Great find! How can this be done in hardware/software? DSP?
Would Audio-Precision be interested as they have implimented multi-tone already?
Last edited:
I think Rich Cabot read that at the time
The makewave utility lets you specify the crest factor you want and then adjusts all levels to avoid clipping to the max output level you spec.
Edit: at AP.com download tech note TN-18.exe (self extracting zip).
jan
Last edited:
Well lets discuss which one we can all do that works well and start using it on a regular basis. This thd/harmonic, single tone test isn't all of what we are capable of doing and wont help in determining better opamp designs. JC just called and suggests at least a triple tone IM test... better than 2 tone IM. But, lets slug it out here and start using IT.
Which can we all do that doesnt cost the earth and tells us a lot more towards the real application of the opamp - multi-tone signal processing/amplifying. Thx - Dick
Which can we all do that doesnt cost the earth and tells us a lot more towards the real application of the opamp - multi-tone signal processing/amplifying. Thx - Dick
It is always the same non-linearity. Interpretation of results (Cabot, e.g.). One can take a twin-tone and move it through audio band. Both IM and HD products are visible then.
I would rather suggest a kind of test that is not limited to audio frequencies only. Like audio band sine + AM modulated or FM modulated HF/VHF carrier. This to be analyzed by FFT analysis in audio band and let's see what products were created.
I would rather suggest a kind of test that is not limited to audio frequencies only. Like audio band sine + AM modulated or FM modulated HF/VHF carrier. This to be analyzed by FFT analysis in audio band and let's see what products were created.
You can use Excel and SoX to directly create .wav files at any resolution and sample rate supported.
SoX. Didn’t know about it. Impressive list of features.
SoX - Sound eXchange | Features
Thanks Scott
Joachim I read the description of the test. The method seems to me very powerful. Congratulations.
George
Once you stop the nose bleeding…
If the very good (stiff and clean) psu is accompanied with some measures for rail isolation/RC filtering btn active stages (local post regulation is another option), PSRR requirements I think can be relaxed.
It remains the problem of rail coupling btn input & output on each stage (e.g. base-collector). RC filter there too.
What else left then to PSReject?
George
If the very good (stiff and clean) psu is accompanied with some measures for rail isolation/RC filtering btn active stages (local post regulation is another option), PSRR requirements I think can be relaxed.
It remains the problem of rail coupling btn input & output on each stage (e.g. base-collector). RC filter there too.
What else left then to PSReject?
George
Last edited:
Bear, I'm not sure I get the question.
As I see it, PSRR always matter because it determines how much of the psu junk makes it to the signal output.
If you have a maximum 'bleed through' goal, like you don't want more than say 1uV of supply junk at the signal output, you need a certain combination of actual power supply junk and PSRR.
Improve your psu regulation and you can relax the PSRR requirement for your goal, but it always will matter.
Probably when you make your goal with a specific supply and PSRR you will want to try for an even lower goal...
jan
Some of us like to also confine currents to loops as local to a given circuit section as possible. Walt Jung discusses this in his recent interview in audioXpress, and Bruno Putzeys mentions it somewhere (possibly on the Grimm Audio site?).
When I designed the electronics for the photodiode array spectrometer in the mid-late 70's, I grappled with circuits that needed to be both fast and precise. There was often some relatively local regulation, but the main power supplies were separate from the primary instrument, and although well-regulated, low noise, and fast response, if the circuitry had substantial variation in currents in operation the system would not have worked well enough.
So in virtually every part, after local R-C decoupling, local shunt regulation acted to produce a system overall with invariant power supply currents. In some cases the shunt conduction was actually driven by the signals, not regulating after the fact. It worked very well indeed, and allowed the boards to share a heavy copper ground plane; each board was attached by a copper angle bracket to the plane and was itself constructed of two-sided copper-clad boards, all hand-wired point-to-point.
