A Battery-Powered Low Distortion "Pocket" Audio Oscillator With Auto Off

[mediatechnology]

The prototype run of "pocket" oscillator PC boards arrived and runs without any mods.

I've tested basic functionality and am waiting on the timer to turn it off. I should have about another 45 minutes left on the meter.

[Tubetec]

Impressive turn around time from conception to prototype to finished pcb .
It beats other designs in terms of simplicity , functionality and component count .

I think it deserves publishing in print , heres a list of the top 30 electronics publications , https://magazine.feedspot.com/electronics_magazines/
Elector or Nuts n Volts I think would snap this up in a second ,
They might recieve an anonymous tip off about it

[AnalogJoe]

Impressive turn around time from conception to prototype to finished pcb .
It beats other designs in terms of simplicity , functionality and component count .

I think it deserves publishing in print , heres a list of the top 30 electronics publications , https://magazine.feedspot.com/electronics_magazines/
Elector or Nuts n Volts I think would snap this up in a second ,
They might recieve an anonymous tip off about it

Why not try JAES? They might be willing to publish it, and it is a indexed peer-reviewed journal.

[mediatechnology]

I've had a chance to run quite a few more tests.

THD is low, around 0.0002% with a JRC 5532.
TI 5532s are a little higher at 0.0005% with more HD2.

The Off timer works fine.
As it turns off level fades eventually into distortion for a couple of minutes but by that time, if the restart button hasn't been pressed, one has probably forgotten about it and has gone to lunch. The batteries get to live another day without it running endlessly.
For continuous tone just turn it on.

Start-up time for oscillation is fast with little under/overshoot.

The areas I've explored are level stability, time to achieve final level and thermal effects.

One of the tests I ran was to learn where the operating point of the H11F1 OPTOFET is when stabilized.

LED Vf measures 1.17V in circuit when the oscillator is at operating level.
I removed the H11F1 and measured it out of circuit at the same Vf.

With Vf at 1.17V Iforward is 107 µA.
Ids on the FET side is 5.27 µA.
Rds, measured indirectly, is 2.14 MΩ.

When the H11F1 is operating in the oscillator this means the FET is attenuating positive feed back by about 0.04 dB to stabilize level.

After a "warm-up" period oscillator level stability at room temperature is about +/- 0.1 dB.

The time to achieve final level from a prolonged "cold" Off state is fairly long:

t=10s -1.4dB
t=30s -1.23 dB
t=1m -1.04 dB
t=2m -0.56 dB
t=3m -0.24 dB
t=4m 0 dB

When the oscillator is switched off by the timer, and immediately restarted, level recovery is almost instantaneous.

To investigate cold start-up time the first thing I looked at was the AGC capacitor, C7, which was 100 µF with no discharge resistor.

I first tried 22 µF which shortened it a little bit but not much.
With 22 µF distortion rose from 0.0002% to 0.001% which is still respectable.

I added a discharge resistor which did not reduce turn-on time.

The AGC cap discharge path (C7) is through the OPTOFET LED.
Once the voltage across the capacitor drops below the LED's Vf, forward conduction is reduced.
C7 has no discharge path - only leakage paths through the LED and bridge diodes.

Adding a discharge resistor did not help.

As it turns out C7, due to it's value and dielectric absorption (DA), stores a portion of it's last value, below the Vf of the LED, for a really long time.
Temporarily discharging it through a short will reduce its voltage for a period of time, but it will eventually grow back to about 850 mV due to DA.
Allowing it to maintain its charge actually reduces turn-on time.
I eventually replaced the 22 uF with 100 uF to lower distortion while I looked elsewhere.

There is almost no supply voltage sensitivity down to around 13V.
That rules out the MOSFET power switch and other things.

I decided to look at thermal effects thinking this might truly be a "warm-up" effect.

The self-heating of the OPTOFET LED is small: The power is about 125 µW.

What about the tempco FET side?
The self-heating is also small.
"Shocking" the H11F1 with freeze mist only reduced the level by about 2 dB.
I ruled out the H11F1.

I then hit the 5532 with freeze mist and the level dropped by more than 8 dB.
If I had sprayed it much longer the oscillator would have stopped.
The 5532's gain decreases with temperature.

"Warmed up" 5532s have more open loop gain.
The 5532 in this circuit dissipates around 200 mW including the pad, feedback loads and quiescent power.

I can hear the rumor now: "Hot op amps sound better than cold ones."
Seriously though could it be that 5532/34 with class-A load resistors simply heat the part more and decrease distortion by providing more open loop gain?
It's something to think about while you mow the lawn or paint.
Or, it could become a highly-debated argument to be found on DIY Audio, GearSpace or Group DIY.

So the warm-up effect appears to be the 5532's gain increasing as it heats up.
Remember that it only takes a 0.04 dB reduction in gain to reduce oscillation from the clipping point to being stabilized.

The AGC in the oscillator is an above threshold limiter with a finite slope.

Varying the gain adjusts the level by pushing the limiter further up its' slope.
This is true of the original pocket oscillator using a thermistor or those using diode limiters.
I suspect that they all have a warm-up period.
Those using TL07X probably have less, but they have lower gain to begin with.

ULDO-Nacho and other line-powered oscillators compare the rectified oscillator level to a fixed reference to provide AGC error correction.
These oscillators reach final level in seconds, not minutes.
ULDO-Nacho regulates its output level to around a couple of hundreths of a dB.

For my purposes I'll trade extended battery life and floating output for having to wait a few minutes to reach that final 0.1 dB of level.
Turn it on, let it warm up and then setup the measurement.
To know if you have signal at all just turn it on and go. Tone will be there.

The only change I've made thus far is to reduce the Zener from 20 to 18V.
I made this change for the benefit of the MOSFETS and to reduce the leakage in the 25V-rated timing capacitor.

[AnalogJoe]

Regarding the Hot VS Cold opamp.

Running a 5532 with pull-up/down resistors does make it class A and that by itself should reduce distortion and increase dissipation. The problem with class AB is that, even though it reduces crossover distortion, you still have the problem of gm doubling, which produces non-linear gain variations when the transistors enter and leave the class A region.

A Hot opamp by itself could indeed produce less distortion as VBE or VGS, and other parameters like IS, etc..., are very sensitive to temperature variations. We can assume that the 5532 designers took into account temperature for their design, so the specs are probably stated when the opamp has reached its working temperature. Increasing the temperature will also change the operating point of the output stage, driving it closer to class A. I guess that if you heat the opamp up to a certain point, the output stage could be operating into very deep Class AB or even Class A, and that could reduce distortion.

[mediatechnology]

I just had a quick look at the TI 5532 datasheet and the minimum DC Av at 25 deg C ambient is 15V/1mV.
Over temperature, 0-70 deg C, it can be as low as 10V/1mV or about 3.5 dB lower.
The 25 degree C minimum DC open loop gain is 83.5 dB.
The typical 25 deg C value is about 94 dB.

[Tubetec]

4 minutes to stabilise sounds like a very favourable result considering a decent bench multimeter might require a hour before its ready to self calibrate. One of the benefits of tube mics is they run warm so very much reducing the chance of condensation occuring around the capsule , the sound also tends to change during the warmup period so at least an hours warm before trying to record is recomended .
Large consoles tend to need even more time with all the metal work to bring up to working temp , powering up the night before a recording session is a good practice .

I just caught up with progress from the E1DA Cosmos site lately , now Ivan recomends performing tests at 768khz as the ADC 's inherent distortion reduces due to self heating . Previously he did offer a miniature programable add on heater module which sat on top of the ADC chip and found at around 50C the situation improved . Elevated temps might also help reduce any variations due to ambient conditions .

As A.J. seems to suggest it might be worthwhile to attempt heating the 5532 by some external means just to see what happens , maybe even thermally insulating the IC could have a desirable effect without detriment to battery life .

So could we say a mic thats below ambient and accumulating moisture on the capsule because of condenseation sounds worse than one thats above ambient temp and not subject to this unwanted effect , yes ,almost certainly .
A large console that hasnt warmed sufficiently where the sound is changing over time sounds worse , that could be a matter of subjective opinion depending on various factors like source ,but at least once its had a chance to warm to opperating temp the rate of change is much reduced and the sound stays where we put it .
Does reducing the distortion on a sound source automatically mean it sounds better , no ,not nessesarily .
Is there pernicious ,pernickety individuals out there who will cause an argument rather than making any meaningfull contribution to the discussion ,absolutely yes

[AnalogJoe]

4 minutes to stabilise sounds like a very favourable result considering a decent bench multimeter might require a hour before its ready to self calibrate. One of the benefits of tube mics is they run warm so very much reducing the chance of condensation occuring around the capsule , the sound also tends to change during the warmup period so at least an hours warm before trying to record is recomended .
Large consoles tend to need even more time with all the metal work to bring up to working temp , powering up the night before a recording session is a good practice .

I just caught up with progress from the E1DA Cosmos site lately , now Ivan recomends performing tests at 768khz as the ADC 's inherent distortion reduces . Previously he did offer a miniature programable add on heater module which sat on top of the ADC chip and found at around 50C the situation improved . Elevated temps might also help reduce any variations due to ambient conditions .

As A.J. seems to suggest it might be worthwhile to attempt heating the 5532 by some external means just to see what happens , maybe even thermally insulating the IC could have a desirable effect without detriment to battery life .

A 4 minute settling time is indeed quite good. I've seen function generators that require up to 15 min or even half an hour to lower the distortion level to its rated spec. So a 4 minute settling time with the THD figures you achieved is quite remarkable.

I also forgot to mention, changing the temperature will also change the operating currents of the stages and their transconductance (also their output resistance) which may increase the overall open-loop gain. So, if the output stage goes closer to pure class A, and the open-loop gain increases, you have a recipe for less distortion when you close the feedback loop. Perhaps if the pull up/down resistors heat the opamp enough, it might be the best solition.

The MCI JH-500 is notorious for running extremely hot. Working on it without AC is practically unbearable. That console was full of 5534s running at +/- 32 V using bootstrap transistors. 5534s are very resilient and can handle the heat, the metal can version used a heatsink but the PDIP didn't. I guess that reduces its lifetime but many are still functioning perfectly after 40+ years. The main issue with increased operating temperature is an increase in noise, so pick your poison. Also, after years of operating at high temperatures the 5534 may become noisier than usual and will require replacement.

Try measuring THD+N to get a more significant measure between distortion and noise. It will probably increase as you heat up the opamps

[Tubetec]

Made an addition to my previous post above

[mediatechnology]

The MCI JH-500 is notorious for running extremely hot. Working on it without AC is practically unbearable. That console was full of 5534s running at +/- 32 V using bootstrap transistors. 5534s are very resilient and can handle the heat, the metal can version used a heatsink but the PDIP didn't. I guess that reduces its lifetime but many are still functioning perfectly after 40+ years. The main issue with increased operating temperature is an increase in noise, so pick your poison. Also, after years of operating at high temperatures the 5534 may become noisier than usual and will require replacement.

I am a follower of the "Swinging Op Amp."

viewtopic.php?t=699
https://proaudiodesignforum.com/forum/p ... .php?t=927

What doesn't handle the heat well in the 500-series are the electrolytic caps near those 5534s.
I've changed those by the thousands.