Low-distortion Audio-range Oscillator

[dimitri]

MPQ6842 is not easy to get:
MPQ6842 | EDX Electronics
Fixed-frequency ultra-low disto oscillator looks ok for this application.

[rsavas]

I looked inside my Amber unit, it is a Bourns 4-gang pot, did not read the #'s but I think it is a 81/82 series pot, as found on the web site. Must be special order, or find 4 of them and stick some single units together as shown in the data sheet, good luck.
Hope I never have to fix that part.
The MPQ6542, near impossible to get, can be sub'd with I think DDMT5401/5551 duals, or some others, not sure if matching NPN & PNP's pairs is an issue.
Yes Victors fixed freq oscillator is the thing to make or buy from him. A few discrete frequencies is fine. It is a very good deal!! All good stuff to read.

Rick

[richiem]

Take a look at the Heath IG-18 I modded to Cordell's SV oscillator here: IG-18 #2, the BIG-18. The results are pretty good overall.

[koja]

I was not aware of this thread, only saw it now...

I also have researched for a while what the best oscillator may be to use. Some people here mentioned (now late) Jim Williams of Linear Systems but they looked too far back at his application notes. Look at AN132 yr2011 ( http://cds.linear.com/docs/Applicati...ote/AN132f.pdf ) which originated with prof.Hill affiliated with Harvard and then was improved by Jim. I comunicated with the other two authors and gathered enough info to change dc offset and the frequency. I am only interested in one (around 2kHz), maybe two discrete frequency settings (2k and 10k) to take a measurement at.

I also have Gerber files for the pcbs. Only the parts are not cheap because they insist on Os-Con caps (brand name by sanyo), which I confirmed through private correspondence. I did not yet check into the availability of all the values since this is a lower priority project for me. If there were enough others to bring the pcb cost down maybe i would move on it quicker. Interested?

I do not believe anything else out there will beat this at 120dB signal S/N ratio.

[coluke]

Originally Posted by koja

I do not believe anything else out there will beat this at 120dB signal S/N ratio.

Do you mean THD+N? It seems that Williams' generator doesn't even get close to that figure, although -105 dB is certainly a respectable figure @ ~1.5 Vrms out. I think a state variable topology working at higher output may be better in this respect.

L.

[trioflex]

Hi

we are using the 1KHz oscillator to test our AD7760 based function blocks, and we are seeing some problems, they may be related to the DC block capacitor but we are not sure yet. We made first measurements with 47uF MLCC capacitor (ouch !!! our bad of course) we got somewhat below 8ppm THD, 101 db SNR, with 3rd harmonic being dominating.

Replacing the DC block with some 0.33uF large ugly blue 1000V rated capacitor of unknown technology and origin, we got 5.3 ppm THD, 105 SNR, but we completely lost 3rd harmonics? At least below 145dB !!



FFT attached, 1M point FFT, AD7760 with 625KSPS sampling rate default digital filters used (cut off 250KHz).

Can I assume safely that the 3rd harmonic what we did see with MLCC capacitor was indeed the capacitor non-linearity related spur?

In that case the 2nd harmonics that dominate with the other capacitor is most likely not any more at all related to the capacitor but some design problem with our ADC?

We do supply the 1KHz generator from four 9V batteries, output level is set to the maximum.

Any ideas how to get the remaining harmonics out? We will try more capacitors, but I bet those will not change the picture any more.

/Antti

[1audio]

1) your ADC will never be perfect.
2) Ceramic caps will limit the performance a lot.
3) only a few types of film caps will be linear enough to not be an issue, Teflon, polystyrene and polypropylene are commonly available.
4) The AGC is the bigger issue in distortion at this level. The Silonex parts are good but CDSE is non-linear with a high voltage coefficient. Keep the voltage across it as small as possible.
5) Some opamps are potentially lower in distortion. It seems hard to predict and you will need to just measure the results.
6) Use a twin T notch to see what the source distortion is. I get well below -130 for all the harmonics with the Shibasoku generator so I know its possible.
7) When you get all the pieces working better you will see more things to make better. There is no end to this.
8) Victors oscillator is more than good enough for almost any kind of testing and costs so little for what it is.

[rsavas]

Take a look at the Heath IG-18 I modded to Cordell's SV oscillator here: IG-18 #2, the BIG-18. The results are pretty good overall.

This get my vote!! Nice implementation. Forget the Amber unit, although you could borrow from some of the circuits it contains.

I wonder how using standard SMT passive parts, although good ones (only NPO & X7R), would affect the performance. I read that Victor changed from standard Yageo MF to some exotic Vishay parts & it made a difference?
Nice web site Mr. Moore, more to read and learn!! Now back to programming for my portable stereo unit!!

Cheers
Rick

[dirkwright]

I got the PCM4222EVM board yesterday. Just as I suspected, it was never intended to be put into a box. The board has rubber feet on it. One part at a time.... sigh.

I'm still waiting on the hi zoot 9V batteries and charger, and the parts for the CCS for Victor's oscillator. It's like watching grass grow.

[trioflex]

Originally Posted by 1audio

1) your ADC will never be perfect.
2) Ceramic caps will limit the performance a lot.
3) only a few types of film caps will be linear enough to not be an issue, Teflon, polystyrene and polypropylene are commonly available.
4) The AGC is the bigger issue in distortion at this level. The Silonex parts are good but CDSE is non-linear with a high voltage coefficient. Keep the voltage across it as small as possible.
5) Some opamps are potentially lower in distortion. It seems hard to predict and you will need to just measure the results.
6) Use a twin T notch to see what the source distortion is. I get well below -130 for all the harmonics with the Shibasoku generator so I know its possible.
7) When you get all the pieces working better you will see more things to make better. There is no end to this.
8) Victors oscillator is more than good enough for almost any kind of testing and costs so little for what it is.

I am using Victors generator.. that is why I am wondering.. I am measuring Victors oscillator directly to the ADC, well, there is AD4940 before the ADC doing single ended to differential conversion..