Electronics > Projects, Designs, and Technical Stuff
OSH Audio THD instrument and live BOM
<< < (5/8) > >>
Kleinstein:
I think it makes absolute sense to use some kind of FFT to look at the signal after the notch. This has a few benefits: one can use it to check the tuning (too much base frequency coming through) and one can differentiate between the harmonics.  So a certain level one could even subtract the levels from the source, if the DUT is broad band and does not alter the harmonics very much.  Looking at it an a FFT way also allows for more tolerance to noise. 

The FFT could also provide another digital notch function, so suppressing residual base frequency. The analog notch would be still needed or at least very useful to keep the demand on the ADC low.

At least for the relatively low frequencies a simple µC base solution might be sufficient.
rhb:
 i do not know of a single DSO with a decent FFT.  I've had experience with MSOX3104T, RTM3104, Instek MSO-2204EA, Rigol DS1102E and LeCroy LC684DLX.  The Instek and the LeCroy are the best of the lot. The Keysight, R&S and Rigol are completely useless.

FWIW the Instek GDS-2000E series has a very nice FFT based SA function.  Instek only supports it on the MDO line, but you can hack the key and run it on a GDS-2072E.  However,  the programmer did not calculate the modulus of a complex number correctly.  So if you sweep a crystal you see a parallel resonance peak, but instead of a notch for the series resonance it's a peak.  Duh!

I strongly recommend just using the DSO to collect the data and then using Octave to do the analysis.
asgard:
Here is a slightly modified version of the track-hold circuit.  Is is configured for a precision peak detector with compensation for the diode reverse leakage.  The MOSFET sets the sample collection period to 8us with an 8us dead time to allow the storage capacitor to completely discharge.  The timing is set with a precision 50% duty cycle pulse train from the 555.  I know it is not a "true" power density measurement, just a peak finder.  But for my purpose it will show in a very simple way how much distorted signal is passing through a DUT, especially if as you say we can accurately calibrate the THD of the oscillators to -90 dB or better.
rhb:
Have you read the Jim Williams paper?  It's starts on page 7

web.mit.edu/6.101/www/s2017/handouts/williams.pdf

His final oscillator design was less than 3 ppm THD.  But the details of how he analyzed the errors should prove helpful.
asgard:
Yes, I read it.  Kinda makes me super smart to make the whole project super modular so if I were to adopt such a type oscillator it would only require a respin of a rather small board assembly.  The only problem is to see about sourcing some of the parts for the specific example oscillator as some are quite probably obsolete.  But other than that it is intriguing.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod