Improving Selectivity In Vintage Receivers

[Radio Wrangler]

The extra transformer gives you an extra pole of LC selctivity and because crystals have spurious responses, you rely on the LC filtration to cover for these, so that's good. Also the non-transformer version has that choke supplying DC bias to the driving anode connected in an asymmetric fashion, the extra transformer fixes the asymmetry. But this form of filte doesn't work without special crystals whether there is a transformer or not.

Dvid

[Bazz4CQJ]

I respect your knowledge/expertise much too much to argue with you; the crystals are not there but have been replaced by ceramic filters and anecdotal accounts (of the QST original configuration) say these give "useful" performance.

This is backed up by Peter's account (on this forum and in his BVWS article) of his trial of the ceramics in an HRO. On that basis, the question is whether my proposed circuit has any obvious flaws, other than "lacking the quartz"?

Now, if I change the IF to 8.1MHz, my junk box certainly has a number of FT243 quartz crystals .

B

[Radio Wrangler]

The purpose of that transformer in the circuit in which it is shown iis to make opposite phase drives for the crystal lattice pair (Half-lattice if being precise)

If you went to a ceramic filter, the transformer could do some impedance transformation if that helps, considering the exact filter chosen.
These ceramic filters are piezo-acoustic devices driven by the voltage applied to a pair of electrodes. They are essentially balanced, but are usually applied unbalanced. If the filter has connectivity from input to output ot to case fro one input pin, then you can't drive it balanced despite the elements involved inside it being so.

The selectivity from the extra IF resonator may still be beneficial.

There is no right or wrong, just a lot of little compromises. It's a maze to be navigated were you want to lose least in the parameters you most value.

I may be mis-understanding the goal, which I think is to get a narrow filter suited a little more to SSB at the cost of less suitability for CW.

If you want a serious 6 or 8 pole SSB filter, Ive probably got a spare one you can have. Impedance will likely be 50 or 200 Ohms, so you'd have to do some transforming. The snag is they are quite big.

David

[Bazz4CQJ]

Many thanks for the offer, but I'm trying to keep it simple and fairly compact.
Yes, my intention is focussed on SSB rather than CW, but people do often talk about the AM that survives on 80m, though it tends to be around mainly about breakfast time when I'm never very "chatty" .

B

[Jon_G4MDC]

After Bazz started the thread about Hammarlund it reminded me that before the Christmas break I had a project in mind to fit a Murata CFJ455K ceramic filter to an HRO.

It is centred on 455kHz with a bandwidth of 2.7kHz. The shape should be reasonably symmetrical so BFO can be tuned high or low of it for USB/LSB reception. The in/out impedance is 2000 Ohms. I plan to abandon the HRO single crystal filter and use the LC components instead in a matching network to suit the new filter.

I found in my folder concerning the HRO the following measurements on the coil which feeds the grid of the first IF stage.
I forget now where they came from - it deserves proper credit if anyone knows. I have modelled this using the spice statement MUI 1 2 4 0 L1=1.13mH L2=0.15mH M=0.28mH where nodes 2&4 are shorted together as the tap. It tunes to 455kHz with about 67pF across it which ties up with other measurements fairly well. I have added 500k in parallel to represent the grid bias resistor.

The tap shows ~27k resistive at resonance. (Reversing L1 & L2 the result is 300k as would be expected with the tap at the hot end instead)

Of course so far I have assumed the inductor is perfect. In reality it has a Q of probably about 100. Is it fair to add a further 530k of swamp resistor to account for that? (XL is about 5.3k at 455kHz)

[Jon_G4MDC]

I have tried modelling Q by adding parallel and series R. No surprise that the result is the same.
It seems likely that the impedance at the tap is in the order of 10 or 20k depending what the actual Q of the coil is.

For terminating the filter I think about 2.5k to ground at the output is a reasonable place to start.

Next it's time to think about the mixer anode IFT. Lawrence has posted that the two coils are identical and 1.5mH
The coil pies are placed side by side so they touch. Does anyone have an idea what the coupling K might be in such a case?

It would be great to measure the inductance connected series aiding and then series opposing. I don't suppose anyone has done that? We could get to M that way.

[Bazz4CQJ]

Re the Murata CFJ455K, have you got one in stock Jon; they seem pretty rare now, I found one guy in the US selling them at ~£70, inc £16 postage.

Re the mixer transformer, I guess those two inductance measurements could only be done with the two caps on the secondary disconnected? I've just held mine against a light and there is a space of about 1mm between the windings, but I guess that they could have been manufactured with variations over time and by different makers during wartime. I plan to re-use that transformer in the Ham. set up.

Each time I handle that assembly, I feel that the looseness of the wire from the coils (they are appreciably longer than they need to be) makes me feel they are vulnerable. I wonder if yours is the same and if you have any ideas about how they could be made more secure? A small dot of glue from a glue gun to fix the mid point of the lead on to the coil might do it, but I'm not at all sure about doing that.

B

[Bazz4CQJ]

Well it's nearly 4am and I am on the prowl and wow wee what I have I found; back in 2014, there was a thread on this forum by Andrew (GW3OQK) who had put a pair of (3-legged?) ceramic filters in to his 1155 and reported how amazingly good the outcome was. Moreover, he later did the same thing with an HRO.

He didn't show a diagram, and the written description is not "crystal" () clear, I will PM him and ask him to join us. He was last on the forum in March '22.

https://www.vintage-radio.net/forum/...d.php?t=107140

B

[Jon_G4MDC]

Yes I already have the filter. Actually I should have two of them, I know I bought a pair when they were plentiful. No idea where the 2nd one has gone. At those prices I should put more effort into finding it!

If will be good if Andrew can tell us first hand what he did. I joined in Nov 2015 so that thread had probably gone well down the list by then.

My mixer anode assembly is shown below. Definitely no gap between the coils.
I agree about the winding ends but it's been OK for ~84years. I'm not going to put them on the vibration table.

I'm wondering whether I can estimate the mutual inductance of those coils using a GDO. I have no other method to measure inductance.

PS Happy New Year to all.

[Bazz4CQJ]

Looking at mine again, the two windings are not quite parallel to each other; there's a distinct gap on one side but less on the other. I was always suspicious that the filter in my HRO was 'faulty', which is one reason why I checked the crystal a few days ago.

I decided against the glue gun but have tacked down the leads using some Shellac.

Andrew has not replied to my PM, (but perhaps not too surprising given the time of year), though the fact that he has not been on the forum for 9 months is a concern. The description of what he did back in 2014 is not quite clear enough to be sure about how to repeat it. He did say he was using the cheap ceramics (not the more expensive ladder types) and the outcome he reported was very encouraging. He didn't seem to get deeply involved in considering impedance matching. The title of his thread just referred to the 1155 IF and the ceramics were only referred to in the text so it had not shown up in "Title" searches.

You raised the question of testing any filter we make, and that is an issue, and again Andrew seemed to have done that but does not give any indication of how he did it, but it seems like he did it with the ceramics installed in the set.

Re the inductance values, I could get a value for the primary on my Wayne Kerr, but not keen to start disconnecting the caps on the secondary. Maybe use a GDO and check that on a frequency counter to get accuracy?

B

[Bazz4CQJ]

Hi Jon. I've done some measurements with my Wayne Kerr bridge, (see attached). Hope those make sense.

B

PS have just recalled the info posted by Lawreance in the Ham. thread;

If it's any use to anyone, some general info I have on file for the HRO, specifically the HRO5T (Xtal hard wired in the phasing unit) for the transformer that couples the Mixer to the Xtal filter, and the coil that couples the filter to the 1st IF amplifier:
Coupling transformer from Mixer to Xtal filter:
Primary:

10/41 Litz....235 turns...3/16" wave wound...1.5mH...7.3 Ohms.

Secondary....Same as above.
Coupling coil to 1st IF amplifier:

10/41 Litz....297 turns tapped at 233 turns...3/16" wave wound...1.93mH total (297 turns)...1.48mH (233 turns)...10 Ohms and 7.2 Ohms respectively.

The numbers don't quite match my measurements; perhaps the effect of the two caps being on play?

[Bazz4CQJ]

I’ve measured the resistance of the coils. I used a Unit UT30B DMM (which is not pulsed) and I checked it against my Avo8 on a 10R 0.5% resistor. The DMM measured the resistor at 10.0 and the Avo at about 10.1R.

My mixer transformer measured 8.4 primary and 7.2 secondary. The data from Lawrence says 7.3R on both sides.

1.2mH needs 102pF to resonate at 455kHz, and 1.5mH needs ~81pF. So on the secondary, with the two fixed caps and the compression trimmer and all the rest of it, both L values look viable.

The data may not be as consistent as I suppose we would like it to be, but we are looking at components which are ~80 years old. War-time tolerances were probably relaxed.

B

[Jon_G4MDC]

Thank you Bazz. A very interesting set of measurements.

Ideally the two capacitors would have been removed before taking them but it seems to have had very little effect. Possibly because the 10kHz test frequency is so low?

Both primary and secondary measured the same inductance alone. Can I assume that the two values for series connection are as follows
3.79mH the two windings connected in aiding "phase"
1.1mH the two windings in opposing "phase"

If that is so then the value for M that I calculate is 0.695mH aiding and 0.65mH opposing. I don't think that's at all bad. Maybe the slight disagreement is an effect of the capacitor being in circuit. Either way I'll take 0.675mH and be happy with it.

I'm now looking at the simulation for the mixer. I have modelled the inductor as before. The secondary is tuned the primary is untuned with a resistor across it to represent the load of the 6C6 mixer Anode. I want to look back from where the filter would be connected on what might be a capacitive tap on the secondary. Any ideas on a representative value for that anode resistance?

Thanks again!

[ms660]

Re: HRO coils, the tolerance of the coils for the R106 MKII (aka HRO-5T) is given in one of the EMER manuals for that receiver, the coil data starts on page 1001 of the manual for all the RF/Osc coils, IFT's, BFO etc, the tolerances are given on the last page:

http://www.vmarsmanuals.co.uk/archiv..._EMER_E164.pdf

The R106 MKII was the last HRO I owned, I had a full set of coil packs for it.

Lawrence.

[Jon_G4MDC]

Thanks for the good gen Lawrence. I'm sure Bazz will find that useful.
For want of any better idea I have taken the published Ra of the 6C6 mixer valve (1 MOhm) and plugged that into the simulation of the Mixer Anode coils. It suggests that the impedance seen across the tuned secondary will be 200k. This ignores the capacitor centre tap so it's a single ended value.

I'm pondering whether a 10:1 Turns ratio transformer is a practical way to get to the 2k the ceramic filter would like to see. Could it be done broadband with ferrite at these impedances / frequencies or maybe it's better to think on another tuned circuit with a tap. I tried the idea of a capacitor tap but it left a reactive component at resonance - for me a forgotten feature of the technique.

[Radio Wrangler]

2k and a 10:1 turns ratio gives 200k Ohms impedance on the full winding section. For a transformer, you want Zl to be at least a few times this, say 600k minimum At 455kHz that's 209mH and would resonate with 0.58pF so you won't be able to do this without self resonance problems.

The only escape is to accept that there will be resonance and abandon the idea of an untuned transformer. You can now use lower L values, the strays would push the SRF up above the target frequency, but then you need more C to resonate the transformer, and that C can gobble up the stray.

Remember that crystal, ceramic and mechanical filters all have various stray responses, so some LC selectivity helps cover them.

David

[Jon_G4MDC]

Many thanks David. That makes perfect sense. I'm coming round to the idea that an extra winding over the secondary might be the best way. 10:1 turns ratio would require about 24 turns not allowing for imperfect coupling. It might make something to play with. I wonder if it would be better to wind over the existing pie or side by side with it?

[Bazz4CQJ]

Thanks to Lawrence for the R106 data; I do have that Base Service document but had not recalled the data about the coils being in there.

In terms of adding an extra winding Jon, I wonder of you might use a sleeve of something that could just slide over the existing polystyrene former, that would make it easy to experiment with? To begin with, maybe just use some paper with a coat of PVA, varnish or Shellac? Alternatively, some heat-shrink, formed on something of the same diameter? I think you'll need to disconnect the 2 caps whatever you do.

No word from Andrew (ceramics) yet.

B

[GW3OQK Andrew]

Hello Bazz
The HRO mods were published in a past Signal but I didn't go into great detail, just the paragraph below. In the final event I am not satisfied with the skirt bandwidth for it is still too wide for CW, which is all I use, and probably too narrow for voice. To work in a proper block filter might be the answer but I don't think its worth the bother when our modern radios have such steep sided IF filters.
73
Andrew

Fitted a narrow band 456 kHz filter in the IF chain which usefully reduces the skirt bandwidth. This consists of 2 tuned top-coupled 455 kHz ceramic resonators on a strip of PCB mounted in a diecast box under the chassis. The HRO’s crystal filter works very well giving nice single-signal reception. Bandwidth is approximately –55 db at +/- 2KHz. (I did similar to my R1155)

[Jon_G4MDC]

Just taken a better look at Lawrence's R106. Very nice especially the box of coil packs.

It's an idea about the paper sleeve Bazz, I might try that.
I think I might leave the tuning caps as they are and just leave the crystal unplugged. I might disconnect the phasing control although I hardly think it's necessary.

I think I will play with the mixer side first.