It was a while since I looked at the exact circuit.
But my intuitive ansver to your question is that your phenomen lies in the change of resistors. The resistors in the RC net will also set the gain of the OP-AMP circuit.
The maximum Q achievable without tuning should not be a first-order function of feedback network impedance level (it shurely depends on opamp o/l gain though). I quickly simulated this and was not able to repeat your findings; peak gain is in both cases a touch below 70 dB. Are you sure that you're not seeing simulation artefacts? As the peak is extremely narrow, I suggest you run the AC simulation with 10k points/octave. If this does not clarify the case I think a schematic of the exact setup might help.
Samuel
Samuel
Hi Samuel,
The effect doesn't show up in simulation as well as other effects. I thinking the op amp models are too close to ideal.
I did this with a real world circuit using to different methods. One method was using a Hp 339A by connecting the oscillator to the input of the filter and monitoring the output on the meter in level mode.The level has to kept quite low, 10-30mVrms as there is a lot of gain at Fr.
The other method is to use a sound card and ARTA fft software. Input a low level PN pink noise, (periodic pink noise), while running the software in single channel frequency response mode. Same setup.
The first is full proof while the second is subject to software maximum resolution.
I can post a schematic if necessary.
David.
This is a preview of the schematics as built.
Probably not bug free, so build at own risk.
Patrick
Distortion magnifier version 2 public
Is this circuit the same as Bob Cordell's DM? I dont find a detailed schematic of Bobs design. I can only find links to the whole vol 0 of linear audio for $30.
Alfredo
Hi avincenty,
If you're thinking of building the DM, it is well worth it to pay for a copy of the LA in which it appeared, as there is a lot of associated information there that will help with understanding its operation, how to use it, and helpful if debug is necessary. Contact Jan Didden at www.linearaudio.net - home. There is also a kit for it available, well worth the money if your time can be better spent on something else (check with Jan re the kit). There is also some information about the DM on my website at CordellAudio.com - Home.
Cheers,
Bob
I've been studying your article in LA and I think I'd like to do some things differently. I need some information though. For the phase adjustment, how many degrees (at which frequency) of range is required for most DUT's? I'm using an all pass filter for the phase adjustment, but I don't know exactly how much range I need to include. Any help is appreciated.
For the all pass filter, I set it up to have a range of about 1 deg of phase lag @ 100kHz to about 12 deg of lag. When I look at the Bode plots for various amplifiers I've designed in simulation, the phase lag is not much, maybe 1 or 2 degrees @ 100kHz. I expect the real world to be similar.
I disagree because a low pass filter will attenuate the harmonics from the oscillator, thus changing the results of the residual distortion output from the DM. The change will result in a reading of more higher harmonics in the spectrum than actually exist from the distortion of the DUT.
Not "equally". Really? You believe that's true for all DUT's? In my limited experience, it's obvious that the LP function of a DUT can be almost any order (first, second, or something else) and it's pointless to try to emulate that in the source channel from the oscillator. It seems to me that it would create just as much error to assume that the DUT behaves as a first order LPF as it would to just ignore it by not using a LPF on the source channel. I do not have any evidence yet to support my theory however. It's just an assumption on my part.
Last edited:
Of course you're absolutely right that it is dangerous to assume anything about the exact low-pass characteristics of the DUT. But we have to assume something, or else we can't put anything to correct it in the subtraction path. That the DUT has all-pass characteristics seems pretty unlikely; for a casual power amp, a first-order model is probably reasonably accurate.
But we may just be splitting hairs here; most DUTs worth throwing a distortion magnifier at will have a bandwidth of at least several hundred kHz. So even slightly above the audio frequency range, the important 2nd and 3rd harmonics will not suffer much from modest errors in the low-pass characteristics. And higher harmonics should, with a decent source, be relatively low to begin with. Last but not least, measuring distortion is never a 1% measurement, more likely 10% or worse below -120 dB. Resolution counts, not accuracy (within limits, of course!).
A more practical reason to stay with the first-order low-pass is that it is easily implemented without adding another opamp, with its distortion contribution. Or did you find a way where the opamp is out of question?
Samuel
But we may just be splitting hairs here; most DUTs worth throwing a distortion magnifier at will have a bandwidth of at least several hundred kHz. So even slightly above the audio frequency range, the important 2nd and 3rd harmonics will not suffer much from modest errors in the low-pass characteristics. And higher harmonics should, with a decent source, be relatively low to begin with. Last but not least, measuring distortion is never a 1% measurement, more likely 10% or worse below -120 dB. Resolution counts, not accuracy (within limits, of course!).
A more practical reason to stay with the first-order low-pass is that it is easily implemented without adding another opamp, with its distortion contribution. Or did you find a way where the opamp is out of question?
Samuel
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.