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

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Television Standards Converters, Modulators etc

Notices

Television Standards Converters, Modulators etc Standards converters, modulators anything else for providing signals to vintage televisions.

Closed Thread
 
Thread Tools
Old 18th Apr 2007, 11:48 pm   #1
Andy Green
Hexode
 
Join Date: Jan 2005
Location: Waterford, Republic of Ireland
Posts: 259
Default Vestigial sideband system 'A' modulator

Has anyone built or modified an existing modulator design for VSB operation, and can anyone shed any light on how best to do it?

I presume that it was done using cavity or lumped constant bandpass filters, possibly at an IF frequency in the transmitter?

I am in the process of ordering a precision return loss bridge for my work as a broadcast engineer (we have to use cavities on the output of al FM transmitters here in Ireland) and was thinking about building a VSB modulator for use on 70 cm / 433 MHx or 405 line amateur TV.
If anyone can shed some light on this, or can point me to any other leads it would be appreciated - there's no point in trying to reinvent the wheel!

Andy EI3HG
Andy Green is offline  
Old 19th Apr 2007, 6:50 am   #2
ppppenguin
Retired Dormant Member
 
ppppenguin's Avatar
 
Join Date: Dec 2003
Location: North London, UK.
Posts: 6,168
Default Re: Vestigial sideband system 'A' modulator

The only amateur built VSB Band I modulator I know of is the one built by Darius (Oldeurope in this forum). This was done with conventional lumped filters after a DSB modulator.

For our interests in vintage TV there is simply no need for VSB unless we hit the jackpot with a special events licence to radiate 405 line telly from Allly Pally.

I'm sure that some searching in the BATC's bulletin (CQ-TV) will yield a solution or two. I think these archives are available on CD at a fairly modest cost. http://www.batc.org.uk/

I doubt if there were any practicable IF solutions for the amateur. This would require a very linear PA.
ppppenguin is offline  
Old 19th Apr 2007, 2:24 pm   #3
Ray Cooper
Retired Dormant Member
 
Join Date: Jan 2007
Location: Porthmadog, Gwynedd, UK.
Posts: 199
Default Re: Vestigial sideband system 'A' modulator

Thoughts on VSB filters from memory (it's all such a long time ago...)

The old high-power Band I TXs at Sutton and the Moss were high-level modulated and produced DSB outputs. They needed external VSB filters to produce the product as stated on the packet.

I have an IEEE paper somewhere (Radio section 1118 by E.C. Cork) which describes in fulsome detail how the Sutton filter was put together. Anybody who wishes to do a reconstruction might care to note before starting that it was about 30 feet long and weighed 800 pounds, of mostly copper and brass. Basically it was a coaxial-line affair having two sections, a low-pass one for the wanted signal, and a high-pass one diverting the unwanted sideband into a water-cooled 5kW load. (Not that there was ever anything like that power level under normal picture conditions, though under test conditions there could be a couple of kW.) The whole affair was cunningly constructed to ensure that the TX saw a more-or-less constant impedance across both sidebands.

Most other Band I TXs were in fact low-level modulated (with the honourable exception of the S.T.& C. CG1 a la Gerry Wells) and from what I remember they didn't have VSB filters at all: the required response was obtained by the tuning of the linear amplifiers.

When we moved up to UHF, which is what you're more interested in, the early TXs were carrier-frequency modulation devices and since it was not easy to produce a stable VSB response by the tuning of the linear amps, we were back to discrete VSB filters for a while - resonant cavity types. Mercifully, IF modulation drives appeared quite quickly before too many staff went insane, and this helped enormously. Modulating at IF (frequencies between 35 and 40 MHz were usual) meant a single filter design would do the job irrespective of final output channel: modulation took place at very low levels in these beasties, around 1 mW peak sync, so there were no worries about constant-impedance devices, and everything was done with lumped components and was an absolute doddle to set up with a sweep generator.

Of course, they caused problems elsewhere: any filter which works near the IF frequency will introduce phase errors, which had to be sorted out using a group-delay corrector unit which was a long way away from being easy to set up. A sweep generator was of little use for this since it was not phase-sensitive. The sound/vision combiner units also caused group delay problems which were sorted out in the same corrector. (You were thinking of including a G.D. corrector weren't you? No?...)

ppppenguin remarked:
Quote:
...This would require a very linear PA....
Quite. One of the problems about using a non-linear amplifier with a VSB signal is that an effect called 'sideband reinsertion' tends to happen: in effect, the sideband that you've been at great pains to remove mysteriously reappears, albeit at a rather lower level. You can overcome the effect by using linearity precorrectors in the modulator, but quite suddenly what started off as being quite a simple system suddenly becomes complex, and correspondingly hard to set up.

Unless you've got a shelf-full of highly expensive (and not particularly common) test gear, I'd keep away from the idea at UHF. The game probably isn't worth the candle...
Ray Cooper is offline  
Old 19th Apr 2007, 7:25 pm   #4
jim_beacon
Retired Dormant Member
 
jim_beacon's Avatar
 
Join Date: Mar 2003
Location: Bishop's Waltham, Hants, UK.
Posts: 939
Default Re: Vestigial sideband system 'A' modulator

It should be possible to build a mixer type modulator, to give VSB on any channel in either band I or band III. Most of the required technology is detailed in the RSGB's VHF / UHF manual, and, as we are talking about small signal output, it shouldn't be too difficult to make a wideband linear ouput stage to foolow a double-balanced mixer.

Jim.
jim_beacon is offline  
Old 19th Apr 2007, 10:21 pm   #5
Andy Green
Hexode
 
Join Date: Jan 2005
Location: Waterford, Republic of Ireland
Posts: 259
Default Re: Vestigial sideband system 'A' modulator

Quote:
Originally Posted by Ray Cooper View Post
Most other Band I TXs were in fact low-level modulated (with the honourable exception of the S.T.& C. CG1 a la Gerry Wells) and from what I remember they didn't have VSB filters at all: the required response was obtained by the tuning of the linear amplifiers..
- would not tuning a linear amo in such a way as to reject one of the sidebands not have the same effect on the phase response as would a cavity bandpass filter tuned so as to reject one sideband.

How does phase error or group delay look like as distortion of a picture - presumably it affects transitions around edges in the picture?

I can see how the filter described in the first part of Rays post work, as I recall, a lowpass and highpass summed acts as an all pass filter (constant amplitude), so is set up to compensate for the group delay - and thus presents a constant impedance across the RF channel - surely a scaled to UHF version and much lower power would be feasable - but as you say, probably a lot of trouble to implement. I'm not doubting you Ray (you obviously know what you are talking about), just trying to get my head round a few things! - I appreciate the detailed answer!
Andy Green is offline  
Old 20th Apr 2007, 7:55 am   #6
ppppenguin
Retired Dormant Member
 
ppppenguin's Avatar
 
Join Date: Dec 2003
Location: North London, UK.
Posts: 6,168
Default Re: Vestigial sideband system 'A' modulator

Quote:
Originally Posted by jim_beacon View Post
It should be possible to build a mixer type modulator, to give VSB on any channel in either band I or band III....
Doing SSB this way is relatively common with a phasing type modulator but surely VSB isn't possible. Also a phasing modulator is hard because of the wideband phase shifter needed for the video. The filter method would work and is effectively the same as Ray Cooper described for professional use.
ppppenguin is offline  
Old 20th Apr 2007, 1:24 pm   #7
Andy Green
Hexode
 
Join Date: Jan 2005
Location: Waterford, Republic of Ireland
Posts: 259
Default Re: Vestigial sideband system 'A' modulator

Quote:
Originally Posted by ppppenguin View Post
Doing SSB this way is relatively common with a phasing type modulator but surely VSB isn't possible. Also a phasing modulator is hard because of the wideband phase shifter needed for the video. The filter method would work and is effectively the same as Ray Cooper described for professional use.

Well I know that any waveform can be synthesised once you have control of the in phase (I) and quadrature (Q) components, so in theory it should be possible. As you say though building the phase shifter to do it in the analogue domain would be tricky enough - I suppose you could build it such that the 90 degree phase shift was not implemented for the frequency bandwidth of the vestigal sideband. I have read recently ( I think it was in Pat Hawker's column in Radcom) that there are some nifty design programmes that can take the pain out of designing these phasing networks, and will allow you to use standard values as well. (see for example this pdf document
http://home.planet.nl/~niess153/Poly...works_V3.5.pdf
and see links in this page
http://home.planet.nl/~niess153/Poly...rk%20links.htm
The best way to do it in theory would be in DSP form - Maybe Darryl could say how practical this might be in realistic terms?

Last edited by Andy Green; 20th Apr 2007 at 1:43 pm. Reason: add link, correct spelling mistake
Andy Green is offline  
Old 20th Apr 2007, 5:37 pm   #8
jim_beacon
Retired Dormant Member
 
jim_beacon's Avatar
 
Join Date: Mar 2003
Location: Bishop's Waltham, Hants, UK.
Posts: 939
Default Re: Vestigial sideband system 'A' modulator

Jeffrey,

most amateur transmitters use crystal filters to eliminate the unwanted sideband, and a double balanced mixer to eliminate the carrier. I meant to suggest that this approach would be suitable to make a VSB system, in fact lumped LC filters could be used, if the IF is sufficiently low.

My Philips pattern generator uses a mixer type system to cover all of bands I & III, but produces a DSB ouput. The video is modulated onto an IF signal equal to the intercarrier spacing, then the video IF and sound signal are mixed up to the required ouput.

Jim.
jim_beacon is offline  
Old 20th Apr 2007, 6:49 pm   #9
Ray Cooper
Retired Dormant Member
 
Join Date: Jan 2007
Location: Porthmadog, Gwynedd, UK.
Posts: 199
Default Re: Vestigial sideband system 'A' modulator

Andy Green:
Quote:
- would not tuning a linear amo in such a way as to reject one of the sidebands not have the same effect on the phase response as would a cavity bandpass filter tuned so as to reject one sideband.
Any VSB filter system tends to produce phase anomalies in the regions where the amplitude response is not flat. But in general terms, the phase errors in Band I monochrome were acceptable: in Band IV/V colour, they were not - and became even less so when 'refinements' like teletext were loaded onto the vision signal.

You tended to get away with it at the VHF end, because the 'fractional bandwidth' of the roll-off part of the VSB response was much greater than that of a similar response up at UHF: eg a 1MHz rolloff at 50MHz (2%) gave far less problems than a 1MHz rolloff at 500 MHz (0.2%) - the implied 'Q' of the VHF filter is much lower than that for the UHF one. The lower 'Q' tends to produce less phase distortion.

Quote:
How does phase error or group delay look like as distortion of a picture - presumably it affects transitions around edges in the picture?
Generally speaking, if you are looking at a sharp edge, the leading edge of the pulse may be slowed down, and the trailing edge have overshoots (or vice-versa). This looks like 'black-after-white', or close-in ghosting. You can easily see the effect by de-tuning the local oscillator of a domestic TV RX. In very severe cases, it can produce 'ringing' - multiple images.

(I just can't resist relating an embarrassing incident here - the early Band I TXs, Sutton and the Moss, were high-level modulated units having external VSB filters. They always used to struggle a bit for output power, in fact Sutton was only run at 35kW peak white when first installed. When the later low-level-mod units, Shotts and Wenvoe, came into action, there was a great deal of crowing that they were much better: 'look, just turn up the knob, we can get 50kW, 55kW with ease...' Unfortunately, after a few years, some ******* in the Post Office devised the Pulse and Bar test, which told us much more definitively just what was going on inside these TXs. Sutton and Moss gave quite decent results: but Shotts and Wenvoe were truly terrible... they had to turn the output power down to about 35kW to give the same sort of results that the others were giving. Har har. There is also a legend that Shotts spent the first few years of its career with the vision amplifier tuned to the wrong sideband, which did at least explain the complaints received of 'fuzzy pictures'. I don't think I should really be telling you all this...)
Ray Cooper is offline  
Old 27th Jul 2007, 7:10 pm   #10
Andy Green
Hexode
 
Join Date: Jan 2005
Location: Waterford, Republic of Ireland
Posts: 259
Default Re: Vestigial sideband system 'A' modulator

For anyone interested in VSB modulators, I just found this very intereting article on an American Amateur radio site which explains what happens if you try to amplify a VSB signal with a typical amplifier. You get spectral 'regrowth' is the short answer .
http://www.hamtv.com/pdffiles/ATVDSBvsVSB.pdf

If anyone did want to transmit 405 line TV on the 70 Cm ham band, then a VSB filter from DCI looks like it would do the trick!
http://www.hamtv.com/pdffiles/ATVDSBvsVSB.pdf

Last edited by Andy Green; 27th Jul 2007 at 7:20 pm.
Andy Green is offline  
Old 27th Jul 2007, 7:38 pm   #11
jim_beacon
Retired Dormant Member
 
jim_beacon's Avatar
 
Join Date: Mar 2003
Location: Bishop's Waltham, Hants, UK.
Posts: 939
Default Re: Vestigial sideband system 'A' modulator

I would suggest that the linear amp in that article is not set up correctly, or has a relativly high Q in the tank circuits (hence nartrow bandwidth). If that were the case with a linear amp used on SSB, the second sideband would be re-generated!

Jim.
jim_beacon is offline  
Old 27th Jul 2007, 9:37 pm   #12
Ray Cooper
Retired Dormant Member
 
Join Date: Jan 2007
Location: Porthmadog, Gwynedd, UK.
Posts: 199
Default Re: Vestigial sideband system 'A' modulator

It's all down to Power Amplifier linearity, rather than bandwidth. The linearity requirements to avoid excessive sideband reinsertion are really quite tight - you're looking at an overall system non-linearity no worse than a few percent to get decent results.

Fortunately, the key word is 'overall' - the P.A. linearity can still be poor, but if the RF going into it is pre-corrected in the opposite direction, satisfactory overall linearity and freedom from reinsertion can still be obtained. Unfortunately the price you pay for this is system complexity - well worth it (in fact, economically unavoidable) if you're talking about a professional amplifier putting out kilowatts of power: but for low-power amateur usage, highly debatable.

If you wanted to put out 405-line in the 70cm band, it might be simplest to go straight for a high-level modulated setup and avoid the problems that you'll otherwise get with any linear amplifier system. You'd probably be wise to be fairly modest in the amount of any pre-correction that you'd apply to this - possibly just some sync stretch to get around the triode-like non-linearity of any valved modulated amplifier, and perhaps just a bit of peak-white stretch to overcome that flattening that you'd get if you drive peak-white into any grid current, though this is perhaps gilding the lily a bit. And now you're back to a DSB signal, so any VSB filter is going to have to be on the output of the mod amp and running at carrier frequency. It could be done with a suitable multipole cavity filter, but it's not going to be that easy unless you've got suitable gear to set the thing up properly after you've built it. Such a filter will also have the highly desirable property of removing the harmonics that are an inevitable product of a high-level modulated TX.
Ray Cooper is offline  
Old 28th Jul 2007, 3:04 pm   #13
Andy Green
Hexode
 
Join Date: Jan 2005
Location: Waterford, Republic of Ireland
Posts: 259
Default Re: Vestigial sideband system 'A' modulator

I just noticed that I mis-posted the link to a provider of suitable VSB filters in my post abocve - see
http://www.dci.ca/?Section=Products&SubSection=Amateur
The MCI filters are about $370, so are within the realms of affordability. Although I have the equipment in work (Spectrum analyser with tracking gen and precision return loss bridge), I don't have suitable metal bashing facilities to build a cavity filter.
How would one go about providing the pre-distortion in practice Ray?
Andy Green is offline  
Old 28th Jul 2007, 4:53 pm   #14
Ray Cooper
Retired Dormant Member
 
Join Date: Jan 2007
Location: Porthmadog, Gwynedd, UK.
Posts: 199
Default Re: Vestigial sideband system 'A' modulator

Quote:
How would one go about providing the pre-distortion in practice...
Many ways of doing it, but about the simplest is on the attached file.

Sorry that I'm using valves - but due to my advanced crumbliness, I'm happier with them. No reason at all why it can't be done with transistors/fets etc.

Referring to diagram - first thing off, the box labelled 'clamp'. You can implement this how you will - possibly a DC restorer would work, but I have to say that a nice solid 4-diode clamp will be better. The most important thing of all is that the DC 'sit' of the input waveform is tightly controlled at the valve grid - else it makes complete nonsense of the pre-stretch circuit operation.

Forgetting about all those diodes in the cathode circuit, for a moment - we have a straightforward voltage amplifier circuit with cathode degeneration, and it is easy to see that the voltage gain will be dependent on the value of the cathode resistor Rc.

However, when the cathode voltage dips below the potential set by VRs, the diode Ds will start to conduct, placing resistor Rs in parallel with Rc: the total cathode resistance therefore falls, and the stage gain rises, 'stretching' the syncs.

Similarly, if the cathode potential rises above that set by VRw, Dw starts to conduct, placing Rw in parallel with Rc: again the total cathode resistance falls, the gain rises and we get 'peak-white stretch'.

In practical terms, Rs and Rw can also be variable resistors, so the gains in the sync and white stretch regions can be tweaked.

This is all highly simplified of course, and all the refinements added to professional gear explains the large number of adjustment pots visible on the front of the modulator chassis...

As a total aside, let me mention that it is possible to provide very large amounts of sync stretch with this sort of arrangement - but you only need so much to cut off a modulated amplifier. If you overdo it, you can end up with extremely sharp corners on the sync pulses, which result in sideband 'splatter' audible beyond the channel limits. So don't.
Attached Thumbnails
Click image for larger version

Name:	predistort.png
Views:	135
Size:	15.8 KB
ID:	11208  
Ray Cooper is offline  
Old 28th Jul 2007, 5:34 pm   #15
ppppenguin
Retired Dormant Member
 
ppppenguin's Avatar
 
Join Date: Dec 2003
Location: North London, UK.
Posts: 6,168
Default Re: Vestigial sideband system 'A' modulator

Ray's approach is very similar to some gamma correctors. They also give stretch towards black level. The main difference here is that we want the syncs to be stretched too.

The modern way of doing this is all digital. It would be conceptually trivial to add something like this to the Aurora standards converter. At a practical level, assuming you are Darryl and have access to the source code, it's just a lookup table with whatever mapping you want. Need to have enough memory resource in the FPGA and beware of running out of resolution in the DAC. The more quantising levels you give to the syncs, the fewer are left for video.
ppppenguin is offline  
Old 28th Jul 2007, 6:27 pm   #16
Ray Cooper
Retired Dormant Member
 
Join Date: Jan 2007
Location: Porthmadog, Gwynedd, UK.
Posts: 199
Default Re: Vestigial sideband system 'A' modulator

Quote:
...it's just a lookup table with whatever mapping you want....
Errm - yes -

However it would be a bit of a bind to have to re-burn your hardware every time the linearity of your mod amp changed (which they do, with monotonous regularity), so on the whole I think I'd prefer to stick to a nice twiddlable pot labelled 'more' and 'less'...

And, conceptually, I'd have thought that the sync waveform only needed one bit of resolution: a tweak on the DAC bit that produces syncs should suffice, to produce the required sync amplitude (watch out for the rise-time, though): or have I missed something?

Of course, if you're working with home-made waveforms rather than industry-standard stuff, you don't need sync-stretch at all: you just set the picture/sync ratio to about unity somewhere in your video chain. There's very good historical precedent for this: Alexandra Palace, pre-war, used no pre-distortion whatever - they just set up P/S ratio to unity in the mixer channel and left it at that. There wasn't any peak-white stretch, either. Sutton Coldfield was the first 405 tx in this country to incorporate pre-distortion techniques (they didn't get it quite right, but they learned a hell of a lot...)..
Ray Cooper is offline  
Old 28th Jul 2007, 8:02 pm   #17
ppppenguin
Retired Dormant Member
 
ppppenguin's Avatar
 
Join Date: Dec 2003
Location: North London, UK.
Posts: 6,168
Default Re: Vestigial sideband system 'A' modulator

Quote:
Originally Posted by Ray Cooper View Post
However it would be a bit of a bind to have to re-burn your hardware every time the linearity of your mod amp changed (which they do, with monotonous regularity), so on the whole I think I'd prefer to stick to a nice twiddlable pot labelled 'more' and 'less'...

And, conceptually, I'd have thought that the sync waveform only needed one bit of resolution: a tweak on the DAC bit that produces syncs should suffice, to produce the required sync amplitude (watch out for the rise-time, though): or have I missed something?
The lookup table could easily be in RAM and readily adjustable. The biggest problem would be the user interface to allow this. I agree that an nice analogue pot is far more intuitive though rather less stable and consistent. I'm really an analogue person at heart, it's just that I've been doing digits for too long

When doing direct digital synthesis of a waveform (which is what the Aurorae do) it's nice to have a straight DAC with enough resolution to do the whole job. An exception for sync (I've done it this way myself in designs for clients) can be a nuisance resulting in quite a lot more circuitry. You mention rise time; the sync waveform is meant to have nicely shaped edges but I suspect this refinement can be sacrificed in a transmitter. Got to watch the bottom half of the burst in that nasty colour stuff.

As somebody who has always worked on studio equipment where 1Vp-p of video is the rule, I've always had a mixture of respect, fascination and amazement for transmitter engineering where video is flying around at kilowatt levels.
ppppenguin is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 10:29 am.


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 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.