UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > General Vintage Technology > Components and Circuits

Notices

Components and Circuits For discussions about component types, alternatives and availability, circuit configurations and modifications etc. Discussions here should be of a general nature and not about specific sets.

Closed Thread
 
Thread Tools
Old 30th Oct 2017, 12:58 pm   #21
G8HQP Dave
Dekatron
 
Join Date: Sep 2008
Location: Solihull, West Midlands, UK.
Posts: 4,339
Default Re: Interesting Armstrong AM detector circuit.

Yes, it is strange that the envelope detector is so poorly handled by most textbooks. I have lots of radio books, and so far I have only found one which gives the formula for the ideal voltage-fed envelope detector. Both Terman and Langford-Smith (the two 'bibles' of radio design) move too quickly to non-ideal detectors without properly dealing with the ideal case. One of my hobby-horses is that you cannot properly understand the real thing until you have first understood the ideal thing.

The one exception I have found so far is 'Radio Communication' by D C Green. He derives a formula giving the upper limit for the time constant:
CR <= sqrt(1-m^2)/(m 2 pi fm)
where C and R are the envelope detector components, m is the AM modulation index, fm is the modulation frequency. This shows that distortionless demodulation is not possible for 100% (m=1) AM, because the CR upper limit for this would be zero but there is also a lower limit imposed by the carrier frequency. Any external loading on the output makes things worse.

For 50% modulation the ideal envelope detector starts to distort for modulation frequency above 3/(4 pi RC) = 0.24/RC. 90% would be 0.033/RC - so for normal AM modulation and IF frequencies the window between the upper and lower limits for CR is quite narrow. For good IF filtering we need fc >> 1/(2 pi RC) - say, fc > 1/(RC). 470kHz IF then requires RC > 2.1us. 5kHz modulation at 90% requires RC < 6.7us. In practice it seems that many sets use a high RC value and accept some treble distortion - the frequency chacteristics of music and speech come to our aid so full modulation is unlikely above a few kHz. Modern broadcast audio compression may not help!
G8HQP Dave is offline  
Old 31st Oct 2017, 12:34 am   #22
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,625
Default Re: Interesting Armstrong AM detector circuit.

G8HQP Dave,

Yes I agree. Due to the fact the RC discharge profile is an exponential, as all the equations indicate it is impossible to demodulate a 100% modulated carrier without distortion.

When I was playing around with AM detectors some years ago I built one where the capacitor was charged and discharged from a split power supply with a current source so the discharge curve approached zero as a straight line (and could cross it). With that I was able to demodulate a close to 100% modulated wave without negative peak clipping of the high frequency modulation content.

Also for a lot of detectors they try to save on parts. With a 455KHz IF for example with the detector time constant short enough for a close to HiFi response, there are serrations from the 455KHz on the signal, and it really needs another RC filter after that. The tendency for most manufacturers was to just hammer that with a bigger detector RC time constant.It was a reasonably big problem with a 262kHz IF's in some early radios for audio.

I can't attach it right now but I built a small detector module to replace the diode detector in one of my Eddystone EC10 Comms radios, so it will work for very low signals without square law distortion, I could attach it when I get home. It converts the IF voltage into a current and uses that to drive a pair of balanced diodes, and it's transfer function is much better than a 1N60 diode for low level signals but it's still relatively resistant to overload. It just has one transistor and a transformer.

One beauty about diode detectors, despite the issues we have been talking about, they operate over a very wide dynamic range and are very resistant to overload. I have built a number of OP amp precision detectors but in this respect they are poor.

Another interesting detector, for very small signals is the Tunnel diode (used in reverse) and in this mode it's called a "Back Diode", They describe this in Howorwitz & Hill, I have made a few of these before.

Last edited by Argus25; 31st Oct 2017 at 12:45 am. Reason: Irrelevant info removed
Argus25 is offline  
Old 31st Oct 2017, 2:35 pm   #23
G8HQP Dave
Dekatron
 
Join Date: Sep 2008
Location: Solihull, West Midlands, UK.
Posts: 4,339
Default Re: Interesting Armstrong AM detector circuit.

The negative peak problem can be reduced (possibly eliminated) by adding some extra current to drag the capacitor down, as you say. This current can also serve to pull a semiconductor diode onto its knee at around 0.6V, yet some designs achieve this by sending the current into the 'input' side of the diode - thus making the negative peak problem worse instead of better! A thermionic diode has the opposite problem: it starts conducting just below forward voltage so this trick does not work.
G8HQP Dave is offline  
Old 1st Dec 2017, 4:50 pm   #24
TrevorG3VLF
Octode
 
Join Date: Apr 2011
Location: Matlock, Derbyshire, UK.
Posts: 1,075
Default Re: Interesting Armstrong AM detector circuit.

Quote:
Originally Posted by Argus25 View Post
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.
Please tell me where this can be found. I have an early copy of "The art of electronics" and also the third edition.

Thank you
TrevorG3VLF is offline  
Old 1st Dec 2017, 9:17 pm   #25
ukcol
Dekatron
 
ukcol's Avatar
 
Join Date: May 2003
Location: Harlaxton, Lincolnshire, UK.
Posts: 3,793
Default Re: Interesting Armstrong AM detector circuit.

I have a copy of The Art of Electronics second edition. It dosn't appear to be in there either.
__________________
Best Regards,
Colin McLaughlin
BVWS Member
ukcol is offline  
Old 1st Dec 2017, 11:48 pm   #26
PJL
Dekatron
 
Join Date: Sep 2005
Location: Seaford, East Sussex, UK.
Posts: 3,776
Default Re: Interesting Armstrong AM detector circuit.

Isn't this just a transformerless IF amplifier with the aim of reducing the signal on the previous AGC controlled IF stage so reducing distortion for high signal levels? The only other benefit I can see is that the loading on the previous IF transformer should be reduced.
PJL is online now  
Old 2nd Dec 2017, 12:26 am   #27
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,071
Default Re: Interesting Armstrong AM detector circuit.

Indeed. For example, see posts #3 and 6.

And the attachment describes it in Armstrong's own words:


Cheers,
Attached Thumbnails
Click image for larger version

Name:	Stereo-Twelve p.02 AM Circuit.jpg
Views:	22
Size:	107.2 KB
ID:	153224  
Synchrodyne is offline  
Old 2nd Dec 2017, 3:10 am   #28
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,071
Default Re: Interesting Armstrong AM detector circuit.

An example of a current-driven diode demodulator may be found in the envelope demodulator part of the Brook outboard synchronous demodulator unit described in Wireless World 1989 September, page 856ff.

This was described as having a low-distortion, with the diode driven by a grounded-base transistor at constant current. Here is the circuit:

Click image for larger version

Name:	Brook Envelope Demodulator.jpg
Views:	30
Size:	51.3 KB
ID:	153232

Brook quoted a maximum THD of 0.6% for both envelope and synchronous demodulation modes over the frequency range 20 Hz to 7 kHz.

Whilst this type of demodulator was evidently chosen for its low distortion (say as compared with a conventional diode circuit), it does seem to be an interesting choice for the application. The envelope demodulator was required for use during initial tuning, before switching to the synchronous mode. As well as having selectable sidebands in the synchronous mode, there was also a “window” facility in which the receiver’s passband could be offset to minimize interference, in which case the outboard demodulator was receiving an asymmetric sideband signal. The optimum offset in any situation would be found during tuning, when the envelope demodulator was in use. But even perfect diodes distort when presented with a corrupted AM signal, such as one with sideband asymmetry. So, for example a quasi-synchronous demodulator with wideband reference channel might have been preferable for the “envelope” mode, given that it produces much less distortion when operating with asymmetric sidebands.

Certainly, Hershberger used a quasi-synchronous envelope demodulator in his broadly similar outboard synchronous demodulation unit described in Popular Electronics for 1982 April, page 61ff. Here is the block schematic:

Click image for larger version

Name:	Hershberger Demodulator Block Schematic.jpg
Views:	25
Size:	46.6 KB
ID:	153233

Hershberger claimed that this form of envelope demodulator got around the diode limitations of diagonal clipping and diode threshold distortion. He also provided a diagram showing the difference between synchronous and envelope demodulators when presented with less-than-ideal AM signals:

Click image for larger version

Name:	Hershberger AM Corruption Effects.jpg
Views:	18
Size:	78.8 KB
ID:	153234

In context, envelope demodulator in this diagram is taken to refer to the rectifying types.

Also, Sherwood used a quasi-synchronous demodulator for “envelope” demodulation in SE3 outboard synchronous demodulator unit. This relied upon offset rather than phasing and matrixing for sideband selection, so I imagine that had it used a diode demodulator, it would have sounded very distorted during the setup and offset selection. And Drake had used a quasi-synchronous AM demodulator in its R7 HF receiver, I imagine in connection with its claim that the R7 could provide a 3 kHz AM audio bandwidth with a 4 kHz IF filter, which clearly required significant offsetting and so asymmetric sidebands at the demodulator.

But these examples were from the solid-state era, where from c.1970, active device count was not much of an issue except perhaps for battery-operated equipment. Back in the valve era, more sophisticated forms of AM demodulator were probably too complex, too costly and required too many valves for consumer equipment in general, although used professionally, hence the efforts put into getting better results from the diode and its relatives. In that regard, I wonder if constant-current drive might have been obtained by using a grounded grid triode ahead of the diode. A double diode-triode (DDT) valve of the 6AJ8 type, with all three units separate, might have worked in this role, with one diode as demodulator and the second used to provide AGC delay. Whilst such DDTs were plentiful in the American valve range, they seem to have been unknown in the European range.

Click image for larger version

Name:	6BJ8 Pinouts.PNG
Views:	17
Size:	31.5 KB
ID:	153235


Cheers,
Synchrodyne is offline  
Old 2nd Dec 2017, 8:08 am   #29
Radio Wrangler
Dekatron
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 10,540
Default Re: Interesting Armstrong AM detector circuit.

Quote:
Originally Posted by TrevorG3VLF View Post
Quote:
Originally Posted by Argus25 View Post
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.
Please tell me where this can be found. I have an early copy of "The art of electronics" and also the third edition.

Thank you
The Hewlett-Packard Journal is available on the web. Find the issue with the HP 8901A modulation analyzer article in it (for search engine purposes, mis-spell analyser with a zee)

This instrument is a very high performance receiver for measuring modulation depth and deviation. They get used for calibrating sig gens. The HPJ article has a section on its current-driven diode AM detector which makes rather good reading.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 2nd Dec 2017, 9:38 am   #30
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,625
Default Re: Interesting Armstrong AM detector circuit.

Quote:
Originally Posted by Radio Wrangler View Post
Please tell me where this can be found.

It is in my second edition of H & H it is on page 890, figure 13.30, ( cited proc. IEE,122,3 249, 1975). And the description of it is a paragraph on page 889.

If you don't have this in your edition, let me know and I will scan the relevant paragraph and the diagram. They say the circuit will operate to 100MHz or more and it solves the problem of detector non-linearity.

I mentioned this circuit before on a thread about how to make a low level RF milli-volt meter.

Hugo.
Argus25 is offline  
Old 2nd Dec 2017, 11:48 am   #31
ukcol
Dekatron
 
ukcol's Avatar
 
Join Date: May 2003
Location: Harlaxton, Lincolnshire, UK.
Posts: 3,793
Default Re: Interesting Armstrong AM detector circuit.

Quote:
Originally Posted by ukcol View Post
I have a copy of The Art of Electronics second edition. It dosn't appear to be in there either.
Well, I got that spectacularly wrong.

Perhaps this scan will make up for it.
Attached Files
File Type: pdf Current-source drive detection..pdf (685.0 KB, 38 views)
__________________
Best Regards,
Colin McLaughlin
BVWS Member
ukcol is offline  
Old 2nd Dec 2017, 1:35 pm   #32
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,625
Default Re: Interesting Armstrong AM detector circuit.

This was the circuit that gave me the idea for the Supadetector circuit I used in one of the radios I posted in the homebrew section.

There wasn't the power supply available in the radio to do it the H&H way but I considered that dynamically an inductor as a collector load tends to behave as a current source. So I simply converted the IF signal voltage into a current with the transistor and used the inductor to drive the diodes. The diodes are driven into conduction at very low signal voltages, but due to the loading the frequency response and linearity is good.

The H&H circuit of course is the answer to a sensitive and linear wide band RF millivolt meter. To make one I think it merely requires a broadband RF amplifier/buffer & attenuator at the input and this circuit, plus an OP amp and meter bridge.
Argus25 is offline  
Old 4th Dec 2017, 7:44 am   #33
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,071
Default Re: Interesting Armstrong AM detector circuit.

Ukcol, thanks for posting that Horowitz and Hill excerpt.

I notice the H&H example current-driven demodulator had two diodes in a voltage-doubling mode. The voltage-doubling demodulator was in and of itself claimed to have lower distortion than a conventional diode, as for example shown in this article from Radio-Electronics for 1959 October:

Click image for larger version

Name:	RE 195910 p.58 AM Demodulator.jpg
Views:	15
Size:	90.7 KB
ID:	153323 RE 195910 p.58 AM Demodulator.pdf


Heathkit used the voltage-doubling AM demodulator in quite a few of its hi-fi tuner models, and also in some communications equipment.


Cheers,
Synchrodyne is offline  
Old 4th Dec 2017, 2:04 pm   #34
ukcol
Dekatron
 
ukcol's Avatar
 
Join Date: May 2003
Location: Harlaxton, Lincolnshire, UK.
Posts: 3,793
Default Re: Interesting Armstrong AM detector circuit.

Unless I haven't fully understood the explanation of Geiler's voltage doubler detector in the article above, this circuit will still suffer from distortion caused by the R.C. time constant of the load.

I expect that the performance of the British 405 television sound AM detector was much better than that of broadcast radio receivers because the high IF frequency would have allowed a much lower value capacitor in the diode load.

No doubt the detector performance in a radio could be improved by up-converting the I.F. to a much higher frequency; a rather expensive solution however.
__________________
Best Regards,
Colin McLaughlin
BVWS Member
ukcol is offline  
Old 4th Dec 2017, 2:56 pm   #35
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,625
Default Re: Interesting Armstrong AM detector circuit.

Quote:
Originally Posted by Synchrodyne View Post
I notice the H&H example current-driven demodulator had two diodes in a voltage-doubling mode.
No they are not.

The deceptive part of the H&H circuit, I've mentioned this on another thread, is that it looks exactly like a voltage doubler but it isn't and it doesn't double voltage.

Notice the the very low value of resistor that these diodes ultimately load into. It is actually, for practical purposes, a "shorted out voltage doubler" the output resistor 100R load is doing the "shorting" and is actually acting as a current sense resistor on the output.

This is why it is not like any other type of rectifier-detector circuit you will see, because its operating principle is entirely different.

The diodes are current driven (not voltage driven) into a low R load. This is why not only is the linearity excellent but it goes to an extremely low input voltage levels. As I mentioned before, its the perfect solution to the problem of an RF volt meter, but it appears ill understood or overlooked.
Argus25 is offline  
Old 5th Dec 2017, 7:32 am   #36
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,071
Default Re: Interesting Armstrong AM detector circuit.

Argus25: – Thanks for the correction. Appearances can be deceptive!

Ukcol: - I am inclined to think that the Geisler article was somewhat overstated as to the benefits of the voltage-doubling AM demodulator. Perhaps there is some small gain as compared with the simple diode. Heathkit claimed lower distortion for it, but did not quantify or explain further.

As you say, TV AM sound does appear to be an easier case, with RF in the 30 MHz range. That is true even if one considers the extended “audio” bandwidth needed for proper operation of rate-ot-rise noise limiting diodes, say around 100 kHz. Mostly I think conventional diode demodulators were used for TV AM sound in the valve and discrete semiconductor eras. In the IC era, for French Systems E/L and then L/L’, quasi-synchronous demodulators, both with and without carrier tank circuits, generally took over. It would appear that the with-tank-circuit type simply followed established vision demodulator practice, and that the without-tank-circuit type was developed to cover multi-system receivers in which L and L’ had different sound IFs. Nevertheless, National developed an IC for TV AM sound that used a rectifying demodulator, whose exact nature I hesitate to categorize:

Click image for larger version

Name:	National TV AM Sound IC.jpg
Views:	21
Size:	39.6 KB
ID:	153398



Cheers,
Synchrodyne is offline  
Old 5th Dec 2017, 11:01 am   #37
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 4,517
Default Re: Interesting Armstrong AM detector circuit.

Quote:
Originally Posted by Argus25 View Post
The deceptive part of the H&H circuit, I've mentioned this on another thread, is that it looks exactly like a voltage doubler but it isn't and it doesn't double voltage.

Notice the the very low value of resistor that these diodes ultimately load into. It is actually, for practical purposes, a "shorted out voltage doubler" the output resistor 100R load is doing the "shorting" and is actually acting as a current sense resistor on the output.
It's good! It's a charge pump, and as the input voltage to the diode bit is not defined, it can't be called a voltage doubler.

If the signal source input has high enough compliance, the output burden resistor could be raised in value to get more voltage output. As long as the input current isn't affected by the higher voltage drop across the diodes+burden, the good performance will be retained. The limit comes when the reverse leakage of the diodes starts to become significant at the consequential higher reverse voltage, or the current drive starts to be affected by the higher voltage.

An even more elegant output sense might be to use an op-amp based current-to-voltage converter, with the diode feeding into the virtual earth node. Then, both diodes would see the same operating conditions; reverse voltage would be limited to the forward voltage drop of the 'other' diode, each half-cycle. And the current source would need only to drive into a diode-drop's worth of voltage.

It certainly would make the basis for a good RF millivoltmeter.
Attached Thumbnails
Click image for larger version

Name:	image.jpg
Views:	23
Size:	44.9 KB
ID:	153410  

Last edited by kalee20; 5th Dec 2017 at 11:15 am. Reason: Add sketch
kalee20 is offline  
Old 5th Dec 2017, 11:19 am   #38
Argus25
Octode
 
Argus25's Avatar
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 1,625
Default Re: Interesting Armstrong AM detector circuit.

Quote:
Originally Posted by kalee20 View Post
An even more elegant output sense might be to use an op-amp based current-to-voltage converter, with the diode feeding into the virtual earth node.
Marvelous idea ! I love it.
Argus25 is offline  
Old 5th Dec 2017, 12:32 pm   #39
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 4,517
Default Re: Interesting Armstrong AM detector circuit.

And you could use silicon PN diodes, such as the fast but cheap-as-chips 1N4148, secure in the knowledge that the forward voltage drop is immaterial. No need for more expensive Schottky or Germanium point-contact!
kalee20 is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 2:29 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2018, vBulletin Solutions, Inc.
Copyright ©2002 - 2018, Paul Stenning.