Help me understand this circuit better. (Please)

[ukcol]

The circuit below shows the reactance stage from the ATC system in a Murphy A40C receiver. ATC stands for Automatic Tuning Correction and is different only in name to the modern label AFC.

The circuit is built around a Mazda AC/SP1 valve. Here is a short note from the AC/SP1 data sheet which describes this particular application.

When this valve is used in automatic tuning arrangements, the tuned circuit of the [local] oscillator is included in the anode circuit of the valve, and a resistance of the order of 100,000 ohms, in series with a blocking condenser, is connected between the anode and grid G1, while a small condenser is connected between grid G1 and cathode; in this manner a voltage in quadrature is applied to the control grid, and this gives an anode circuit admittance equivalent to a negative inductance across the oscillator tuning circuit. This admittance is directly proportional to the working slope of the grid G1. By initially biasing grid G3 in the middle of its operating characteristic it is possible to vary the frequency of the oscillator either up or down by applying either additional positive or negative volts from a frequency-discriminating circuit to grid G3. The initial biases on grid G1 and G3 should be obtained by the self-bias [cathode-bias] method, and the resistance in the grid G3 should not exceed 3 ½ megohms.

Most of the circuit is fairly straightforward to understand.

R37 is the cathode bias resistor for grid G1, and R37 plus R38 provide the cathode bias for grid G3. R38 is adjustable so that the bias can be set in the middle of the G3 characteristic as mentioned in the application note above. C50 is the cathode decoupling capacitor. The control voltage from the discriminator is feed via the time constant R40/C51 to the grid G3. S6 is a switch fitted via a clutch mechanism to the tuning gang and closes to defeat ATC action during operation of the set's tuning control. S2 is part of the selectivity switch and S3 a small switch on the back of the set. This allows the ATC to be defeated in the first two positions of S2 depending on the setting of S3. Although R39 is in the time constant circuit its value is very small compared to R40. Its actual function is to limit the discharge current of C51 when any of the defeat switches are closed; C51 is a relatively large value (4uF).

In its simplest form as described in the application note the 90 degree phase shift between anode and grid G1 would be provided by R36 and a small value capacitor to cathode. The designers of this circuit have chosen to use a network of components as can be seen in the diagram, namely R36, C48 L25 and C47 the latter only being in circuit when the local oscillator is running at medium wave frequencies. C49 is the DC blocking capacitor.

Before I ask my questions a little note about S1 is appropriate here.

S1 is a 3 position wave-change switch. Going clockwise the positions are SW, MW and LW respectively. The reason that the first two positions of S1g are linked is that in both positions the local oscillator is running at MW frequencies; this is because the A40C is a dual conversion receiver on SW that uses the MW section of the receiver as the first I.F.


My Questions.

1] I don’t understand is the function of L25. Clearly there needs to be a DC path from the grid G1 but why not a resistor?

2] C47 is switched in on MW and out on LW, this seems counter intuitive to me because I would have expected less capacitance to be required at the higher (MW) local oscillator frequencies.

3] Why is C47 returned to cathode and not to the junction of R37/R38 as is C48?

Extra information that may help with providing answers.

1] The local oscillator frequency range on MW is approximately 650 KHz to 1.5MHz and on LW 265 KHz to 320 KHz.

2] The inductor L25 has a value of about 1.5mH and a resistance of about 6 ohms. With the 45pF capacitor C48 this gives a resonant frequency (if that’s even relevant) of about 600 KHz.

[kalee20]

Some of this does not make sense! If the LO is running at 265 - 320kHz on LW then this is BELOW the resonant frequency of C48 and L25 (612.5kHz). So the grid circuit looks inductive, and on LW the reactance valve would look like a negative capacitor, not a negative inductor. Are you sure that S1g is as you say, ie open on LW, closed on MW? As you observe, the reverse would make more sense. (It is of course possible, though I feel unlikely, that on LW the thing is designed to look like the opposite reactance to that at MW).

I can see the reasoning behind an inductor rather than a resistor to maintain the DC path, g1-k. A resistor would stop the grid circuit looking purely capacitive. This would reduce the overall 90 degree phase shift between Va and Ia, so the thing would look like a more lossy reactance than with the inductor as grid leak.

Can't comment right now as to things are returned to cathode in some cases and to the bias chain in others. I'm observing that C50 is large (0.5uF) so the RF voltage on it is for all intents and purposes, zero. C47, when switched in, goes to, what is for me, the ideal place (cathode); if I was designing the circuit I'd have C48 returning to cathode too. L25 obviously has to go to R37/R38 junction. Maybe somebody else will come in with a subtlety which escapes me, I can't wait to learn something new!

[Radio Wrangler]

Perhaps because it's an active gyrator, inverting the impedance in the g1-k limb to form the anode impedance. Less becomes more?

David

[Argus25]

Maybe they added R36 to help isolate the circuit capacitance ( which provides the grid drive voltage) from the main tank circuit and having added that resistance it caused a phase shift which they then had to at least partially correct by adding the inductor to get the grid g1 drive voltage back to the correct phase relation ( or close) to what it would have had without the series resistor present ? Just a guess, though it probably would be easy to figure out with a good dual trace scope connected to it.

[Radio Wrangler]

R36 is quite a large value and the capacitor is rather small, and operation is needed across a few octaves. There's a lot going on!

David

[ukcol]

Originally Posted by kalee20

Some of this does not make sense! If the LO is running at 265 - 320kHz on LW then this is BELOW the resonant frequency of C48 and L25 (612.5kHz). So the grid circuit looks inductive, and on LW the reactance valve would look like a negative capacitor, not a negative inductor. Are you sure that S1g is as you say, ie open on LW, closed on MW? As you observe, the reverse would make more sense. (It is of course possible, though I feel unlikely, that on LW the thing is designed to look like the opposite reactance to that at MW).

Good point that man. I had made the mistake of relying on the A40C manual. I should have been aware of this possibility because I was already aware that the manual contains errors. Examination of my set confirms that S1g should indeed be closed on LW and open on MW and not as drawn in the manual. I have corrected the drawing below.

Originally Posted by kalee20

I can see the reasoning behind an inductor rather than a resistor to maintain the DC path, g1-k. A resistor would stop the grid circuit looking purely capacitive. This would reduce the overall 90 degree phase shift between Va and Ia, so the thing would look like a more lossy reactance than with the inductor as grid leak.

Thanks, that makes sense to me and was aspect of this circuit that i found most puzzling.

Originally Posted by Radio Wrangle

Perhaps because it's an active gyrator, inverting the impedance in the g1-k limb to form the anode impedance. Less becomes more?

I'm sorry David I wasn't able to follow that argument at all . It must be marvellous to have had a proper education .

[ukcol]

Originally Posted by ukcol

Good point that man. I had made the mistake of relying on the A40C manual. I should have been aware of this possibility because I was already aware that the manual contains errors. Examination of my set confirms that S1g should indeed be closed on LW and open on MW and not as drawn in the manual. I have corrected the drawing below.

The drawing error associated with S1g is NOT in the Murphy full manual (available from the banner, top right of this page). In the full manual, which also covers the A40RG radiogram, S1 is shown as a 4 position switch (to include the gram position) and S1g is correctly drawn as closed only on LW.

The circuit diagram available from Dave Grant's diagram dungeon site is drawn for the consul only and S1g is shown incorrectly closed on MW. I used the latter document to make my snip of V8.

[Radio Wrangler]

By using an amplifier with appropriate feedback, you can simulate impedances. With a circuit using just an amplifier and resistors, you can make a circuit which presents what looks like a negative resistance at a pair of terminals. This gets used for compensating losses in transmission lines or small signal inductors, but mostly it's used in oscillators.

Using an amplifier, some resistors and a capacitor, you can simulate an inductor. There is a style of active filter design where you make what amounts to a normal L-C design, but all the inductors are simulated! This can make inductors with better imperfections in some cases, or be cheaper.

The converse of this is to change the impedance of an inductor to look like a capacitor. This may look like wasting money, but the 'gain' of the conversion process can be varied so that you wind up with a voltage-controlled variable capacitor which in the days before varactor diodes were needed for AFC and FM transmitters.

The circuits which generally convert one sort of impedance into its opposite are called Gyrators.

Note that they can't turn resistors into capacitors or inductors or vice-versa Reactances stay reactive, resistances stay resistive (positive or negative)

You can also convert one sort of component into the same sort, but of a scaled value, so a capacitor can be converted into different value which can be electronically varied.

Monkeying about with the G3 voltage of the pentode changes the valve's natural gain. This acts with the feedback to convert the reactance seen at the grid into a variable reactance seen at the anode.... and that is what pulls the tuning of the local oscillator around.

Hence the name 'Reactance valve"

Educationally, this sort of stuff doesn't get taught very well if at all - certainly not the matter of seeing the similarities in different areas. It's something you've got to find out for yourself. If you're lucky you get a few hints.

David

[ukcol]

Thanks for that David, that helps considerably. I don't believe i had ever heard of "Gyrators" in the context of electronics, either in my education or 50 years of practical experience.

One interesting aspect of the A40 design is that there are no adjustable inductors. The setting up of the reactance stage is the only alignment adjustment for the local oscillator at the low end of the MW band. Only capacitive trimmers are provided for the rest of the alignment although on SW the manual suggests that it may be necessary to move the end winding on some of the coils to maintain "calibration" [tracking].

[ms660]

W.W. did at least one article on the gyrator.

Lawrence.

[ukcol]

Thanks Lawrence I'll look it (them) up. Have you any idea what year, or better still what issue?

[kalee20]

Originally Posted by Radio Wrangler

Perhaps because it's an active gyrator, inverting the impedance in the g1-k limb to form the anode impedance. Less becomes more?

Maybe.

Further to my earlier post, and having been thinking about it, I'm not confident in the circuit description in the manual, ie that it behaves as a negative capacitor. Such a thing has the phase shift (voltage and current) of an inductor, but the magnitude of the impedance goes down with frequency, ie like a capacitor. I can't see the circuit doing that; I reckon it behaves just as an inductor, with an inductor's phase shift and an inductor's Z versus f characteristic.

It will have a loss component - impedance can't be greater than the ra of the valve, which is effectively a parallel loss resistance. And at low frequencies there will be a series loss resistance because the voltage on the grid can't be quite 90degrees out of phase with voltage on the anode. Adding a bit of extra resistance in the grid circuit makes this worse, not better!

Instead, I'm wondering if the connections of C47 and C48 are for purely practical reasons - C47 going to cathode is the 'better' option to me, but sticking C48 across L25 and making a neat 2-lead sub-assembly was easier for production, with the slightly inferior circuit performance being more of academic interest than a shortcoming in practice? Or... Maybe even that the loss resistance varying in frequency was actually used to good effect, perhaps countering a change in natural oscillator amplitude with frequency, and this thing made it constanter!

[ms660]

Re: Post 11, there's a theory one that starts on mag page 423 in W.W. September 1957:

https://www.americanradiohistory.com...ld-1957-09.pdf

Lawrence.

[ukcol]

Originally Posted by kalee20

Further to my earlier post, and having been thinking about it, I'm not confident in the circuit description in the manual, ie that it behaves as a negative capacitor.

Hello Peter,

I'm not sure what manual you are referring to here. The A40 manual doesn't have a circuit description and the A28 (an earlier model that uses the same reactance circuit) only refers to the reactance of the circuit and the conductance of the valve.

In the application note from the AC/SP1 data sheet (that I quoted in post 1) reference is made to negative inductance. Could you clarify what you mean by "the manual" please?

______________________________________________

Hello Lawrence,

I will have a read of that, thank you very much.

[kalee20]

Ah. By 'manual' I meant the text you quoted from yourself in Post #1, ie the application note.

I don't have access to anything else, honest! Nor even the app note. Except background knowledge of how a reactance valve works, as outlined by Radio Wrangler above.

[ukcol]

Sorry Peter. I asked if you were referring to a different document because the application note is in terms of "negative inductance" and does not speak of a "negative capacitor". Have you made a typo or have I missed your point completely?

[Radio Wrangler]

Capacitors produce negative and imaginary values of reactance

Inductors produce positive and imaginary values of reactance.

In a tuned circuit, at resonance, the capacitor is producing the same amount of reactance as the inductor is, but of the opposite sign. Where the two components are in series, the two reactances cancel each other leaving zero.... in other words, a series tuned circuit looks like a short circuit at resonance.

You could say that capacitors are sorts of negative inductors, and inductors are sorts of negative capacitors. But be careful, it isn't as easy as reversing the sign on the number of picofarads, you have to equate the reactances. And saying that X picofarads is like -Y henries is only true at a spot frequency, move a bit in frequency and the equivalent values shift.

David

[kalee20]

Originally Posted by ukcol

I asked if you were referring to a different document because the application note is in terms of "negative inductance" and does not speak of a "negative capacitor". Have you made a typo or have I missed your point completely?

I think we're suffering from phase shift!

Post #2 was typed in a hurry, by that I meant I had not thought about everything, and I took the app note quote as gospel and factual, which I later came to doubt.

The app note talks about the circuit behaving as a negative inductance, this is with a capacitor between g1 and k. If this was indeed correct, then putting an inductor between g1 and k would give the effect of a negative capacitance. But I now reckon the app note is wrong; it's just a straight (positive) inductance.

The way a reactance valve works is as follows:

1) the anode current is virtually independent of anode voltage, for a pentode, so its impedance is virtually infinite.

2) There is a capacitor between g1 and k.

3) There is a resistor between a and g1, chosen so that its resistance is much larger than the reactance of the capacitor g1-k. (There will be a DC blocking capacitor as well, but ignore this).

If you wiggle the anode voltage up and down, then the current in the a-g1 resistor will vary in sympathy because one end is connected to anode, and the other to the g1-k capacitor which has a small reactance. The wiggling of the anode voltage will itself make no difference to anode current.

The current through the resistor will cause a (small) voltage to appear g1-k, but it will be 90 degrees out of phase because the voltage across the capacitor is out of phase with the current through it. So the grid voltage wiggles are 90 degrees out of phase with the anode voltage wiggles.

The anode current through the valve will now be forced to follow the grid voltage. So the anode current variation will be 90 degrees out of phase with the anode voltage variation. And the magnitude of the current variation will depend on the gm of the valve.

But that's exactly what you would expect from an inductor, a current which is 90 degrees out of phase with voltage. (A capacitor also has phase shift, but it's in the other direction).

So whatever is connected to the anode, and imposing voltage wiggles, will 'think' it's connected to an inductor. The value of inductance is dependent on the gm of the valve, so can be controlled.

It will behave as an inductor, because as frequency rises, the grid voltage will fall, so the anode current variation will fall, exactly as an inductor does (current falls with frequency). And the phase shift, V vs I, is exactly right for an inductor - not a negative inductor. App note wrong!

If you had a small inductor between g1 and k (which is how we thought would be on LW till you proved the switch operation was wrong), then the circuit would behave as a capacitor.

[ms660]

Another reference on reactance valves in 29.2 (iii) here:

http://frank.yueksel.org/other/RCA/R...cy-Control.pdf

Lawrence.

[Radio Wrangler]

That's a much better way of putting it than I did, Kalee.

I went for the general case, but the specific description ought to be found much clearer.

David