Pocket Oscillator

[mediatechnology]

I changed the oscillator topology to all-inverting Wien using both halves of a 5532.
This also provides a balanced output that, when operating on batteries, is also floating.

Turns out the OptoFET has a distortion mechanism that is not just delta-Rds.
That mechanism is what appears to be providing the higher-order harmonic content.
Some of it is delta-Rds and can be mitigated at higher levels by operating it at high values of Rds.
But, to achieve the lowest distortion, the terminal voltage needs to be relatively low.

I figured out a way, using a second positive feedback loop, to have the FET operate in shunt mode on a 40 dB attenuated version of the output.
(The Wien network is the first positive feedback path. The AGC positive feedback enters the 5532's non-inverting input on pin 5.)
Thus, an output level increase reduces positive feedback and overall loop gain to stabilize the oscillator.
The required delta-gain is very, very small.

Measuring one of the outputs single-ended (unloaded) at +12 dBu I'm reading THD levels of <<0.02%.
The overall balanced level is >+24 dBu with +/-15V supplies.

The active components are a single 5532, an H11F OptoFET, and a sprinkling of diodes.

I'll tweak it some more and post a drawing.

[mediatechnology]

Simple Low Distortion Balanced Output Oscillator

This is all there is to it: A single 5532 and H11F OptoFET.





With +/-9V battery supplies a +20 dBu balanced output is easily reached.
The THD at +18 dBu into 7K32Ω is around 0.0009%.
Something around 0.0003% is A/D.
Although they can add or subtract the final measurement, confirmed with the "Nacho" filter, is that they usually subtract.

The trick is to have 46 dB attenuation applied before having the H11F FET connected in parallel across a 1K resistor.
[2MΩ + (1KΩ||H11F)]/1KΩ||H11F.
The resulting voltage divider output is then applied as positive feedback.
As the shunt resistance decreases so does the gain.
The amount of delta-gain required to sustain oscillation is very small.
The trim pot adjusts the desired output level.
Ideally the oscillator would be run at elevated level and it's output adjusted with a stepped attenuator or fixed pad.
https://ka-electronics.com/shop/index.p ... uct_id=114

The advantages of using an OptoFET are:

1) The detector is a simple FW bridge rectifier connected across the balanced outputs consisting of 4 diodes, 3 resistors and an electrolytic.
2) The H11F is the only through-hole "FET" available at Mouser and is relatively low cost.

I'll post a schematic of the work in progress. I still have some tweaking to do...

[mediatechnology]

A Simple Low Distortion "Pocket Oscillator."

Here is what I have so far. I may do some more tweaking...



The top two op amps are an all-inverting Wien bridge oscillator similar to ULDO-Nacho, Janasek and Vicktor Mickevics.

There is a small amount of positive feed back at the non-inverting input of IC1 that is reduced by the AGC control loop.
This configuration allows the H11F1 to operate at a terminal voltage that is 1/200th of the output to reduce the H11F's distortion contribution.
As the FET's Rds is reduced, the amount of positive feedback is also reduced which in turns lowers stage gain by a tiny fraction.
The output level is detected by a bridge rectifier, smoothed by the 100 uF cap and fed to the H11F's LED thus providing level feedback closing the AGC loop.
The 10K resistors feeding the bridge provide not only a time constant and current-limiting but also isolate the bridge from the output.

Due to diode drops the oscillator wants to run at higher levels where it is both quieter and cleaner.

Though it is not shown an output pad is recommended.
1) It reduces level to 0 dBu or mic-level if needed.
2) It provides resistive isolation for external fault voltages. (e.g. phantom power)
3) It isolates one output being grounded when the load is single-ended.
If a pad is installed the 49R9 resistors need to be made larger...

I used a 5532 for IC1 and ran it off of +/-9V.

The THD is pretty good:

[mediatechnology]

A Simple Low Distortion "Pocket Oscillator."

Here is what I have so far. I may do some more tweaking...



The two op amps are an all-inverting Wien bridge oscillator similar to ULDO-Nacho, Janasek and Vicktor.

There is a small amount of positive feed back at the non-inverting input of IC1 that is reduced by the AGC control loop.
This configuration allows the H11F1 to operate at a terminal voltage that is 1/200th of the output to reduce the H11F's distortion contribution.
As the FET's Rds is reduced, the amount of positive feedback is also reduced which in turns lowers stage gain by a tiny fraction.
The output level is detected by a bridge rectifier, smoothed by the 100 uF cap and fed to the H11F's LED thus providing level feedback closing the AGC loop.
The 10K resistors feeding the bridge provide not only a time constant and current-limiting but also isolate the bridge from the output.

The trim pot is initially adjusted to sustain operation.
Once oscillation is achieved it can be used to adjust level.
It might not make the best front panel control allowing for a wide range of adjustment owing to the fact that if adjusted too far oscillation will stop.
It can however be used in conjunction with the pad to trim level.

Due to diode drops in the AGC loop the oscillator wants to run at higher levels where it is also both quieter and cleaner.
If one did want to run it at lower levels the 10K resistors feeding the bridge could be reduced.

Though it is not shown an output pad is recommended.

1) It reduces level to 0 dBu or mic-level if needed.
2) It provides resistive isolation for external fault voltages. (e.g. phantom power)
3) It isolates one output being grounded when the load is single-ended.
If a pad is installed the 49R9 resistors need to be made larger...

I used a 5532 for IC1 and ran it off of +/-9V.

The THD is pretty good:

With +/-9V battery supplies a +20 dBu balanced output is easily reached.
The THD at +18 dBu into 7K32Ω is around 0.0009%.
Something around 0.0003% is A/D.
Although they can add or subtract the final measurement, confirmed with the "Nacho" filter, is that they usually subtract.

Thinking about making a board for this if enough people are interested.

I need to see how it compares to using a small "jump drive" USB-based player.

[AnalogJoe]

Very nice!!!

[mediatechnology]

Very nice!!!

Thanks!

I just had to see if I could do something as good or better than one using an unavailable thermistor.
I realize Ian in his original post is probably looking to retrofit one he has to use a thermistor.

I may build one for myself to have a battery-powered low distortion source of balanced output tone that floats from ground.
I'm thinking a small box with two TRS outputs. One line level and one mic level. Maybe add a third RCA at -10 dBV.

The only floating mic-level tone I have is Shure A15TG and it's distortion is relatively high.
I do have a Cheng Instruments battery-powered oscillator but it's unbalanced.

A second project using this oscillator is a pair adjusted for 19 and 20 kHz that are resistively-summed.

[ruffrecords]

Very nice work. As you correctly guessed what I am really looking for is a simple retrofit solution for my existing oscillator but given the difficulty of finding a suitable thermistor this may be a pipe dream anyway.

One minor point. My version works from a single PP3 battery. Given the implied elevated operating level of your device what battery option would be appropriate?

Cheers

Ian

[mediatechnology]

Thank you Ian.
I was thinking two 9V batteries.

I added pads to the output schematic and prototype.
The pad loads the output in 732Ω. (Differential load.)

At +12 dBu driving the pad I'm measuring 0.0015% at 1 kHz.

I also tested 1nF for 10 kHz and 100 nF for 100 Hz and the THD is around 0.0025%.

The design doesn't lend itself to frequency switching unless a more sophisticated detector is used.
The levels from capacitor tolerance vary too much requiring re-trimming.

I also tested variation among H11F optos - 4 of the same lot.
The level varied from -0.45 dB to +2.5 dB compared to reference OptoFET.
The variation is easily calibrated out by the trimmer.
I have a few more coming from Mouser today from a contemporary production run. Mine are from the 1990s...

[mediatechnology]

I got some modern ON Semiconductor H11F1 today and had a chance to try a couple.

They seem to have a higher transfer ratio: These were 8.5 dB lower than the reference NOS Harris H11F1.
Adjustment of the gain trim brought it back into range.
In the 100 Hz oscillator distortion was almost double but still acceptable.
Raising the positive feedback resistor from 2MΩ to 4MΩ lowered it the same range as the reference H11F1.
(I corrected the schematic which showed the shunt resistor to be 1KΩ. It's actually 10KΩ.)
Using 4M/10K the positive feedback is attenuated by 52 dB.
The newer ON devices are more sensitive but just as good.

I did notice that the 100 Hz HD3 was higher than the 1 KHz and remembered that I needed to increase the smoothing cap across the LED from 100 uF to 200 uF. These are the results with the ON H11F1 at 100 Hz:



The 100 Hz THD is 0.0014% with the oscillator driving 732Ω at +12 dBu.
The spurious free dynamic range, "SFDR," is -100 dBc.



BTW and FWIW the copyright on the schematic are for the artwork and not the design. Feel free to use it: Attribution would be appreciated.

[mediatechnology]

I reset the oscillator frequency back to 1 kHz and increased the positive feedback attenuator to -60 dB using a 10MΩ/10K divider. Oscillator stability was not affected and the THD dropped to essentially that of the A/D which is around 0.0003%. I'll need to use the notch filter and dig deeper.

The operating current is about 14 mA which should provide 24-36 hours on two 9V batteries.
5-6 mA of that is current in the line-level -12 dBu pad.

With the positive feedback attenuation increased to 60 dB I could trim it down to 0 dBu (direct, no pad) but adjustment was a little touchy.
The 10K resistors connected to the bridge could be lowered with a slight increase in THD.

At the moment I'm running it on two "lightly used" 9V. One was around 8.6 the other 8.8V. I wanted slightly spent batteries with a higher internal resistance to see if it affected THD.
I'll leave it on batteries overnight.

This is the THD running on batteries. It measures 0.0004%



Having both a U-pad output and a "double L" output provides some flexibility for termination by active and transformer line inputs as well as mic-level terminations.