UK Vintage Radio Repair and Restoration Discussion Forum

UK Vintage Radio Repair and Restoration Discussion Forum (https://www.vintage-radio.net/forum/index.php)
-   Components and Circuits (https://www.vintage-radio.net/forum/forumdisplay.php?f=40)
-   -   Interesting Armstrong AM detector circuit. (https://www.vintage-radio.net/forum/showthread.php?t=140848)

ukcol 27th Oct 2017 2:00 pm

Interesting Armstrong AM detector circuit.
 
1 Attachment(s)
I have been looking at the circuit of the Armstrong "London Stereo Twelve MKII." This is an AM/FM stereo receiver chassis that uses no less than three ECC85s.

One ECC85 is used for the usual RF amplifier and mixer functions in the VHF tuner. One is used in the stereo audio pre-amplifier. Half on the last ECC85 is used in the VHF tuner providing AFC.

The remaining half of the last ECC85 is used in the AM detector. A hand drawn copy of the circuit is attached below. The triode is used as a buffer RF amplifier to feed the diode type AM detector.

I assume, in my ignorance, that this arrangement reduces distortion by providing a larger voltage swing to the input of the detector diode and that it has a better performance at high modulation depths. What does the team think?


Details of this chassis, including the circuit, are available on the Radiomuseum web site.

G6Tanuki 27th Oct 2017 6:11 pm

Re: Interesting Armstrong AM detector circuit.
 
I guess having an extra bit of gain would present a higher voltage-swing to the [semiconductor] diode - but to be honest the poor-linearity 'threshold knee' distortion effect in diode detectors is only ever a problem when you're feeding them with really low-level signals - tens/hundreds of millivolts, not the tens of volts which I'd expect a competent superhet receiver to provide!

Synchrodyne 27th Oct 2017 6:29 pm

Re: Interesting Armstrong AM detector circuit.
 
My recollection is that Armstrong did it that way in its ST3 tuner and related equipment in order to minimize modulation rise distortion, which could occur in final IF stages to which full AGC voltage was applied when operating with large signal inputs. A non-AGC’d final IF stage relieved the burden on the AGC’d IF stage, so that it no longer found itself chasing its own tail, being required to deliver a large signal to the demodulator and AGC rectifier whilst back-biased to the very bent part of its characteristic. I suspect that it might have come about not as an ab initio idea, but following a search for a productive way to utilize a spare half-ECC85.

Somewhere I might have a reference to this circuit feature – I’ll look and revert.


Cheers,

ms660 27th Oct 2017 6:32 pm

Re: Interesting Armstrong AM detector circuit.
 
To maintain bandwidth shaped by previous IFT's without going to any extra expense/bother of neutralizing the triode if a final IFT was employed in the anode circuit?

Lawrence.

G6Tanuki 27th Oct 2017 6:37 pm

Re: Interesting Armstrong AM detector circuit.
 
Quote:

Originally Posted by Synchrodyne (Post 986442)
My recollection is that Armstrong did it that way in its ST3 tuner and related equipment in order to minimize modulation rise distortion, which could occur in final IF stages to which full AGC voltage was applied when operating with large signal inputs. A non-AGC’d final IF stage relieved the burden on the AGC’d IF stage

That makes sense: something similar was done in quite a few 'communications' radios from the AM-era that had multiple IF amps. The last IF-amp was not subjected to AGC action [though it was invariably included - by way of sharing a common cathode-resistor pot with the other IF stages - in the manual IF-gain-control schema].

This was sometimes nicknamed a "virtual AGC amplifier" since the final pre-detector non-AGC-controlled IF-stage was within the AGC-loop and helped provide additional overall loop-gain. It became a real pain-in-the-proverbial when you wanted to provide any kind of AGC action on CW or SSB signals though.... and even worse if you needed a noise-blanker[1] the smart designers split the signal and used a separate parallel AGC gain/detector-strip and suitable time-constants.

[1] Narrow-band IF stages tend to 'ring' and so lengthen clipped noise-pulses, making them harder to subsequently eliminate from the output. So many receivers split the IF signal and fed some through wideband unlikely-to-ring-or-saturate stages before detecting them and using the peak-level signal to 'clamp' the audio that had made it through the narrow-band signal-stages.

Synchrodyne 27th Oct 2017 6:45 pm

Re: Interesting Armstrong AM detector circuit.
 
2 Attachment(s)
Here are a couple of references to the Armstrong circuit:


Cheers,

G8HQP Dave 27th Oct 2017 7:30 pm

Re: Interesting Armstrong AM detector circuit.
 
Whether the extra gain stage boosts the signal sent to the envelope detector depends more on the AGC arrangements than the gain of the extra stage. Having an uncontrolled stage last avoids the problem of the last stage being backed off so much by AGC that it can no longer provide enough signal to generate that AGC.

ukcol 27th Oct 2017 8:39 pm

Re: Interesting Armstrong AM detector circuit.
 
Quote:

Originally Posted by ukcol (Post 986395)
.... in my ignorance....

Please help me out with concise explanation of the term "modulation rise distortion" and how this circuit gives an improved performance with respect to same.

Synchrodyne 28th Oct 2017 6:07 am

Re: Interesting Armstrong AM detector circuit.
 
When a modulated signal is passed through an amplifier with a curved input-output characteristic, that curvature will add harmonic distortion to the modulation. This addition to the modulation will have the effect of increasing the modulation depth. The increase in modulation depth, known as “modulation rise”, is a convenient way of quantifying the added distortion.

Generally, modulation rise distortion is not a major issue, but it can become one when a final (or sole) IF amplifier stage is in receipt of a relatively large signal, and it has a very curved transfer characteristic. Thus, the signal swing traverse will cover a range with a significant change in slope. The very curved transfer characteristic is representative of a valve that is back-biased by a large AGC bias to be operating at the tail end of its curve.

One solution, as used by Armstrong, is to insert a non-AGC’d gain stage between this IF amplifier and the demodulator/AGC rectifier. As this will be operating in the centre, relatively straight part of its characteristic, it will not suffer from modulation rise. Then as, for a given signal level into the demodulator/AGC rectifier, the main IF amplifier is dealing with a lower signal input level (reduced by whatever gain is provided by the non-AGC’d stage, it will also be operating clear of the modulation rise region.


Cheers,

ukcol 28th Oct 2017 9:00 am

Re: Interesting Armstrong AM detector circuit.
 
Thank you very much for that excellent explanation Synchrodyne, thats another hole in my knowledge base filled :)

G8HQP Dave 28th Oct 2017 12:34 pm

Re: Interesting Armstrong AM detector circuit.
 
Many amplifying devices compress signals. Devices with an exponential signal response (such as remote cutoff valves and BJTs) do the opposite: they expand signals. It basically all comes down to the sign of the third-order term in the series expansion of the in/out function. The reason why an exponential response is (approximately) designed-in for a remote cutoff valve is that this response means that signal handling does not degrade as gain is reduced (unlike a 'straight' valve).

Argus25 28th Oct 2017 1:21 pm

Re: Interesting Armstrong AM detector circuit.
 
My take on this circuit is that it is an attempt at a current source driver. Before I'l explain it I could mention that there are two basic types of problems that cause non linearity or distortion in AM detectors which use diodes. (These problems can be cured using op amp precision detectors, but that is outside the scope of this response)

Firstly there is the problem (for a detector with a positive output polarity) of negative peak clipping that always occurs when the modulation reaches some frequency threshold. It occurs because of the RC time constant that the filter the detector feeds, has an exponential discharge profile. The frequency at the point this distortion starts to appear is approximately 0.375/RC when the modulation is around 50%. It's a lower frequency when the modulation is higher. Basically the detector diode gets reverse biased due to the charge stored on the filter capacitor on the negative modulation peaks and it's not able to track the modulation envelope.

Secondly , the distortion caused with low input level to the diode detector that affects all modulation frequencies when the signal level to the detector falls too low. The detector, for low level signals has an approximate square law property, or close, rather than a linear transfer function and also for very low level signals fails to detect at all.

The cure for this problem is to drive the detector with a current source rather than a voltage source. This is elegantly explained by Horowitz & Hill, with an example circuit which converts the incoming modulated RF voltage to a current source to drive the detector diodes.

This works because in the case of a circuit attempting to provide a current drive, the voltage rises as high as it needs to develop conduction in the diode even with an extremely small RF input signal voltage. But there are other ways to do this, one way is to source the current from an inductor or transformer when the voltage has been stepped up prior to the diode, this apparrent mismatch achieves the same effect.

In the case of the circuit provided by the OP, it could be regarded as a circuit that converts the modulated RF into a plate current. From the voltage perspective the plate current is converted to a voltage by the 10k anode resistance, and the higher voltage swing at the plate ensures that the detector diode current remains high even for very low level signals out of the RF transformer. In addition this configuration puts little load on the IF transformer's secondary, helping to maintain the gain and selectivity. So I would expect this type of circuit to be better than most in detecting low level signals with little distortion.

ukcol 28th Oct 2017 3:12 pm

Re: Interesting Armstrong AM detector circuit.
 
Quote:

Originally Posted by Argus25 (Post 986634)
Firstly there is the problem (for a detector with a positive output polarity) of negative peak clipping that always occurs when the modulation reaches some frequency threshold. It occurs because of the RC time constant that the filter the detector feeds, has an exponential discharge profile. The frequency at the point this distortion starts to appear is approximately 0.375/RC when the modulation is around 50%. It's a lower frequency when the modulation is higher. Basically the detector diode gets reverse biased due to the charge stored on the filter capacitor on the negative modulation peaks and it's not able to track the modulation envelope.


Thanks for that gem Argus25. I have often wondered why changing the time constant of the components in the (effective) diode load can have an effect on distortion figures.

G8HQP Dave 28th Oct 2017 8:11 pm

Re: Interesting Armstrong AM detector circuit.
 
Quote:

Originally Posted by Argus25
In the case of the circuit provided by the OP, it could be regarded as a circuit that converts the modulated RF into a plate current.

The output impedance of this circuit is probably lower than an IFT, so it would look more like a voltage source than an IFT. I think it works by providing a little non-AGCed voltage gain, and allowing the final IFT to work into a highish impedance.

Argus25 28th Oct 2017 10:28 pm

Re: Interesting Armstrong AM detector circuit.
 
Yes I agree it's just a matter of proportions. But I would guess even for small signals out of the IF transformer the voltage swing at the plate would be much higher (well over a few hundred millivolts) than the value required to operate the detector diode in its linear region and the AC component of the plate current drives the detector diode via the coupling capacitor.

It might have been even better with an RFC as the plate load rather than the 10k resistor, then it would have much better approximated a current source drive as the inductance as a load tends to resemble a current source as a load.

In this circuit it's very important that the coupling capacitor has no leakage.

Argus25 29th Oct 2017 12:22 am

Re: Interesting Armstrong AM detector circuit.
 
Ukol,

That formula should have read 0.275/RC. That is if one goes by remarks in Terman about the AC impedance and DC resistance of the detector. I calculated it two other ways and got between 0.5/RC and 0.3/RC, so it's in this vicinity.

Right now I'm on an iPad having some difficulty copying and pasting addresses. If you go to one of my websites which is just a repository for articles in PDFs format,

www.worldphaco.com

and hunt around on the right side of the page for the article on Radio Grid Leak Detectors in the 1920's , it is explained in there how the formula was calculated. Then that link could replace the general website link above in this post.

I investigated it because I was curious about why grid leak resistors in the 1920's had colourful marketing claims for different tones etc.

ukcol 29th Oct 2017 9:56 am

Re: Interesting Armstrong AM detector circuit.
 
Thank you for that Hugo, I have downloaded the article and will have a read.

I have read at least one of your articles before (on the HMV 904) but I've not visited your electronics page until now; there looks like there is a wealth of goodies there and so I have bookmarked your page for future reading.

ukcol 29th Oct 2017 1:45 pm

Re: Interesting Armstrong AM detector circuit.
 
Excellent.

Having read the above article I feel I have, at last, a good grasp of how the diode load time constant effects distortion will respect to modulation depth and frequency.

It is a rare "teacher" that has both a through understanding of his (or her) subject and can also put that subject across in a way the student can understand. This involves starting at the students level and by small increments working towards the goal.

Argus25 29th Oct 2017 11:00 pm

Re: Interesting Armstrong AM detector circuit.
 
Colin,

For the most part I have found that the standard text books cover topics pretty well and I have a good collection of books from the UK & USA. However over the years I came across really only three topics that were not covered to my satisfaction;

1) Time constants of AM detectors
2) Energy recovery or efficiency diode circuits for TV
3) THe frequency response of interstage audio transformers in vintage valve radios.

So I did at one point an article on each of these topics, if for nothing more than to help myself understand them better.

If you look on the website you will see an article on the Evolution of the Damper Diode, it explains how we ended up with the common configuration line output transformer design, which is certainly not an obvious thing , the steps to get there.

The transformer analysis and the equation for the audio interstage transformer frequency response is hiding inside the article on the Grebe MU-1 radio, this equation is not available eslewhere because generally the modelling is not done for a transformer with an unloaded secondary.

Hugo.

ukcol 30th Oct 2017 10:29 am

Re: Interesting Armstrong AM detector circuit.
 
Thanks again Hugo,

I have a much better understanding of efficiency diode circuits than I did of AM detectors; as to audio transformers theory, that is a very weak area of my knowledge.

I shall have a read of both those articles. No doubt I will find I didn't know as much about the efficiency diode as I thought I did and I will definitely benefit from the transformers article.


All times are GMT +1. The time now is 8:09 pm.

Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.