn't be spending as much time as i do on these forumsTrying to communicate engineering concepts without local graphics storage seems a major handicap in your thread and this forum
buffers with simple topologies are hard to compare when the feedback is affected by the bilateral nature of the devices - input impedance has a quite noticeable effect on emitter follower/Darlington/Szilkai performance
just downloaded the Linear Technology SwCad III to post sims over on DIYaudio.com - i'm quite impressed after < 1 week
Trying to communicate engineering concepts without local graphics storage seems a major handicap in your thread and this forumActually there is local storage. Though it's a bit trickier to display the graphics in messages compared to how it's done on say diyAudio.com.
Just start your own Inmate Picture Gallery and then you can link to the uploaded images in your posts.
se
Thanks!
here it is
When I try to access your picture gallery (as well as several others), I get a bandwidth exceeded message. Is this image a really big one size-wise or something?se
and all I get is the yo-yo now???
I still get a yo-yo and I still get the "User bandwidth has been exceeded Please try again tomorrow" message when I try and view your gallery pictures directly. Same as with some other users' gallery pictures. But on others, the pictures display just fine.Don't know what the problem is. Have to ask Rod.
se
Hmmmm... All I see is an Audio Asylum yo-yo. Yet the URL's pointing to Sziklaicomp10ohm_.2uFload8v1050square_1v5Khzsine.JPG.se
mfc, I agree that SE is a real resource in helping us to put difficult equations and pictures on this website. I had no idea that it could be done.
"To get significant TIM, you have to have a dominant pole within a feedback loop that is significantly lower than the working bandwidth of the system. Therefore, if you have an open loop bandwidth (dominant pole) of 1MHz, then you won't have any TIM distortion below 1MHz. However, you can have plenty of harmonic distortion that is frequency sensitive, due to non-linear cap effects."You are right. OTOH, if the non-linear C effects are built into the Spice model, the distortion could also be due to phase modulation (PIMD). Someone will have to take a closer look at that model. If, however, the model does not include non-linear C effects, then we have to assume that there is some form of slew-rate limiting (AIMD).
Hi,> Therefore, if you have an open loop bandwidth (dominant pole) of
> 1MHz, then you won't have any TIM distortion below 1MHz. However,
> you can have plenty of harmonic distortion that is frequency
> sensitive, due to non-linear cap effects. Is this making sense to
> you?I thought part of TIM's cause was due to non-linear cap effects...
OK. I could be way off on attributing the effect of the square
wave to TIM. In the papers I've read so far, I haven't come across
anything about TIM not appearing below the dominant pole. If you
know of a reference, that would be well appreciated.Back to non-linear cap effects...this is one difference between
a simple emitter follower and a Sziklai. The Sziklai is more
sensitive to this effect and produces more distortion (see sims
in transient test thread). I had to increase the peak voltage on
the Square wave from 4 to 8 volts in order to see it while leaving
the sine wave at 1v peak.> TIM is an op-amp dominated kind of distortion
Since amps don't have these low frequency open loop dominant poles,
and opamps do, then it sounds like TIM isn't much of a problem to an
amp?
"In the papers I've read so far, I haven't come across anything about TIM not appearing below the dominant pole. If you know of a reference, that would be well appreciated."For this I would recommend Jung's overview of SID and TIM (Audio, Jun-Aug 1979) or Cordell, "Another View of TIM" (Audio, Feb-Mar 1980). Cordell stresses that for a given size comp C an input stage with a higher output current capability will provide an increased slew rate.
"Since amps don't have these low frequency open loop dominant poles,
and opamps do, then it sounds like TIM isn't much of a problem to an
amp?"It was before the mechanism was well-understood. Several authors (esp Garde and Leach) showed how to set up a first stage to ensure minimal TIM even for high feedback factors.
I did not mean to be unclear. If you are ONLY talking about an output stage, NOT the whole amp, then TIM is not a significant contributor. Many power amps and preamps are based on the OP AMP MODEL. That is: high gain, single pole dominant equization, low open loop bandwidth, modest slew-rate, etc. For clarity, you might want to read Jung's AES preprint from 1977 on SID. I can send it to you if you send me an E-mail.
This post is made possible by the generous support of people like you and our sponsors: