Improving Selectivity In Vintage Receivers - Page 3 - UK Vintage Radio Repair and Restoration Discussion Forum

Bazz4CQJ · 29th Oct 2022, 12:15 pm

Thanks for that. Jon has got the numbers, so he may now be able to make a stab at estimating what load a ceramic would see if placed where he suggested earlier in the thread. I think that the comments that Peter made in his published article do suggest that even if the environment for the filter is less than perfect, it still improves selectivity, though the loss incurred may be significant.

B

Jon_G4MDC · 29th Oct 2022, 8:14 pm

Yes Bazz, now I definitely have something to play with. Thinking on how to make best use of it. So far I have not tried to model the tuned circuit in LTspice although I do have it.

A bit lacking in experience with it. I have modelled some DC and Slow AC stuff by graphing Voltages and Currents. So far I have not explored how to look at impedances. (maybe LT Spice needs a thread of it's own!)

For fun I have put it in ARRL Radio Designer, of which I have somewhat more experience. Modelling a 2.2mH inductor with Q of 100 @ 0.5MHz in parallel with 55.7pF shows resonance at ~455kHz. Real Part of Impedance ~560k Ohms Imaginary 0 and of course so far without a tap. So assume a perfect transformer of turns ratio 5:1 with impedance transformation of 25:1 it will be 22.4k at the tap and that is too high. Thoughts turn to adding a 50k shunt resistor across the trimmer. That would look like 2k at the tap.

A better way to model the tapped inductor might be the MUI statement which allows entry of L1 and L2 inductances along with Coupling K or Mutual Inductance M. This one does not allow Q to be entered.

Which is the best approach?

Radio Wrangler · 29th Oct 2022, 9:27 pm

ARRL radio designer was Ulrich Rohde's contribution to the ARRL, because he not only owns 50% of R&S, he also owns what was Compact Software one of the earlr RF sim firms. It was quite something to be told I'd got a nice mention in one of Rohde's books, me working at the time for HP... You just can't buy that sort of publicity.

I've pushed LT spice to microwaves and it works. You just have to concoct effective models where you need them. You can model general coupled-resonator filters and with appropriate values handle crystal ones, LC ones or ceramics.

Mis-termination of a multi-order coupled resonator filter usually manifests itself as increased ripple.

If you model a quartz crystal fundamental and C0, you can build ladder filters easily. Use shunt coupling capacitors in a string of crystals in series to make a ladder. Make the coupling capacitors in the pattern of ratios as in that WW article by Monseiur Fochet F6BQP, and then play about with simulations of varying source/load resistance to minimise ripple and see what response you get. Scale all the capacitors, keeping the ratios and then play with the terminating resistors and you can empirically move your simulation to wider or narrower bandwidths. Bigger Cs mens narrower.

You can have a lot of fun before getting the soldering iron warm. Designing a crystal filter by hardware hacking in simulation.

It works!

I've done overtone ones as well.

A far cry from designing full-lattice crystal filters and then having to get a lort of different crystals made.

Oh, yes, and with the crystal ladder filter, try adding inductance across individual crystals to tune out their C0 by resonating it at the centre frequency. You now get a symmetrical filter rather than theSSB one you started with.

Jack 'JIR and myself had some good chats at Rochdale.

David

Bazz4CQJ · 29th Oct 2022, 9:32 pm

Quote:

Originally Posted by Jon_G4MDC

Yes Bazz, now I definitely have something to play with. Thinking on how to make best use of it. So far I have not tried to model the tuned circuit in LTspice although I do have it. Which is the best approach?

Sorry Jon, but I am out of my depth at this stage. All I can offer is encouragement!

B

Jon_G4MDC · 30th Oct 2022, 3:49 pm

I have now added series resistors to L1 and L2 in the MUI statement as Q killers and now I have it in a tuned circuit which gives the same results as the 2.2mH Inductor with a Q of 100.

Looking at the the Top of the Coil 560k resistive at resonance.

Looking instead at the tap about 60k. This is much to high to load a ceramic filter correctly. I think I would probably just add a 2k resistor here instead or there will be lots of ripple in the response.

Bazz4CQJ · 30th Oct 2022, 5:45 pm

OK, sounds as though you are refining the requirements. I should have asked earlier - what is an MUI statement?

B

Jon_G4MDC · 30th Oct 2022, 6:29 pm

In Spice it is the abbreviation for the Mutual Inductor Model. It is where you enter L1 and L2 Values along with either Mutual Inductance or Coupling Factor K.
I was having trouble with the fact that it won't accept a Q value so the inductors will appear perfect. Since I have access to the terminals I have added series resistors there to damp the Q down to a likely real world value of about 100 given the coil construction. (Does anyone have an opinion on that?)

I'm still thinking that resistive termination is the right way. Also still in favour of an emitter or source follower. That could be made to still look like 60k as load to the crystal and phasing control but it would give a low impedance output with unity Voltage gain.

The low impedance output would drive the ceramic filter through a series resistor of 2k.
The far side of the filter would have 2k to ground as termination and that would connect to the coil tap through a suitable DC blocking capacitor since that point is carrying AVC Voltage.

Any comments about preference for FET or Transistor follower anyone? Any suggestions for suitable devices?

G6Tanuki · 30th Oct 2022, 6:41 pm

My go-to JFETs in times past were the J310 or U310 types , used in grounded gate mode to keep the impedances low and avoids the need for neutralising even up to 144MHz.

They are rather expensive these days, because they are long obsolete. But still worth a look.

Jon_G4MDC · 30th Oct 2022, 6:45 pm

Yes I have some I think. Maybe a TIS88 too. They seem to have a bit too much "go" for this though.

Maybe a humble 2N3819? Idss range a bit wide 2-20mA but i have a few so I could select.

Jon_G4MDC · 30th Oct 2022, 7:14 pm

Playing also with LT Spice. I discovered you can make a sweeper using the modulator block.

Setup a 0 to 1V ramp waveform over 0-1Sec and apply that to the FM terminal.
Specify that 0V is 400kHz and 1V is 500kHz. It did this.

Well it's a start!

Bazz4CQJ · 30th Oct 2022, 8:25 pm

Jon, you might to take a look at the FET circuit Radiofun uses after the ceramic filter in this video. He specs either a BF245 or a BF256.

The video is 17mins 35 seconds long, the key bits are close to the end, but it's all worth watching, jumping forward occasionally!.

https://www.youtube.com/watch?v=mTqXWB6RSOA

B

Radio Wrangler · 30th Oct 2022, 9:23 pm

2N4391 2N4392, 2N4393 a set graded in Idss and Ron. Quite high current beasties. Used by Marconi in some of their better transistorised receivers. HP used the 20 Ohm one as an analogue switch.... rather a good one. It also does high dynamic range RF work as an amplifier. Bit pricey now, inevitably.

David

Jon_G4MDC · 30th Oct 2022, 9:30 pm

That's interesting. I have not come across them.
Many thanks David.

Bazz4CQJ · 30th Oct 2022, 10:34 pm

I took a screen shot of his FET circuit and that is attached. It does look a mess, but that reflects his progressive refinement of what he started with. There are others besides myself on the forum who think he (Radiofun) is worth a looking at.

B

Cobaltblue · 31st Oct 2022, 7:27 am

Three off topic posts removed.

Cheers

Mike T

Jon_G4MDC · 31st Oct 2022, 9:32 am

Looking at the 3 FETs David suggested I think the 2N4392 might be the one to try.

Jon_G4MDC · 1st Nov 2022, 10:01 pm

Just for interest I decided to look at the HRO 1st IF transformer - I expect the others will be much the same.

The variable capacitor measured ~80pF as it was tuned. So maybe the inductors are ~1.5mH or so to resonate with it + strays? There are no taps on these coils.

There is a screen between the two capacitors and approx 1.5" spacing between the coils (presumably inductive coupling dominates).

Bazz4CQJ · 1st Nov 2022, 10:55 pm

I wonder how much the HRO was designed (in the early 30's) on engineering principles and how much on trial and error? That space between the coils is quite something!

B

Radio Wrangler · 2nd Nov 2022, 12:50 am

Oh, that 20 Ohm, highest Idss selected part was HP part number 1855-0020, you might come across them marked 5-020. Amazing what floats up out of old memory.

David

Jon_G4MDC · 2nd Nov 2022, 1:09 pm

An idea is forming that might avoid the need for a FET.

If the crystal filter is not required then the two 100pF capacitors which make the balanced feed which drives it could instead be made into a capacitive tap. That might be made to transform the impedance down to the 2k which the ceramic filter would like to see.

I'm quite tempted to try this on the spare filter box I have.

29th Oct 2022, 12:15 pm	#41
Bazz4CQJ Dekatron Join Date: Apr 2009 Location: Oxfordshire, UK. Posts: 4,935	Re: Improving Selectivity In Vintage Receivers Thanks for that. Jon has got the numbers, so he may now be able to make a stab at estimating what load a ceramic would see if placed where he suggested earlier in the thread. I think that the comments that Peter made in his published article do suggest that even if the environment for the filter is less than perfect, it still improves selectivity, though the loss incurred may be significant. B __________________ Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch.

29th Oct 2022, 8:14 pm	#42
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers Yes Bazz, now I definitely have something to play with. Thinking on how to make best use of it. So far I have not tried to model the tuned circuit in LTspice although I do have it. A bit lacking in experience with it. I have modelled some DC and Slow AC stuff by graphing Voltages and Currents. So far I have not explored how to look at impedances. (maybe LT Spice needs a thread of it's own!) For fun I have put it in ARRL Radio Designer, of which I have somewhat more experience. Modelling a 2.2mH inductor with Q of 100 @ 0.5MHz in parallel with 55.7pF shows resonance at ~455kHz. Real Part of Impedance ~560k Ohms Imaginary 0 and of course so far without a tap. So assume a perfect transformer of turns ratio 5:1 with impedance transformation of 25:1 it will be 22.4k at the tap and that is too high. Thoughts turn to adding a 50k shunt resistor across the trimmer. That would look like 2k at the tap. A better way to model the tapped inductor might be the MUI statement which allows entry of L1 and L2 inductances along with Coupling K or Mutual Inductance M. This one does not allow Q to be entered. Which is the best approach?

29th Oct 2022, 9:27 pm	#43
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,901	Re: Improving Selectivity In Vintage Receivers ARRL radio designer was Ulrich Rohde's contribution to the ARRL, because he not only owns 50% of R&S, he also owns what was Compact Software one of the earlr RF sim firms. It was quite something to be told I'd got a nice mention in one of Rohde's books, me working at the time for HP... You just can't buy that sort of publicity. I've pushed LT spice to microwaves and it works. You just have to concoct effective models where you need them. You can model general coupled-resonator filters and with appropriate values handle crystal ones, LC ones or ceramics. Mis-termination of a multi-order coupled resonator filter usually manifests itself as increased ripple. If you model a quartz crystal fundamental and C0, you can build ladder filters easily. Use shunt coupling capacitors in a string of crystals in series to make a ladder. Make the coupling capacitors in the pattern of ratios as in that WW article by Monseiur Fochet F6BQP, and then play about with simulations of varying source/load resistance to minimise ripple and see what response you get. Scale all the capacitors, keeping the ratios and then play with the terminating resistors and you can empirically move your simulation to wider or narrower bandwidths. Bigger Cs mens narrower. You can have a lot of fun before getting the soldering iron warm. Designing a crystal filter by hardware hacking in simulation. It works! I've done overtone ones as well. A far cry from designing full-lattice crystal filters and then having to get a lort of different crystals made. Oh, yes, and with the crystal ladder filter, try adding inductance across individual crystals to tune out their C0 by resonating it at the centre frequency. You now get a symmetrical filter rather than theSSB one you started with. Jack 'JIR and myself had some good chats at Rochdale. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

30th Oct 2022, 3:49 pm	#45
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers I have now added series resistors to L1 and L2 in the MUI statement as Q killers and now I have it in a tuned circuit which gives the same results as the 2.2mH Inductor with a Q of 100. Looking at the the Top of the Coil 560k resistive at resonance. Looking instead at the tap about 60k. This is much to high to load a ceramic filter correctly. I think I would probably just add a 2k resistor here instead or there will be lots of ripple in the response.

30th Oct 2022, 5:45 pm	#46
Bazz4CQJ Dekatron Join Date: Apr 2009 Location: Oxfordshire, UK. Posts: 4,935	Re: Improving Selectivity In Vintage Receivers OK, sounds as though you are refining the requirements. I should have asked earlier - what is an MUI statement? B __________________ Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch.

30th Oct 2022, 6:29 pm	#47
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers In Spice it is the abbreviation for the Mutual Inductor Model. It is where you enter L1 and L2 Values along with either Mutual Inductance or Coupling Factor K. I was having trouble with the fact that it won't accept a Q value so the inductors will appear perfect. Since I have access to the terminals I have added series resistors there to damp the Q down to a likely real world value of about 100 given the coil construction. (Does anyone have an opinion on that?) I'm still thinking that resistive termination is the right way. Also still in favour of an emitter or source follower. That could be made to still look like 60k as load to the crystal and phasing control but it would give a low impedance output with unity Voltage gain. The low impedance output would drive the ceramic filter through a series resistor of 2k. The far side of the filter would have 2k to ground as termination and that would connect to the coil tap through a suitable DC blocking capacitor since that point is carrying AVC Voltage. Any comments about preference for FET or Transistor follower anyone? Any suggestions for suitable devices? Last edited by Jon_G4MDC; 30th Oct 2022 at 6:47 pm.

30th Oct 2022, 6:41 pm	#48
G6Tanuki Dekatron Join Date: Apr 2012 Location: Wiltshire, UK. Posts: 14,007	Re: Improving Selectivity In Vintage Receivers My go-to JFETs in times past were the J310 or U310 types , used in grounded gate mode to keep the impedances low and avoids the need for neutralising even up to 144MHz. They are rather expensive these days, because they are long obsolete. But still worth a look. __________________ I'm the Operator of my Pocket Calculator. -Kraftwerk.

30th Oct 2022, 6:45 pm	#49
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers Yes I have some I think. Maybe a TIS88 too. They seem to have a bit too much "go" for this though. Maybe a humble 2N3819? Idss range a bit wide 2-20mA but i have a few so I could select.

30th Oct 2022, 7:14 pm	#50
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers Playing also with LT Spice. I discovered you can make a sweeper using the modulator block. Setup a 0 to 1V ramp waveform over 0-1Sec and apply that to the FM terminal. Specify that 0V is 400kHz and 1V is 500kHz. It did this. Well it's a start! Attached Thumbnails

30th Oct 2022, 8:25 pm	#51
Bazz4CQJ Dekatron Join Date: Apr 2009 Location: Oxfordshire, UK. Posts: 4,935	Re: Improving Selectivity In Vintage Receivers Jon, you might to take a look at the FET circuit Radiofun uses after the ceramic filter in this video. He specs either a BF245 or a BF256. The video is 17mins 35 seconds long, the key bits are close to the end, but it's all worth watching, jumping forward occasionally!. https://www.youtube.com/watch?v=mTqXWB6RSOA B __________________ Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch.

30th Oct 2022, 9:23 pm	#52
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,901	Re: Improving Selectivity In Vintage Receivers 2N4391 2N4392, 2N4393 a set graded in Idss and Ron. Quite high current beasties. Used by Marconi in some of their better transistorised receivers. HP used the 20 Ohm one as an analogue switch.... rather a good one. It also does high dynamic range RF work as an amplifier. Bit pricey now, inevitably. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

30th Oct 2022, 9:30 pm	#53
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers That's interesting. I have not come across them. Many thanks David.

31st Oct 2022, 7:27 am	#55
Cobaltblue Moderator Join Date: Dec 2010 Location: Exeter, Devon and Poole, Dorset UK. Posts: 6,875	Re: Improving Selectivity In Vintage Receivers Three off topic posts removed. Cheers Mike T __________________ Invisible airwaves crackle with life or at least they used to Mike T BVWS member. www.cossor.co.uk

31st Oct 2022, 9:32 am	#56
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers Looking at the 3 FETs David suggested I think the 2N4392 might be the one to try.

1st Nov 2022, 10:01 pm	#57
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers Just for interest I decided to look at the HRO 1st IF transformer - I expect the others will be much the same. The variable capacitor measured ~80pF as it was tuned. So maybe the inductors are ~1.5mH or so to resonate with it + strays? There are no taps on these coils. There is a screen between the two capacitors and approx 1.5" spacing between the coils (presumably inductive coupling dominates). Attached Thumbnails

1st Nov 2022, 10:55 pm	#58
Bazz4CQJ Dekatron Join Date: Apr 2009 Location: Oxfordshire, UK. Posts: 4,935	Re: Improving Selectivity In Vintage Receivers I wonder how much the HRO was designed (in the early 30's) on engineering principles and how much on trial and error? That space between the coils is quite something! B __________________ Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch.

2nd Nov 2022, 12:50 am	#59
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,901	Re: Improving Selectivity In Vintage Receivers Oh, that 20 Ohm, highest Idss selected part was HP part number 1855-0020, you might come across them marked 5-020. Amazing what floats up out of old memory. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

2nd Nov 2022, 1:09 pm	#60
Jon_G4MDC Nonode Join Date: Nov 2015 Location: Welwyn Garden City, Hertfordshire, UK. Posts: 2,015	Re: Improving Selectivity In Vintage Receivers An idea is forming that might avoid the need for a FET. If the crystal filter is not required then the two 100pF capacitors which make the balanced feed which drives it could instead be made into a capacitive tap. That might be made to transform the impedance down to the 2k which the ceramic filter would like to see. I'm quite tempted to try this on the spare filter box I have.