Modifying discriminator coil for narrow band FM.

[Biggles]

Just a quick question for the experts. Would it be possible to use a standard 10.7MHz FM broadcast IF discriminator coil from a valve set to demodulate narrow band (12.5KHz chan spacing/dev at 2.5KHz) signals? Could this be as simple as adjusting the associated component values in the discriminator circuit, or is it an inherent design with the coils inside the transformer? Obviously the audio would be well down on normal wide band FM, so this could be compensated with an extra gain stage, but this doesn't feel the right way to do it. Any suggestions?
Alan.

[Jon_G4MDC]

In principle it would be as you suggest. Audio output will be proportional to deviation and tuned circuit Q.

In the Pye Westminster discriminator there are shunting resistors across each half of the coil which set Q for different system deviation values R28 & 29 in the attachment. Value given is 10k for 25kHz and 1.8k for 50kHz.

Of course these are values for 455kHz IF. For 10.7MHz and Broadcast BW you would scale as appropriate.

[G8HQP Dave]

75kHz deviation at 10.7MHz is 0.7%. 2.5kHz deviation at 455kHz is 0.55%. You are attempting 2.5kHz at 10.7MHz, which is 0.023%. To a first approximation, you need a discriminator coil Q which is about half the inverse of these fractions, so about 70, 90 and 2200 respectively. The first two are easy and commonly done. To do the latter you would need to use a crystal or ceramic resonator, which is why most narrowband FM receivers convert to 455kHz for the final IF.

The alternative is to remember that people listen to BBC Radio 3 and that probably often has deviation at the 2.5kHz level so all you need is some audio amplification.

[Biggles]

Thanks for the replies. I might just try and hook up an IF strip and try some reduced deviation tests to see if it works. I had forgotten that most PMR sets use dual conversion so the discriminator is running at 455KHz. There is normally good reason behind most circuits being the way they are. Its really because I have quite a few old 10.7MHz IFT's from domestic sets and was wondering if I could put them to use on amateur/NBFM home made gear. Maybe not so good an idea after all. The performance probably wouldn't be all that good.
Alan.

[Jon_G4MDC]

Been mulling this one over. What about a Q multiplier approach for a gentle Q boost?
I have never seen it done but that doesn't mean it's impossible...
If you have some bits and you fancy playing with them perhaps.

[Biggles]

It's a job for a rainy day (so shouldn't have to wait long!). One of those random ideas you have when your mind starts wandering. I have plenty of bits, but not enough time to get round to my ideas sometimes. My mind works far faster than my hands. Am I right in thinking a Q multiplier has another name...a gyrator? or am I getting mixed up. A Q multiplier will just have a little positive feedback applied to peak up the selectivity a bit, like reaction on a TRF? I will investigate.
Alan.

[Radio Wrangler]

A Q-multiplier is essentially a negative resistance generator used to cancel some (but not all) of the losses of a tuned circuit, leading to highet Q and narrower bandwidth.

Accuracy to be sure of stability will be difficult. Stability of the centre frequency will be the biggest problem, especially with a narrow band signal.

It's probably easier to down convert to 455 and do it there.

David

[turretslug]

The MC3357's raison d'être.

[Jon_G4MDC]

And you get a squelch thrown in too.

I still think trying the Q multiplier might be interesting.

Somewhere on here is a thread about receiving FM with super regen.
That could be viewed as an extreme use of Q multiplication making an injection locked oscillator maybe?

[G6Tanuki]

"Q" (and hence slope) is the heart of 'classic' FM-detectors: as noted the 'percentage frequency-shift' is greater when the IF is lower; I recall some professional FSK RTTY-demodulators designed for use with really narrow shift (150Hz or so) which downconverted to an IF of around 15KHz before feeding the discriminator.

If you can get a high-enough Q you can do the detection at a high frequency though - I recall once seeing a proof-of-concept Band-II FM receiver that used a really high-Q cavity [silvered copper, physical geometry optimised for maximum Q] resonant at the signal-frequency, whose Q was high enough that it would effectively slope-detect 75KHz-deviation FM.

So it can be done...!

[G8HQP Dave]

If you live near the transmitter you could make a Band II FM crystal set!

[Jon_G4MDC]

Beer barrel cavities would be suitable for that although I have never tried it.
This one is for 70MHz.

[Radio Wrangler]

When building a keg filter cavity, it's important not to use nicked or empties. You don't have a clue about their condition. Buy new full ones and empty them yourself just to be sure. For a good 2m repeater, you may need five or six cavities.

David

[G8HQP Dave]

How do you drill a hole in the right place after six kegs?

[Radio Wrangler]

Not sure I remember, but after you've finished, everyone in the group will agree that it's a VERY good repeater.

David

[CambridgeWorks]

The nbfm handbook rsgb, from 1970s? I seem to remember a cystal discriminator circuit? A good practical reference book if you can get a copy.
It was a book I often referred to back then in my heady construction days. My 1W 2 metre 3 channel portable all on one home etched and desined pcb was one achievement. Still have it somewhere, built around 1976. A 6 channel scanning mobile as well for 2M that used a Pye pa board for 15W out. Though both used ics, a CA3089 I think.
The thought of making a discriminator coil just was not on back then!
Rob

[G6Tanuki]

The mid-60s "Stornophone 500" walkie-talkies used by MetPol etc. used a crystal discriminator at 10.7MHz; I guess that back then when everything was discrete-components this took up less space than a conversion to 455ish MHz (which would have needed a crystal anyway).

You may be able to crib the circuit.

http://www.storno.co.uk/stornophone_500.htm

[Radio Wrangler]

Special 10.7MHz two-pole crystal filters were made as phase shift elements for quadrature detectors for NBFM demodulation.

David

[Biggles]

Some interesting ideas there. I have used various FM IF chips down the years, including most of the MC3357/3359/3361 series with great success. For this project I just wanted to use up some of the old tat I have in boxes removed from various sets in the past. Seems like it may be more of challenge than I anticipated. Obviously the use of valves had it's appeal too. I built an all valve dual conversion airband receiver a few years back, mainly for the challenge and the fact that I had all the bits. That was AM though so not such of a problem. I had to leave behind a load of those massive piston type barrel cavities as I didn't have room to store them a while back during a de-commissioning job. They probably all went for scrap. No good for "/portable"!
Alan.

[Synchrodyne]

The differences between Foster-Seeley discriminators for wideband and narrowband FM seems to have been covered only in a very general fashion in the literature. E.g. one can find that for the wideband case, overcoupling was used along with damping to achieve the desired bandwidth. Perhaps for the narrowband case, critical coupling was used?

There were cases where the same Foster-Seeley discriminator was used for both wideband and narrowband reception, with no apparent switched adjustment between the two. Examples were the Eddystone 770R and 990R. One might infer that using a wideband discriminator for NBFM was not seen as unduly compromising the SNR. Of course, the IF channel selectivity was adjusted accordingly, and for NBFM, the 990R offered a choice of 30, 15 or 7.5 kHz bandwidth crystal filters.

In the quadrature demodulator case, the CA3089 application note covered NBFM at 10.7 MHz, indicating that with the higher Q coils, drift was a problem and that this could be corrected with an AFC circuit acting on the quadrature tank via a varactor diode. Another approach, although not in the application note, was to use a crystal for the quadrature tank circuit. This was done in the Wireless World 1977 November, December 2-metre transceiver, and occasioned some subsequent discussion in the letters pages.

Eddystone used a CA3089 (at 21.4 MHz) for both wideband and narrowband FM in its 1990R receiver, again with no apparent switching between two at the demodulator, and no AFC needed for NBFM, given that the quadrature circuit was wideband.

None of this helps with the quest to modify a wideband discriminator coil for NBFM use. But the Eddystone precedents at least suggest that it might be worth trying the discriminator coil “as is” behind a narrowband IF strip.

Something else that come to mind is that somewhere I have a GE (American) portable receiver, long unused, that covers the old US VHF analogue TV bands (sound only) and NOAA weather radio as well as FM and AM. NOAA weather radio, NBFM at around 162 MHz, was covered by a downward extension of the TV high band. So the same FM demodulator, presumably wideband, was used for NOAA weather radio. My recollection is that the sound volume when tuned to the local NOAA station was not hugely lower than it was for FM broadcasts, which at the time surprised me, given the big difference in maximum deviation. I am not sure whether, back in the 1990s, NOAA was ±5 or ±2.5 kHz. I guess though that it was highly compressed voice, always close to maximum modulation.


Cheers,