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

kdausman said:

So the writeup I am working from says to adjust the circuit such that the air/open setting on the capacitor is about 1.3 MHz.

Click to expand...

kdausman said:

As I went to higher frequencies and amplitudes, I found that the waveform shifted from an almost sawtooth form (rounded on the rise) to what you see here. I was able to get the frequency up to 1 MHz and the amplitude up to about 500 mV, but as I did so, the bottom of the waveform seemed to get cut off (like you see in the picture, but more extreme) and the smaller extra peak before the main peak grew in prominence. And that extra peak seemed to depend as well on the external cell capacitance, leading me to worry that the frequency counter might start picking them up as separate pulses, screwing up the measurement.

Click to expand...

As discussed earlier, although this oscillator circuit works, it does have its limitations.

Scratching my head, I remembered a circuit for a 1MHz oscillator which I knocked up for someone a few years ago.
Look here:
https://www.electro-tech-online.com...lex-microcontroller-coms.134988/#post-1132639
It runs at 1MHz and is easily adjusted by varying the turns on a single layer coil.
The output from the emitter of the transistor is a close approximation to a sine wave with an amplitude of nearly 2.5v p-p.

To re-iterate what I said yesterday:

JimB said:

If I were starting from scratch, I would select an oscillator circuit which used a simple inductor (ie just one winding).
Then using a reel of wire and suitable former, it would be easy (five minute job) to make whatever inductor was required.

Click to expand...

JimB

PS, I think that I have "oscilloscope envy".

JimB said:

PS, I think that I have "oscilloscope envy".

Click to expand...

Don't we all, when we see something better-spec'd than what is currently sat on the bench in front of us?
I have a cal-failed TDS220, for obvious reasons, bought from an e-waste seller for £20 around 2011, IIRC.
It was spotted on a display table at Bowlers Computer Fair, in Manchester, with CH1 BNC socket completely adrift. (I swapped the EXT-TRIG BNC socket to CH1).
He informed me that it was a signal generator and randomly-twiddled all of the knobs whilst un-powered, muttering "This is sine, square, triangle, and this is speed and other mumblings. £50 or you can make me an offer..." - so I did.
Swapped the BNC, made sure the recall had been done, reset the factory settings and ran through the self-cal, which passed fine.
It does all that I personally need so far, but it sure would be nice to have something more sexy, like a bandwidth-hacked DS1054Z to look at.

At work I have a Tek Series 5. But then I found out the Keysight UXR existed.

kdausman said:

Looking at the **broken link removed**, I don't see specs on the voltage amplitude necessary to count as a pulse. Another site with the same product suggests that it has sensitivity "better than 60mVPP," which I presume means 60 mV peak-to-peak. Given that, should I be aiming for a cleaner waveform at the expense of frequency and total amplitude? Or am I likely to be ok with this waveform?
Thank you all for the amazing help!

Click to expand...

The counter should have an input attenuator so you can limit the minimum input you are interested in. If it doesn't have one, is easy make.
You should aim to get the cleanest waveform you can anyway. I don't know if harmonics would actually affect your measurement, but they might, especially if they change when you adjust the cell.

(Edit) I just followed your link. Oh, one of /those/ sort! You probably do need to make an attenuator. A simple potential divider will do it.

dr pepper said:

I've wondered about those hobby shop various coloured copper wires, are they actually copper?, I wondered if they'd be aluminium or something.

Click to expand...

so far, the only ones i've seen are copper.

I'll get a set of those then.

Ok, I have everything together and stable, fully shielded with coaxial cables connecting the component=s. I can trim the frequency using the internal variable capacitor, and then can use the external capacitance cell in both open and closed positions, and see different frequencies when I do. I'm still measuring the frequency with an oscilloscope, because my frequency counter hasn't come in yet, but at least things are looking like they are working relatively well. I had a devil of a time tracking down an intermittent problem that resulted from me failing to ground the housing for the RF transformer, but I seem to have all of the bugs worked out. The waveform is quite nice, and I have an amplitude of about 100 mV. My one concern is that the frequency is about an order of magnitude below what I wanted. I'm getting about 150 kHz with both capacitors open, and the lab writeup suggested we wanted 1.3 to 1.5 MHz. Of course, I have never done this experiment before, so I don't know how critical that is.

Looking at the figure in post #29, I followed that diagram pretty closely with the following modifications: (a) 222 pF capacitor between signal and ground instead of 300 pF (because that's what my stockroom guy got me a lot of for some reason), (b) a 0.1 microF capacitor in the signal line as suggested by Nigel, (c) the transistor is a 2N3904 (more on that in a moment), (d) the transformer is the 42IF106-RC suggested by JimB, and (e) the variable capacitor is actually two variable capacitors in parallel, one in the shielded case to allow frequency trimming and one external for the liquid cell. Note that I am using an NPN transistor despite following the suggestion made by a couple of folks to switch to a positive bias voltage (as post #29 indicates). I did that because I had a whole bag of them my stockroom guy got me, and I didn't have any pnps, so I thought I would just try it and see what happens. And it worked, so I assume that I misunderstood something, and I would love to learn what I misunderstood.

Any thoughts?

The frequency will be low, as you're using a 455KHz IF transformer, and then adding extra capacitance to it - so it's going to be considerably less than 455KHz.

kdausman said:

My one concern is that the frequency is about an order of magnitude below what I wanted. I'm getting about 150 kHz with both capacitors open, and the lab writeup suggested we wanted 1.3 to 1.5 MHz.

Click to expand...

The original circuit suggested a coil with an inductance of 0.6mH.
After various experiments etc, you said that you would prefer to buy a coil (transformer) so I suggested the 42IF106-RC from Mouser, which according to the datasheet has an inductance of 680uH (0.68mH), near enough to 0.6mH

The datasheet also states that the transformer ( 42IF106-RC ) is for use at 455kHz using a 180pF parallel tuning capacitor.

The circuit that you are working to has a 0.6mH inductor and a 300pF capacitor plus whatever capacitance of the test jig, the frequency can only go down when adding capacitance. (0.6mH and 300pF will give 377kHz before considering all the usual circuit stray capacitance etc).
Which is why I suggested an alternative oscillator circuit here:
https://www.electro-tech-online.com...lex-microcontroller-coms.134988/#post-1132639
A circuit which uses a single layer coil, which is easily wound to give the inductance you require and hence the frequency that you require.

JimB

A capacitor marked 222 is 2200pF ie 2.2nF, not 222pF.

throbscottle said:

A capacitor marked 222 is 2200pF ie 2.2nF, not 222pF.

Click to expand...

A good point.
But that 300pF cap across the 2.2k resistor is an odd one, it is effectively decoupling the output to the frequency counter, not what you really want.

JimB

So I wanted to provide a huge thank you to the community, and ask a follow-up question. First, the thank you: I got the circuit working and my students have successfully been able to use it to measure dielectric constants of solvents and solvent mixtures. We used the oscilloscope determination of frequency because the frequency counters were taking forever to get shipped to us, but that was sufficient to be able to get the dielectric constant to within a couple of percent, and then the dipole moment of molecules to within about 10 percent, which is awesome if you think about it.

Now that the semester is over, and the frequency counters have come in, I am trying to use them. This are the styles of counter that I am working with:

I have two that look like this: https://www.cwtd.org/frequency/RF Signal Frequency Counter.pdf (although the versions I have don't seem to have the "Move a bit to the right" menu option)

I have one that looks like this: https://www.amazon.com/0-1MHz-Signal-Frequency-Counter-Cymometer/dp/B00PQ30GV4 (I don't like this one because I need the extra digits).

It seems that when I T off of the signal output into the oscilloscope and the frequency counter, the waveform gets disrupted dramatically. Amplitude drops quite a bit, and the waveform shape really goes to hell. When I try using a signal generator in the same amplitude and frequency range and my home-built oscillator, I *sometimes* get an unmodified signal, but sometimes that gets screwed up too, and I have not been able to identify what governs that behavior.

Is there a termination issue? Is it an impedance-matching problem? How could I fix this? Thanks!

kdausman said:

I got the circuit working and my students have successfully been able to use it to measure dielectric constants of solvents and solvent mixtures.

Click to expand...

It is always good to hear a success story. Thank you.

kdausman said:

It seems that when I T off of the signal output into the oscilloscope and the frequency counter, the waveform gets disrupted dramatically. Amplitude drops quite a bit, and the waveform shape really goes to hell.

Click to expand...

OK.

First of all, can you tell us which circuit variant you are actually using?
We discussed several variations earlier in the thread.

Also, how are you connecting the 'scope and counter to the test jig/oscillator?
Are you using
an oscilloscope probe?
a coax cable?
a pair of simple wires?

The oscillator will be susceptible to anything being connected to it, the best thing you could use is possible a X10 scope probe.
This will have a low resistive and capacitive loading effect on the oscillator.

The basic counter circuit is likely to have a fairly low input impedance, which is not specified in the Amazon advert. This is why you are seeing odd effects when it is connected to the oscillator.

It may be possible to build an input attenuator, with high impedance, for the counter.
Or it may be better to make a simple buffer amplifier to provide isolation so that the test equipment does not load the oscillator.

JimB

Here is the final circuit that I used (I think... unfortunately I have slept since then. If it doesn't look right, I can open up the box and check):



The middle part is in a cast aluminum box. The power (+15 V) and signal output (marked as a voltmeter in the diagram because my circuit diagram software doesn't have an oscilloscope symbol) are connected by chassis-mounted BNCs. The external variable capacitor is a pass-through pair of wires to a banana-plug/BNC adaptor, which then goes via coaxial cable to a BNC plug on the external capacitor cell. The transistor is a 2N3904, and the transformer is a 42IF106-RC. In a future iteration of this project, I will look at using a different inductor to get a higher frequency, but for the initial test-run, this worked pretty well.

When I run the signal output directly to the oscilloscope, I run a BNC coaxial cable directly to the scope. I have the scope set to 1 MOhm termination. When I try to put the frequency counter in, I put a BNC-T at the output of the circuit box, and run one coaxial cable to the oscilloscope, and the other coaxial-to-banana-plug cable to the pair of wires that go into the frequency counter.

Does that give you enough information?

kdausman said:

Does that give you enough information?

Click to expand...

Yes, I understand how it is connected.

I suggest that you make a buffer amplifier to prevent loading of the oscillator by the counter.
I had a bit of a play around today and came up with this:


A simple amplifier with a gain of one, but it will provide isolation to stop the counter loading the oscillator.

kdausman said:

I will look at using a different inductor to get a higher frequency

Click to expand...

If you decide to do that, I fervently suggest that you select a different oscillator circuit.
I have two main reasons for suggesting that:
1/ As discussed earlier in this thread, that oscillator circuit is a bit odd in some respects.
2/ The circuit uses a transformer, it is much easier to make and modify a coil with a single winding.

JimB

Thank you! I will give your amplifier a try.

And yes, I will use a different oscillator circuit if/when I look to move to a higher frequency. Thank you for the reminder.

Hmmm. I’m getting some strange behavior with your circuit. When just the scope is hooked up, I am getting a superposition of a ~200 kHz oscillator and a 71 MHz oscillator.



Then when I T off the frequency counter, the ~200 kHz signal is damped nearly completely, and the high-frequency signal is all that is left. It goes up to 77 MHz, and seems mostly independent of the variable capacitors.

kdausman said:

I’m getting some strange behavior with your circuit. When just the scope is hooked up, I am getting a superposition of a ~200 kHz oscillator and a 71 MHz oscillator.

Click to expand...

Well, one of the old jokes about radio/electronics goes something like:

When you try to make an oscillator, it will not oscillate.
When you try to make an amplifier it oscillates.

Where you are getting 200kHz and 70Mhz from I hesitate to speculate.

Here is the one which I built, it is the same oscillator tidied up a bit, and the amplifier with long spidery leads on the resistors.



On the oscilloscope, the output of the amplifier looks like this:



JimB

JimB said:

Well, one of the old jokes about radio/electronics goes something like:

When you try to make an oscillator, it will not oscillate.
When you try to make an amplifier it oscillates.

Click to expand...

I love it!

So my circuit has given up the ghost. Something shorted, and it looks like I would need to do some significant debugging to find exactly what. Instead, I will take your advice and try a better oscillator circuit to begin with.

So, I am taking a look at the thread you pointed me to earlier, and that certainly looks doable. Just to make sure, I am looking at this:



In that thread, you set:

C parallel = 330 pF
C coupling = 1500 pF
C feedback = 3000 pF
Inductor = 23.5 uH

Assuming this is the right starting point, I have some further questions/requests for clarification:

1) You talked about winding your own inductor. Can you point me to a good resource on how to do that? I also notice that there are commercially-available variable inductors; would that be a good way to go? Or if I know what inductance I want, can I just buy a fixed inductor?

2) Recall that for my application, I need to include a variable capacitor that will modify the oscillator frequency. The variable capacitor I currently have in my external liquid cell is this product, which has a capacitance that varies from 12-144 pF. In the open position (12 pF), in air, I want the oscillator frequency to be around 1 MHz. I will be measuring the frequency open and closed, in air and in liquid, for a total of four measurements. The calculation I will be doing depends on the frequency of the oscillator being given by f=1/(2 pi Sqrt(L C)), where C here is the capacitance of the system as a whole. By making these four measurements, we can remove the effects of the capacitance of the portions of the system that are NOT due to the variable capacitor. We then get the dielectric constant of our sample by:



Here a and b correspond to open and closed positions for the variable capacitor. So in your circuit diagram, where does this variable capacitor go? (My guess would be C parallel, but I'd rather ask and be sure).

3) What effect will changing the capacitance of the variable capacitor be on the resulting waveform? It sounds from your discussion on the other thread that all of the capacitance values are chosen carefully to give stable waveforms at a specific frequency, but the goal here is to modify the frequency by changing the capacitance. Will the oscillator still work at multiple settings of the variable capacitor?

4) Does the specific transistor (PBC108) matter?

5) The discussion in the other thread you mention that the amplifier could probably be improved by connecting the collector directly to the supply. Is that something I should consider?

6) You also talk about the output having harmonics. Is that something that will cause me problems for my application?

What else have I forgotten?

Thank you so much!

kdausman said:

Just to make sure, I am looking at this:

Click to expand...

Correct.

kdausman said:

In that thread, you set:

C parallel = 330 pF
C coupling = 1500 pF
C feedback = 3000 pF
Inductor = 23.5 uH

Click to expand...

Yes.

kdausman said:

You talked about winding your own inductor. Can you point me to a good resource on how to do that?

Click to expand...

Have a look at these.
A bit verbose and waffling but he get there in the end.

http://info.ee.surrey.ac.uk/Workshop/advice/coils/#cor A bit dry, and at a quick read, short on practical detail.

If you look back at my old post from 2013, you will see that I wound the coil on a simple cardboard tube.
That tube was actually the centre of of a roll of aluminium cooking foil.
The wire I used was (probably) 28SWG, which is about the same as 27AWG, but the exact gauge is not critical.

kdausman said:

Recall that for my application, I need to include a variable capacitor that will modify the oscillator frequency.

Click to expand...

Yes I have redrawn my schematic to include the test cell capacitor:

I would suggest that you reduce the value of Cparallel by maybe half of the capacitance of the test cell, so that Cparallel becomes around 270pF.

kdausman said:

What effect will changing the capacitance of the variable capacitor be on the resulting waveform?

Click to expand...

Within the tuned circuit, the waveform should be very close to a clean sinewave.
Depending on the circuit conditions the amplitude of the oscillation may vary.

kdausman said:

Will the oscillator still work at multiple settings of the variable capacitor?

Click to expand...

Yes, no problem at all.

kdausman said:

Does the specific transistor (PBC108) matter?

Click to expand...

No, most small signal transistors will work OK, I think that you used 2N3904 in the original circuit, you can use that here also.

kdausman said:

The discussion in the other thread you mention that the amplifier could probably be improved by connecting the collector directly to the supply. Is that something I should consider?

Click to expand...

I will do a quick "lash-up" of this circuit in the next day or two and let you know how it works.
At the moment I am thinking that it may work better with a higher supply voltage, say 15v as used on the other circuit.

kdausman said:

You also talk about the output having harmonics. Is that something that will cause me problems for my application?

Click to expand...

I don't think so.
Harmonics are caused by distortion in the amplifier, but as I said earlier the waveform within the L-C tuned circuit will be very close to a pure sinewave.

JimB