Crystal Filters

[Wendymott]

Hi peeps. I bought a Module on Ebay containing 3 x 9.0 Mhz I have tried looking on google for the bandwidth of the filters, but no success.
I wonder if anyone else bought one of these modules from Israel, they were £10 ish.
The types have the following on the case :-

KRISTALL VERARBEITUNG.. NECKARBISCHOFSHEIM Gmh. XF-9 S 42
KRISTALL VERARBEITUNG.. NECKARBISCHOFSHEIM Gmh. XF-9 S 43
KRISTALL VERARBEITUNG.. NECKARBISCHOFSHEIM Gmh. XF-9 S 44

I am hoping that at least one is suitable as a SSB filter.
Fingers crossed.

[G6Tanuki]

These were originally produced by KVG in Germany: they had a range which covered SSB/AM/FM bandwidths. The XF-9B was the favourite one for SSB: I used just such a beastie in my rendition of the "G4CLF" 9MHz SSB board using Plessey SL600 chips.

I gather that KVG dropped this side of the business some time in the early-1990s so you may have some difficulty tracking down the specs.

There are references to fitting these in Drake receivers - https://groups.yahoo.com/neo/groups/...messages/21504

Looks like the S42 is for AM double-sideband-style bandwidths, the other two are USB and LSB filters. All centred on 9MHz.

[Wendymott]

Hi Tanuki. Wow. Thanks VERY Much. I am also looking at the Plessey G4CLF as a project, see my other posting in Circuits ref SSB exciters. I wonder if you were please with the result or did you move on.
Regards

[G6Tanuki]

Glad to be able to point you in vaguely the right direction; you could say I've got a bit of professional history tracking down this sort of stuff.

My experience with the G4CLF design was somewhat mixed: there were various versions offered as kits which didn't fully decouple the AGC line to RF/IF/AF and as a result local-oscillator leakage got past the IF filter, amplified and detected and led to a residual static DC level on the AGC line which meant the RX was about 15dB deafer than it should have been.

I solved this by removing the front-end double-balanced-mixer stage from the main PCB and fitted it in a separate diecast box - and included a bunch of switch-selected series-tuned traps for the LO frequencies between the output of the balanced-mixer and ground, so that the balanced-mixer output always had a low-impedance path at the LO frequency to earth.

Xtal filters like the KVG ones may have a good 'near-in' signal rejection but that doesn't always apply at other frequencies. 80dB down within +/- 10KHz of the centre frequency but only 40db down at 5MHz away... and the balanced-mixer's only giving 30dB attenuation of the rather-powerful LO. Do the sums yourself - the residual LO signal can be as strong as the required signal !

The other issue was that following a narrow front-end IF filter with lots of IF gain and a product-detector means that any 'other sideband' noise from that 100-odd dB of IF gain will be resolved by the product-detector as an untunable 'mush' background. If you're running a 'hot' high-gain IF strip after the primary selectivity you really need a *bit* of extra selectivity to filter out the other-sideband noise before the product-detector. This selectivity doesn't need to be too tight but it does need to be able to handle significant signal-levels without itself introducing cross-modulation. I found a simple ceramic filter did just fine for SSB, but it did horribloe things to the overall IF-strip group-delay characteristics when trying to use it on digital modes.

Some 1980s Japanese ham transceivers *cough*Icom*cough* were horrible in this 'other sideband' IF-strip-generated mush sense.

People assume filters are simple, linear, symmetrical, bi-directional things.

Simple, no!
Linear - only at low signal-levels.
Symmetrical - only if terminated properly for all source/destination frequencies.
Bi-directional, if you're lucky.

[Radio Wrangler]

Back in the 70s/80s people would have killed for KVG filters at affordable prices and three for a tenner is unbelievable.

In terms of what KVG could do with crystal filters, I reckon only the C.E.P.E. arm of Thomson-CSF could do better.

What Tanuki's said about adding filters late in a high gain IF to avoid double sideband noise aggregation in the product detector is right on the money.

James did his board when he worked for Plessey, and it was a vehicle for demonstrating the SL600 series. I've not met up with him for several years. He used to come Northwards occasionally in his guise as the European market manager for Analog Devices.

The SL600 devices are now hard to obtain, but not quite impossible. They were great examples of what could be packed onto ICs, but their performance was limited in several ways. I think it's fair to say they aren't up to the standard of KVG XF-9 filters. I think you should start reading stuff by Colin Horrabin and Jacob Makhinson.

Now you've got the crown jewels, what are you going to do with them?

David

[Wendymott]

Hi David. The intention was to make a SSB Exciter, and I thought these filters would provide the basis, coupled with a MD108 DRM. As I now know, thanks to Tanuki, they are 9 Mhz +/- 1.5Khz devices, thus I can switch between the two for upper and lower Sidebands keeping the L/O at 9 Mhz. Its early days yet and I am certainly not in your league's. I have a number of the SL600 devices, but only two SL640 DBMixers. I like the RX/TX by G4CLF but from notes from Tanuki, it leaves a lot to be desired.
I have made a SSB Exciter using Audio phase modulation but that has its own problems... see my other Posting

[G6Tanuki]

Heh - don't get me wrong, I'm not 'dissing' the G4CLF circuit - just saying that it needs to be viewed with reference to the era and context in which it was designed.

It's very much presented as a "sample circuit" intended to demonstrate how - in the early-1970s - the then-available new ICs could be used to produce a compact SSB generator/demodulator with a price/performance that would have been hard for an amateur - who may well have been unfamiliar with IC-based circuit design and construction - to achieve using valves or discrete semiconductors of the era.
As a way to "get something up and working" relatively easily it was great - but it was never intended to represent design perfection. You were meant to use it as a basis and adapt it to your needs! Alas, too many people in the amateur fraternity/sorority saw the circuit as representing "the be-all and end-all" of SSB generation using the SL600-series chips (in the same way as audiophools slavishly replicate early valve-amplifier designs which were never really intended to be the 'gold standard' and have since been surpassed).

If you compare the G4CLF bare-bones SL600 circuit with something like that of the Clansman PRC320 you'll see what I mean! 40 years on since the SL600 series was designed, there are chips and devices available with performance figures that the Plessey guys could only imagine in their wildest dreams - your mobile-phone or satellite-TV receiver is stuffed full of them, and at dirt-cheap prices too.
By all means build a G4CLF circuit if you want to re-live the experiences of 40 years ago. Just don't expect it to handle like a modern design!

[Wendymott]

Hi Tanuki. Sorry for mis quoting you. The nice thing about the SL 600 series is the physical size and pcb foot print, to make a prototype lashup is so easy, relative to the newer parts which in some cases are tiny SMD devices where a pallet is required as a chip carrier. If I cannot improve the Phasing modulator, I will jump to the Diode mixer and Xtal filter option. I have a number of the SL600 series devices.

[Radio Wrangler]

Have a look at "H-mode" mixer.

It uses CMOS bus switches and you can wind your own transformer. Blows the socks off diode rings and SL6440 etc, yet low power and a doddle to make.

David

[Wendymott]

Thanks David, you certainly opened a new "Can of worms" with that sentence. In my application it needs to be a SSB modulator. Thus Audio in on one leg, a RF input of 9 Mhz carrier, to a sideband filter. Pointers please.

[Wendymott]

Mod, do you think this should be moved to HOMEBREW cat ?? its getting more off subject.. THanks

[Radio Wrangler]

The H-mode circuit is a double-balanced mixer and one port is coupled right down to DC, so put audio into that port (or take it out in a demodulator)

Demodulators are mixers are modulators.

Once you see it, everything becomes simpler than you thought it was...

David

[Wendymott]

Thanks David. I will get into it and order a couple of chips.

[G0HZU_JMR]

I did a google and found this french site:

http://www.tsf-radio.org/forum/forum.php?show=188929

Eh eh, salut visage pâle!

J'ai mesuré les miens, car je n'ai rien trouvé sur internet, même chez KVG.
Ces filtres étant de provenance surplus militaire, ça n'est pas vraiment étonnant.
Avec le tracking de l'analyseur de spectre, et une adaptation résistive variable de chaque côté, les résultats pour les 3 filtres 42,43 et 44 donnent ceci:
S42: bande passante=+/- 3k6 à -3db : fréquence centrale 9.000000 MHZ
S43: +/-1K55 à -3db : 8.998200 MHZ
S44: +/-1K55 à -3db : 9.001800 MHZ

Capa d'entrée et de sortie: 25 pF
Z entrée/sortie 560 ohms (mesurés)

Pour le 150 S04, c'est un 150 Mhz, mais je n'ai rien à son sujet.

So based on this person's measurements it looks like the S43 and S44 could be used as USB and LSB filters as they have about a 3kHz BW at the -3dB points. I'd recommend you try a few experiements to find the best impedance matching but at least this person provides some info on matching as well

[Wendymott]

Hi Jeremy. Thanks for the info, I'm certainly looking at using these filters, possibly getting some more too only £24 inc postage from Israel. Fortunately Tanuki sent me the info too, and as David says the "crown jewels" of filters.
I have the SN74CBT3125 Quad fet chips now as well to make a "H" Modulator, thus a few options.

[Radio Wrangler]

The nice thing is you only need 9.000000 MHz as a CIO and you can switch in the USB or LSB filter without changing it.

Now if you get a number of 9.000000 MHz crystals cheaply, ladder filter techniques will yield a nice CW filter (several bandwidths if you want). A Monsieur Pochet put an article in wireless world. Jack Hardcastle did some articles for Radcom, and Wes Heyward put info and programmes in "Intro to RF design"

H-mixers will do modulator mixer and demodulator beautifully. To get good mark-space balance, use oscillators at twice the frequency you first thought of and a divide by two.

David

[Wendymott]

Hi David. I have those KVN Filters at 9 Mhz +/- 1.5Khz.... and the circuit I found for the "H" mixer has a Squaring circuit with balance... see attached....
I understand this arrangement negates the use of a divider...
By the way.... I said I bought them for £10.... I was wrong.. they are still advertised at £24 inc postage from Israel...but still a good price.... I am going to get another module.....
Looking at the schematic.... and text... there is no mention of the filters type used in the DC supply..... have you any ideas ?

[Wendymott]

Looks like my first posting excited a few of you.... the guy selling the filter modules has sold out..I bought another, leaving 3 left....and looking this morning.. they have all gone...I bet he wonders why.. after sticking for months, he sold out within days...glad I bought another module.

[Radio Wrangler]

T'wasn't me.

I've got boxes of 1.4MHz marine grade SSB-TTY-CW filters for that 'Doomsday' receiver I've always been intending... yeah, right!

David

[Wendymott]

Hi David I sent you a PM.. ref above