Pocket Oscillator

[ruffrecords]

I chatted with Gemini about a modern replacement of the RA43. After all the motherhood saying I should look for one with similar specs and listing the key specs I asked it to recommend a specific part based on that spec and it came up with the EPCOS / TDK B57862L0202F000 which seems pretty close. It is available at Farnell/Mouser etc and only costs £2.30.

Cheers

Ian

It might require higher signal power to work: The TDK has a dissipation factor of 1.4 mW/°K versus 12.5µW/°C for the RA53.
The TDK is about 100 times less sensitive to self-heating.
But in today's world that might be the best you can get.

Yes, you are right, self heating is the key and although the TDK is a tiny thing it is a lot less sensitive to self heating.

I need to keep looking.

Cheers

Ian

[Tubetec]

The Shibaura type PSB-S9 has 0.15mW/C and a responce time of 0.6 seconds , and is the worlds most compact mass produced thermistor .
Its designed for catheters , ouch

[mediatechnology]

The other small beads I looked at are higher, such as the Amphenol/Thermometrics P20, are 140µW/°C.
The CTS are 100µW/°C with a fast 1s response time.

I posted the links to these earlier.

Here's the CTS: https://www.ctscorp.com/Files/DataSheet ... asheet.pdf

The CTS is spec'd it at "0.1 Mw/°C" but what they really mean is 0.1 mW LOL. (Or 100 µW/°C)

The two things that give the RA53 it's sensitivity are the small chip size and operating in a vacuum.
At 1 mA it's definitely in "glow mode."

[mediatechnology]

Ian -

You might want to think about using an opto-isolated bidirectional FET voltage-controlled-resistor ("VCR") instead of the thermistor:
https://www.onsemi.com/download/data-sh ... 1f3m-d.pdf

The FET side of the opto's control range could be degenerated using a smaller parallel/series resistance that would reduce the FET's distortion contribution.
You only need a fraction of dB's worth of FET delta-R to regulate oscillation.
I use a highly-degenerated VCA in an oscillator here: https://proaudiodesignforum.com/forum/p ... php?t=1265

The LED side of the opto could operate at low current.
Place the LED and filter cap inside of a FW bridge with a series resistor feeding the bridge connected to the output. The distortion contribution of the bridge would be minimal due to the isolation resistor.

I can do a sketch...

I've always been fascinated by the H11F. It would make cool Allpass/Phaser.

[flyboy71]

I have some H11F1M optos. I was messing with them as a variable gain attenuator as in the datasheet but given the response curve they definately need some type of feedback to control. Ive also found them to be very non linear in terms of distortion as you stated when varying the resistance of the FET because they arent tolerant of high signal levels. It would be better if they brought the gate out for bias or provided a matched pair of FETs for using one as feedback. A bunch of folks on other forums had the same issues. I gave up on them and went to some NSL32 optos for any variable gain applications. Id still like to find a use for them.

There is an app note on making the H11F1M optos linear. I havent tried it yet either. https://www.edn.com/use-a-photoelectric ... ntiometer/

[mediatechnology]

Thanks for that EDN article.

Since the Opto FET will operate over a very narrow control range and also has negative feedback control non-linearity may not matter much.

What I need to learn is can it be degenerated enough (if necessary) so the FET terminal voltage remains small.

I don't recall from my 1990's experiments with the HF11 if distortion is that high when Rds >>100KΩ.
My recollection was that it was far worse when Rds was low and delta-Rds became a greater percentage.

A 0-1 mA LED range was what I had in mind.
If the HF11 was placed in parallel with a 10K feedback resistor (R3 below) the pull range at about 1 mA would about -3% which is perhaps 30 times more than needed.
That configuration exposes the H11F to a fairly large terminal voltage but, at the low currents the delta-Rds, and the distortion caused by it, may be small.

In the above figure:
Make R4 a 4K99 composite with a fixed R and a trim.
Parallel R3 (10KΩ) with an HF11.
At the output place a FW passive rectifier with an isolation resistor, LED and smoothing cap to filter ripple.
AC couple the output to the FW rectifier so the LED common is V- to reduce induced distortion from LED return current.

[ruffrecords]

The other small beads I looked at are higher, such as the Amphenol/Thermometrics P20, are 140µW/°C.
The CTS are 100µW/°C with a fast 1s response time.

I posted the links to these earlier.

Here's the CTS: https://www.ctscorp.com/Files/DataSheet ... asheet.pdf

The CTS is spec'd it at "0.1 Mw/°C" but what they really mean is 0.1 mW LOL. (Or 100 µW/°C)

The two things that give the RA53 it's sensitivity are the small chip size and operating in a vacuum.
At 1 mA it's definitely in "glow mode."

The CTS ones look promising but I cannot find anyone that stocks them. I checked all the stockists listed on their web page that are available this side of the pond.

Cheers

Ian

[mediatechnology]

I had to try opto-FET-based stabilization of the "Pocket Oscillator."

Thanks for that EDN article.

Since the Opto FET will operate over a very narrow control range and also has negative feedback control non-linearity may not matter much.

What I need to learn is can it be degenerated enough (if necessary) so the FET terminal voltage remains small.

I don't recall from my 1990's experiments with the HF11 if distortion is that high when Rds >>100KΩ.
My recollection was that it was far worse when Rds was low and delta-Rds became a greater percentage.

A 0-1 mA LED range was what I had in mind.
If the HF11 was placed in parallel with a 10K feedback resistor (R3 below) the pull range at about 1 mA would about -3% which is perhaps 30 times more than needed.
That configuration exposes the H11F to a fairly large terminal voltage but, at the low currents the delta-Rds, and the distortion caused by it, may be small.

In the above figure:
Make R4 a 4K99 composite with a fixed R and a trim.
Parallel R3 (10KΩ) with an HF11.
At the output place a FW passive rectifier with an isolation resistor, LED and smoothing cap to filter ripple.
AC couple the output to the FW rectifier so the LED common is V- to reduce induced distortion from LED return current.

I placed the optoFET across R3.
The output has two diodes and two resistors added to provide FW rectification.
The diodes form the upper arms of a bridge - the resistors the lower arms connected to 0V.
The lower arm resistors are 10K.
The LED connects between the bridge arms.
A 2u2 cap is placed across the opto-FET LED.
R3 is 10K; R4 is a 4K74 + 500 Ohm trim.
The trim adjusts level. Levels above +6dBu are regulated well.
Below +6 dBu setting is a bit touchy owing to the LED being just on the verge of conduction.

The THD measures around 0.015%.

EDIT: Made some adjustments on the ProtoBoard and got 0.007%.

Most of the THD seems to be unrelated to the FET.
I suspect some of it is ProtoBoard-induced.
Still pretty good.
This is do-able with an optoFET...

[Tubetec]

Thats an amazing discovery Wayne . congradulations

[mediatechnology]

Thanks!
I'm going to see if I can improve on it.
The sweet spot for THD is around 4-5V P-P.
And although THD is low there are high-order harmonics >=HD4 that I don't like seeing.

I does not appear to be detector ripple though - slugging it with a high value capacitor makes no difference.
I also doubt it's rectifier currents because I can isolate it with a fairly high value resistor and it also has little effect.

It looks like it may be common mode distortion due to the non-inverting topology.
Although the CM voltage is relatively low the source impedance of the bridge is high.
At resonance it's about 6KΩ.

The rectifier stops working below 3V P-P or so due to the Vf of the rectifier.
I need to see if I can improve on that as well.
That 0.007% THD measurement was made with it right on the edge around 3V P-P.