"Power Grid" A.M. detector.

[ukcol]

Here is another interest AM detector circuit as used in the Marconi 272 (from 1933) and associated models. The circuit is show below.

When the set is switched to the "gram" input, V3 is used as the audio voltage amplifier but I have omitted the circuitry associated with the "gram" function in order to concentrate on the detector function.

Trader sheet 518 describes this circuit as a "power grid detector" but offers no information on how it works. My simplistic understanding of how it works is that the zero grid bias on V3 forces the grid cathode to function as a diode and the resulting anode current then follows one half of the applied RF wave. The signal developed across the anode load is then passed through an RF filter and some tone correction before being applied to the output valve's control grid.

I have never heard the term "power grid detector" before but perhaps this circuit is known more commonly under a different name?

[G8HQP Dave]

Your explanation is essentially correct. You could regard it as a peak clamp circuit directly coupled to a zero bias triode. It would impose only a light load on the tuned circuit, but at the cost of amplifying the IF so potentially causing problems if the amplified IF strays too far away from this stage. I suspect it will have the usual envelope detector problems of distortion on negative peaks.

[ukcol]

Thanks Dave

[Argus25]

I think it is basically a "grid leak detector" much like the type of thing seen in a 1920's valve radio with a 201A, but much more efficient due to the indirectly heated valve which probably has a closer cathode to grid spacing. It is probably called power grid detector to distinguish it from a power plate detector (anode bend) type. Due to the grid-cathode rectification with increasing signal, it cannot be an anode bend detector. C7 and R7 set the detector time constant and carrier serrations are filtered off in the plate circuit which contain both the audio modulation and IF carrier.

[Marconi_MPT4]

Interesting, this circuit is not strictly a power grid detector with the circuit values giving the game away. Looks more cumulative grid type with R7 being 1M and C being 50pf.

A power grid detector would have a much bigger coupling capacitor value, say 200pF to 470pf and smaller R7 value of say about 100k to 330k. The preceding stage needs to provide sufficient drive due to the damping presented, which is a major disadvantage as is the Miller effect.

It was frequently used in high quality pre WWII receivers.

Rich

[Argus25]

The preceding stage needs to provide sufficient drive due to the damping presented, which is a major disadvantage as is the Miller effect.

Actually the Miller effect doesn't trouble this circuit, because the bulk of the IF carrier frequency is filtered off by the 2000pF capacitor from the plate to ground, so the feedback of carrier from the plate to the grid is minimal. That filter comprises the tube's plate resistance in series with the capacitor too, so the filtering is fairly effective.

If the 1 meg resistor was connected across the 50pF capacitor, it would be immediately recognized by most as a grid leak detector, but returning the resistor to ground instead, as they have done, makes no difference to the circuit operation, all that happens is that the capacitor's discharge current flow via the IF coil's secondary as well with the resistor grounded, which has zero effect on the circuit operation. You could confirm this by putting the 1M resistor across the capacitor instead. The small actual grid space charge current leaks to ground in both cases, in one case via the IF secondary coil.

[kalee20]

Originally Posted by ukcol

...the zero grid bias on V3 forces the grid cathode to function as a diode and the resulting anode current then follows one half of the applied RF wave. The signal developed across the anode load is then passed through an RF filter and some tone correction before being applied to the output valve's control grid.

I have never heard the term "power grid detector" before but perhaps this circuit is known more commonly under a different name?

Yes, the grid-cathode functions as a diode. The capacitor gets charged up to (nearly) the peak RF voltage, and very nearly holds that voltage for the remainder of the cycle. Each cycle just 'tops up' the charge in the capacitor.

The anode current should actually follow the whole RF wave, not just half of it. But, the current waveform is offset by an amount determined by the grid capacitor voltage, which depends on the peak RF voltage input.

As the amplitude of the RF rises and falls, the grid capacitor voltage follows this, taking bigger gulps of current as the amplitude rises, and as the amplitude falls the charge leaks away through the associated resistor, so the anode current offset follows the grid capacitor voltage. If you filter off the RF content, you're left with the relatively slowly rising and falling of the mean current, which therefore follows the modulation envelope.

This is really a grid-leak detector, as used in simple TRF sets. The term 'power grid' was used for low-μ valves, working with smaller values of grid-leak resistance and higher anode currents, to handle larger input signals without overloading, but really it's the same thing. F J Camm used the term.

The necessary presence of RF on the grid does mean that the anode current has a big RF content which has to be dumped AFTER it's gone through the valve. But it does have the advantage that the ACC load ratio on the 'diode' is exactly 1, so in theory the circuit can handle modulation right up to 100% without sudden onset of distortion.

[Argus25]

The circuit is the electrical equivalent of figure 4 and 8 in this article and the grid voltage is shown in figure 12 and the anode voltage in fig 13, but of course a lot of carrier is already filtered off at the anode by the 2000pF capacitor in your circuit, then more filtered with the anode choke and second 2000pF cap, just leaving the modulation content:

http://worldphaco.com/uploads/Radio_...he_1920_s..pdf

[ukcol]

Thank you so much gentlemen for taking the trouble to explain this circuit so concisely. I can see that my understanding of the circuit as expressed in post #1 was indeed simplistic but not completely wrong

I now have a much better understanding of the details of its function and that of other AM detectors as explained in other threads.

I spent most of my working life in the servicing trade; I always had a good understanding of circuits at least at the level required to be an effective service engineer, but it is really interesting to go into circuit function more deeply. I do do a lot of reading on the subject but sometimes, as is the case with this circuit, it can be difficult to find sources that cover what I need.

[ms660]

A five page article in W.W about the power grid detector:

W.W. Dec.1931, mag page 631&632, continued on mag page 672 to 674:

http://www.americanradiohistory.com/...ld-1931-12.pdf

Lawrence.

[Argus25]

Originally Posted by ms660

A five page article in W.W about the power grid detector:

W.W. Dec.1931, mag page 631&632, continued on mag page 672 to 674:

http://www.americanradiohistory.com/...ld-1931-12.pdf

Lawrence.

In my view this article really points out that grid leak detection and "power grid leak detection" are one and the same thing.

That is why the circuits are the same.

However, as noted, the terminology "grid leak detection" is largely applied to low signal levels and under these circumstances its permissible to use a higher value RC time constant for the detector, as current flows on both half cycles to discharge the capacitor.

However, with large signals, the bias point of the grid-cathode diode shifts and current flows only on one half cycle, so you need to have a smaller RC time constant, typically a smaller capacitor, to maintain the detectors high frequency response and relative freedom from distortion.

All it means is that you can get away with a longer RC time constant in a grid leak detector for lower level signals, while still maintaining some detector distortion value and you would call it a "grid leak detector" compared to another scenario where the signal levels were higher and then for the same detector distortion you need a shorter RC time constant, typically a lower C and then you call it a "power grid leak detector".

It is actually a bad case of someone trying to claim that two identical circuits should have a different name depending on the signal level, by claiming ideally a circuit component value would be different in the two cases.

[ukcol]

I'll read that article, thanks Lawrence.

[ukcol]

An excellent article, reinforcing the points made by contributors here, thanks Lawrence. Its interesting to learn, from the article, how the requirements for the anode characteristic curve in the "power detector" circuit rendered it unsuitable for battery sets.

[PJL]

Articles from this period are not always to be trusted. Do you really get 50uA grid current if you connect the grid of an AC/HL to the cathode? One paragraph talks about -ve grid bias created across the grid leak resistor the next paragraph it is positive.

The valves are specially manufactured to bias correctly at 0 grid volts. The signal gets rectified on the grid. Small signals are distorted due to the rectification not being ideal. Any 2-3 valve straight tuned set is going to have to have a reasonable sized signal on the detector. If you have a lower impedance driver you can afford to reduce the resistance and increase the capacitance which might reduce distortion.

Only interesting thing is the choice of anode resistor which is quite low resulting in less gain but that is probably down to the valve characteristics.

[kalee20]

Originally Posted by Argus25

...However, with large signals, the bias point of the grid-cathode diode shifts and current flows only on one half cycle, so you need to have a smaller RC time constant, typically a smaller capacitor, to maintain the detectors high frequency response and relative freedom from distortion.

I can't swallow this bit! A good diode detector (whether a proper diode, or the grid-cathode of a more-ode) should pass little gulps of current at the waveform peaks, and be cut-off as a diode the rest of the time.

So, in our detector, the grid is negative most of the time, and just hits a bit above 0V momentarily each RF cycle. The rest of the valve acts as an amplifier, and if this gets cut off, RF distortion will obviously occur.

Since the AF content is the mean of the anode current, any significant distortion of the RF envelope is likely to result in AF distortion. In fact, if you start hitting cut-off as an amplifier, you'll get a bit of anode-bend detection, which gives an AF output of opposite phase to grid-leak detector!

[ms660]

Originally Posted by PJL

Articles from this period are not always to be trusted. Do you really get 50uA grid current if you connect the grid of an AC/HL to the cathode?

Different author similar valve:

http://www.americanradiohistory.com/...search=%22grid current curve%22

Lawrence.

[Argus25]

Originally Posted by kalee20

In fact, if you start hitting cut-off as an amplifier, you'll get a bit of anode-bend detection, which gives an AF output of opposite phase to grid-leak detector!

I agree with this.

In the article cited they indicated the valve's grid current had a significant effect altering the operating point of the grid-cathode diode curve, but I did not check the value they cited, to see if it was realistic. Most valve's grid current (from the grid acquiring electrons) doesn't contribute greatly to generating a negative grid voltage unless the grid resistor is over 4.7M to 10Meg. Some valves are biased this way but its uncommon.

Certainly this is true:

For very low level signals that do not drive the grid-cathode into forward conduction, there is no grid leak detection at all as no asymmetry is produced in the grid waveform. And as you point out, all that happens is you get some anode bend detection which produces an anti phase audio signal at the plate. Then as the drive level comes up you get grid leak detection but any audio that appears initially is phase canceled by any anode bend detection.

As I said in my article on grid leak detectors something like ... "in that boundary between anode bend detection and grid leak detection, no detection can occur".


A lot of valve grid leak detectors would be substantially improved if a silicon signal diode was placed across the grid-cathode connection (anode of added diode to grid) that way you get much better clamping of the positive going carrier peaks to about 0.6V above ground and higher audio output at the plate.

[PJL]

Originally Posted by ms660

Different author similar valve:

http://www.americanradiohistory.com/...search=%22grid current curve%22

Lawrence.

Never tried but I am surprised it is so high. The effect of this is the valve will settle with a bias of -1V but the detection would still be the same. It will just change the anode current.