Need help interpreting an old circuit diagram: Transistor and RF Transformer

[JimB]

I have been experimenting with this circuit over the weekend, and getting what I think are encouraging results.
But before we get all excited, can you give me an indication of the relative permittivity of the liquids which you intend to test in your capacitance cell?

I have been trying to balance good stability and reliable oscillation with sensitivity to change of capacitance,
ie getting a reasonable change in frequency for a given change in capacitance.

JimB

 

[JimB]

I had some electronic playtime today, and came up with this:

Physically it looks like this:

At the collector of the transistor, the waveform is a bit ugly:

But the waveform across the test cell is a nice clean sinewave.

Doing a few tests to see how the frequency varied with changes in capacitance, I had these results:

I tried one set of feedback and coupling capacitors, as per my original calculations.
I tried two different coils, one with 25turns (19.7uH) and one with 35turns (32.2uH).
I used an air variable capacitor to represent your test cell to give the open (12pF) and closed (140pF) values.
I added an external capacitor to represent the cell full of liquid with a permittivity of ~2 (275pF).

Cfeedback =	3000pF		Inductor		Ctestcell	12pF	140pF	275pF
Ccoupling =	1500pF		19.7uH		F (MHz)	1.22873	1.147612	1.065834
Cparallel =	0pF		32.2uH		F (MHz)	0.966634	0.903535	0.839257
Vsupply =	10V

I repeated the tests using feedback and coupling capacitors of half the original calculated values, results as per:

Cfeedback =	1500pF				Ctestcell	12pF	140pF	275pF
Ccoupling =	750pF		19.7uH		F (MHz)	1.703183	1.50594	1.335409
Cparallel =	0pF		32.2uH		F (MHz)	1.340274	1.187051	1.053532
Vsupply =	10V

JimB

 

[kdausman]

Thank you immensely for your work on this! The range we are interest in ranges from 2 to 80. I have two main applications:

1) A teaching laboratory where we measure small changes in the 2 to 5 range. High precision is needed here.

2) A research project that involves validating values for solvent mixtures over the 2-80 range. Less precision is needed here.

Given the work you have put into this, would you at all be interested in being a co-author on the paper we are putting together? You have definitely contributed enough intellectual content and effort to justify that. The paper is primarily on distinguishing between two models in the literature used to explain a strange color-change behavior in solutions/suspensions of buckminsterfullerene in organic solvents, but it is in part rebutting faulty methods used to estimate the dielectric constant of solvent mixtures in those prior papers. The correct theoretical models are well-established, but having experimental verification for our specific system is of course quite nice. And on top of that, if we can get into the literature a better, more reliable oscillator design for this application, then maybe chemists will start moving away from the crappy design that started this thread.

 

[JimB]

The range we are interest in ranges from 2 to 80.

From an electronic point of view that could be "challenging".
Let me give that some more thought.
It may be a good idea to have two or three cells with different sized variable capacitors for the test cells.
A small variable capacitance, in conjunction with a larger value of fixed capacitance (Cfeedback and Ccoupling), LC oscillators don't work well (if at all) with a min-max capacitance ratio of 80:1.

Given the work you have put into this, would you at all be interested in being a co-author on the paper we are putting together? You have definitely contributed enough intellectual content and effort to justify that.

Praise indeed! Thank you.
Let me think about that.

JimB

 

[kdausman]

Could variable capacitors in the feedback and coupling positions, which could then be tuned based on the expected dielectric, help? Or is the issue simply that the open/closed difference would be too great?

 

[JimB]

Or is the issue simply that the open/closed difference would be too great?

Yes, this is the condition that I was concerned about.

When the test cell is full of "air" (permittivity =1), say the capacitance is 140pF.
When the test cell is full of "super stuff" (permittivity = 80), then the capacitance will be 11200pF.
This is a large ratio, much larger than what is generally used in LC oscillators.

Based on previous experience and recent experiments with the test circuit, I was thinking that by using a smaller capacitor, say 20pf in air, when immersed in "super stuff" the capacitance would be a more reasonable 1600pF.

However, having reconfigured the oscillator circuit as:

This seems to work with a much wider range of test cell capacitance, as shown below:

Test Cell Capacitance (pF)   0     275    750   1500    3000    5000   10000   15000
Frequency (MHz)            1.046  0.966  0.867  0.757   0.618  0.520   0.393   0.329

I created the varying values of capacitance by simply connecting fixed capacitors across the open variable capacitor seen in previous set-ups.

How well this would work in your application I do not know, resolution at low values of permittivity may not be sufficient.
I guess that needs to be tested in the chemical lab.

JimB

 

[kdausman]

I have finally been able to get into the lab and start building this. I spent most of today trying to track down a strange issue where the waveform across the test cell was beautifully sinusoidal with a frequency around 1.5 MHz, but the output to the counter was some multiple of that (often 2x, but occasionally up to 6x). I finally tracked the issue down to the power supplies. I am using what are supposed to be high-stability variable power supplies (one driving the oscillator and the other driving the counter), but when I swapped them out for good-old 9V batteries, everything calmed down. The output to the counter is deformed a bit, but not so badly that I cannot use it.

I will post updates once I get it together in a more permanent setup. Thanks again for the help!

 

[JimB]

but the output to the counter was some multiple of that (often 2x, but occasionally up to 6x). I finally tracked the issue down to the power supplies.

I am not sure about this case, but power supplies can sometimes do odd things.
Usually this is due to the output "floating", ie the negative side of the supply is not connected to earth. This is quite deliberate by the PSU manufacturers, because sometimes you need a PSU which is floating, but other times it can cause odd effects.

waveform across the test cell was beautifully sinusoidal with a frequency around 1.5 MHz,

The output to the counter is deformed a bit, but not so badly that I cannot use it.

I would expect the test cell waveform to be sinusoidal, it is part of an LC tuned circuit.
The distortion in the signal to the counter should not matter, the fundamental frequency is still 1.5MHz (In this case).

As I type this it occurs to me that distortion is harmonics.
If there is something about the connection to the counter which effectively creates a high pass filter, that could suppress the fundamental and enhance the harmonics to that the counter is responding to one of the harmonics.

It is good to hear that you are getting some reasonable results.


JimB

 

[kdausman]

Status update: The instrument is looking REALLY good, with only one exception (details to follow). Here's the completed box while it is making a measurement:

The really high frequency I will explain in a moment. And here is the sample cell:

And here are the innards:

I'm using a fixed inductor. I have separate batteries powering the oscillator circuit and the frequency counter, each with a switch. I have a BNC to go to the sample cell, and a BNC to give me an output through which to look at the signal on the oscilloscope should I need to do debugging.

Amusingly, the frequency counter has the signal in backwards from standard; the black wire goes to signal and the red wire to ground, so I have jammed the connected in backwards. It works, but I will probably resolder once I am confident everything is working right. I don't want to mention how long it took me to find that error...

Stability is quite good. I did a run using toluene as a test solvent (literature dielectric constant of 2.39). Here were the frequencies I got:

Air open: 1.6454 +/- 0.0002 MHz
Air closed: 1.5370 +/- 0.0002 MHz
Toluene open: 1.6291 +/- 0.0002 MHz
Toluene closed: 1.4057 +/- 0.002 MHz (yes, the uncertainty on that one is higher. For some reason that reading wasn't particularly stable compared to the others)

When I run through the calculation on those frequencies, I get an experimental dielectric constant for toluene of 2.40 +/- 0.03. Just over 1% uncertainty, and bang-on the literature value! I call that a magnificent success!

Ok, now about the ridiculously high frequency shown above. That was what I got with water in the cell (dielectric constant around 80). Looking at it on the oscilloscope, the frequency counter's answer is bang-on what the waveform looks like. And it is nicely sinusoidal and quite stable. That is on the high end of what the frequency counter seems to be able to measure, and I had to futz with the modes on it to find settings that would reliably give me that frequency. Notice that the frequency is HIGHER than it is in air, acting like water has a dielectric constant that is much less than 1.

So what is going on here? I have an idea, but wanted to run it by the group to see if it makes sense. Water, depending on the level of ionic solutes in it, is conductive. This would short the circuit across the capacitor. Would that cause an increase in frequency from the oscillator? The water I used is "NanoPure" water, which has been rigorously deionized to a resistivity of 18 MOhms x cm. But the water WAS exposed to the air before putting it into the sample cell, and thus undoubtedly had gotten some CO2 dissolved in it from the air, which forms carbonic acid (which is ionic upon dissociation), and so the resistivity drops to below 8 MOhms x cm in about a minute. From an electronics perspective, could this be what is causing these strange results?

Thanks again for all of the help.

 

[JimB]

Wow!

Overall I am impressed.
But wire nuts? Hmmm... not so much.

As for the high frequency when the cell is filled with water, I have no explanation.

JimB

 

[Nigel Goodwin]

Overall I am impressed.
But wire nuts? Hmmm... not so much.

It's an American thing Jim, civilised countries banned them decades ago!

Got to love his hole for the display though - nice!.

 

[JimB]

Got to love his hole for the display though - nice!.

I noticed that, looks like it was done with a milling machine..

JimB

 

[Nigel Goodwin]

I noticed that, looks like it was done with a milling machine..

As he seems to be in a University, presumably he's got access to machines (or even people to make the box for him).

I know in my travels with my daughter in 'University land' they seem to have amazing capabilities for making things, and have on-site engineering departments for making anything you can seem to think off. I was mega impressed to find out they can (and do) manufacture integrated circuits where my daughter is now (University Of Twente), and I imagine it's fairly commonplace in Uni's?.

In fact, as I mentioned before, Melissa had to make a wafer of solar cells to pass her 'clean room' certification.

 

[ronsimpson]

Physically it looks like this:

Now there is a breadboard made by some one that knows RF.

 

[JimB]

Now there is a breadboard made by some one that knows RF.

Ooh Ron, you say the nicest things!

JimB

 

[kdausman]

Overall I am impressed.
But wire nuts? Hmmm... not so much.

As for the high frequency when the cell is filled with water, I have no explanation.

Once I get some testing with some other solvents under my belt, I will replace the wire nuts.

If I were to manually short the circuit across the capacitance cell, would that damage the circuit in any way? I am think that doing that would test whether water conductivity shorting the capacitor is a reasonable explanation.

 

[kdausman]

I noticed that, looks like it was done with a milling machine..

Yup. We have a machine shop that is pretty well-equipped.

 

[JimB]

If I were to manually short the circuit across the capacitance cell, would that damage the circuit in any way?

No.

I am think that doing that would test whether water conductivity shorting the capacitor is a reasonable explanation.

My first thought is that the oscillator would just stop if the test cell capacitor were shorted.
But, a length of coax cable can under some circumstances look like an inductor, that may be what is happening here. (Wild guess).

JimB

 

[kdausman]

No.

Ok, then I will run the test.

My first thought is that the oscillator would just stop if the test cell capacitor were shorted.
But, a length of coax cable can under some circumstances look like an inductor, that may be what is happening here. (Wild guess).

Interesting, and that matches my understanding of the physics. I’ll try shorting both with and without the coax. Thanks for the feedback!

 

[kdausman]

Ran the test, and indeed saw a 5+ MHz signal when shorting the circuit at the sample cell after the coax cable, but not when shorting it right out of the box. This means that I cannot measure dielectric constants by this method on liquids that are conductive, but that is neither surprising nor a problem for my applications. Once I clean up the connections (fixing the signal leads to the frequency counter and replacing wire nuts with direct wiring), I’m going to call this project done. Thank you for all of the help! And my offer of a co-authorship is still open!