Re-purposing a three gang tuning condenser

[Jim McLay]

Hi Peter,

Thanks for that. Will do. Hopefully I'll have enough gain to make this a necessity!

I have found nice aluminium-backed 3mm PVC sheet which is a lot easier to work with than solid Aluminium. I think there's more in the mechanical construction than anything else. Plus I'll be doing some mahogany woodwork.

Cheers

Jim

[peter_scott]

Hi Jim,

I did find one picture that I think shows the screening of the Music Magnet condenser.

Peter

[Mike. Watterson]

The thickness of aluminium affects screening properties.

[Radio Wrangler]

The main problem with a TRF receiver arose when the number of stations had risen to the point when they got hard to separate.

The first assaults on the problem were attempts to make receivers narrower by increasing the number of tuned circuits, by using exotic materials to make the Q of those tuned circuits greater, and also by employing valves to apply a little feedback to cancel part of the tuned circuit's losses, and so to multiply the effective Q.

Q is an interesting property, it can not only be viewed in many ways, it can also serve as a handy way to relate things seen in one view, to things seen in another.

If I have a tuned circuit and I plot a swept frequency response, I'll get a bell-shaped graph. If I measure the spacing of the half-power points of the bell (aka -3dB points, also 0.707 of peak volts, also 0.707 of peak current) and call it the bandwidth, then the ratio of centre frequency to the bandwidth is equal to Q.

If I drive the tuned circuit with a bit of current applied across the L or the C, and measure the current flowing in the loop at resonance, it's bigger and the ratio is Q

If I insert a voltage source in the loop and measure the voltage across L or C, it too is bigger, and the ratio to the driving voltage is Q. Q used to be called 'Circuit Magnification Factor' but Q is snappier..

If I made a perfect resonator with a perfect L and a perfect C and then spoiled it by clapping a resistor across L or C, the the ratio of the resistor divided by the capacitive reactance at resonance is Q. And at resonance, the inductive reactance equals the capacitive reactance (though opposite phase) and so it's also true for the inductive reactance.

If I made the perfect resonator and spoiled it with a little series resistance added to the loop, then the ratio of the reactance of L or C at the resonant frequency to the added resistor is Q again.

If I calculate the energy stored in a resonator, and then divide that number by the amount of energy lost per radian of phase worth of time, Yup, ti's Q again.

So you can flit from one approach to another, just using Q as the key to relationships.

For a capacitor, if you know its Q, for a given frequency you can calculate its reactance then multiply by Q to get its parallel equivalent resistance, or divide by Q to get its equivalent series resistance. Same goes for inductors.

The Q's of connected components add as reciprocals, just like resistors in parallel.

Cardboard formers can give quite a good Q when dry. The trouble is keeping them dry.
There are specialist glasses which are RF lossy for screening, but just about all others have good Q. There are good ceramics (and lossy ones) ditto for plastics.

The other cure for the problems narrowing the TRF receiver was to recognise that narrowness was easier to achieve at lower frequencies, so why not make a nice narrow TRF receiver at a low enough frequency to make it easy, don't bother making it tuneable, just screw a frequency converter on the front of it. Hey presto, the superhet is born. And there is a TRF receiver lurking inside every superhet.

One nice thing about superhets is you don't have to have all your gain on the same frequency, and so keeping the thing stable is easier.

David

[Mike. Watterson]

^^^ What he says.

Q= Quality Factor

It applies also EXACTLY to loss in a sprung or oscillating mechanical system too. A hi Q pendulum needs less power.

https://en.wikipedia.org/wiki/Q_factor

Interesting idea
http://users.tpg.com.au/users/ldbutler/QMeter.htm

[Jim McLay]

Hi Mike / Peter / David,

A wealth of good stuff there. Many thanks guys for all the ideas and pdf's. Who needs text books when you can find it all on Vintage-Radio?

Screening is definitely a must. Seeing as I'm starting from scratch, it should be easy enough to get the coils (whatever they are wound on!) into their own compartments, and hopefully close enough to their respective tuning condenser sections.

Heck, this beats real work!

Cheers

Jim

[Jim McLay]

If I wasn't so unobservant, I'd have seen the makers name not once but SIX times engraved on the (bakelite?) supporting frames. According to Radiomuseum, they were

Bowyer-Lowe Co., Ltd.,
Radio Works,
Letchworth,
England.

Supplier of Components and Wireless Sets in 1927 and 1928.

Getting there slowly!

[Jim McLay]

Originally Posted by Mike. Watterson

There is. 100K is too small
Also a scope probe often has too much capacitance.

The Radio Wrangler has given some pointers.

Q varies for frequency too. The capacitor value and frequency of test depends on inductance and application.

Note that thin wire or many turns reduces Q due to resistance and also skin effects making the resistance higher than at DC. Hence at higher frequencies silver plated tubing works better than solid copper wire. Paralleled windings reduces resistance and increases Q. Multiple fine enamel or
Litz of same weight of copper per metre as solid single enamel of same DC resistance has better Q due to Skin Effect.

This isn't so easy to do properly.

Hi All,

Well, per Mike's advice I have used a 1 meg resistor to measure the unloaded "Q" with my crude method, driving a tuned circuit consisting of a coil of 67 turns of 0.5mm enamelled wire wound on a 50mm cardboard tube. The inductance is 170 uH approx.

I tuned it with a 500pF metal framed variable capacitor in parallel set to approx 250 pF. getting resonance at around 890 kHz with a Q of 110.

I am using a Tek TDS3012 to measure rms voltage & frequency, with a 10:1 probe which probably presents about 10 pF to the circuit under test.

I then tried stuffing a wad of damp tissue into the coil but did not see any significant difference in the Q.

A rough wrap with aluminium foil caused the Q to fall to 55, and if I placed the foil at 25mm on one side of the coil, the Q is 88.

My conclusion would be that the screening arrangements will have a much bigger impact on the Q than the coil former properties.

Any comments?

Jim

[Mike. Watterson]

The aluminium changes the inductance.

Screening is supposed to be at least 1/2 radius out from edge of coil and at best 1 radius.

Philips at one stage used air cored coils slightly too high inductance and fixed capacitors and fine tuned the inductance by denting the aluminium screening can.

Steel/iron shielding raises the inductance and reduces Q a lot due to eddy current magnetic losses. Only Coils in a ferrite or dust iron pot/cup use tinned steel case. Everything else thus uses Aluminium, Brass or copper at suitable spacing. Coils with ferrite or dust iron cores can use closer aluminium shielding than air core, which however have higher Q if lower loss winding/wire is used as the wire length/turns are higher for air core.

170uH is low for MW.
MW RF coil is typically larger inductance than that. LW RF typically uses about 2.5mH

An 455 KHz IF coil about 1.1mH to 1.5mH

The larger the inductance, the more there are issues with wire resistance, (skin effect and actual size), former and core losses.

I'd use about 150 turns of 20 strand or better Litz.

Where is the 1M resistor?

Search out some of the specialist crystal Radio sites to learn about low loss coils.

[Jim McLay]

Hi Mike,

Well, the 3-gang allows me about 1.5 radius gap with the 50mm coil former size.

I was wondering how best to tweak the 2nd & 3rd coils' inductances to get the tuning correct over SOME portion of the band. It should really be done with the screening in place, which would be very tedious if it involved adding or removing turns. I wonder if I could use a plastic rod with a tiny piece of ferrite inserted inside the screen but adjustable from the outside.

My low inductance came about through not reading the coil winding details (W J May, The Boy's Book of Crystal Sets P23, carefully enough! Needs rework.

I ought to use Litz. I'll have a gander & see if I can buy a reel. Having been building switch mode power transformers, I've seen the advantages big time.

The 1M resistor is in series with the (voltage) drive from my signal generator. It provides a cheap current source to drive the parallel tuned circuit.

I appreciate your point about low loss coils, and am balancing the need for best quality design with the time it will take to complete the set . . .

Cheers

Jim

[Mike. Watterson]

Paralleling 8 strands of fine enamel wire is a cheap alternative to Litz.

There are online calculators that tell you length, inductance etc.

Just wind slightly low and pad with trimmer capacitor, or slightly high and push in shield to reduce inductance.

[Jim McLay]

I've been hunting for some details on the RF transformers needed to go with the 3-gang capacitor and finding commercial types like Repanco, Osmor and Denco, all of which are obsolete, and their datasheets tell you very little, and certainly not what turns or turns ratios were used. I imagine these were carefully-guarded trade secrets, values acquired by much design and test.

Other older posts show similar difficulties so I'm not the only one looking for a "cookbook" solution to avoid having to go back to basics!

As far as I can tell, the aerial coil was generally 1:1 with the tuning coil, and often wound side-by-side with it. In TRFs like the Music Magnet, a series variable capacitor was used in the aerial connection, perhaps to attempt to match whatever aerial was being used to the front-end impedance of the 1st RF stage.

Or something.

The RF transformer coupling the anode of the first valve to the grid circuit of the 2nd is similarly clouded in mystery. I've trawled the Forum with various combinations of search terms, each resulting in lists of Forum threads to look through. Trouble is, coming from a transistor background, I don't have any real idea about the impedances to be matched. Again, it could be that 1:1 was generally used.

At some point, I suppose I'd better put up an aerial . . .

[ms660]

Not a cook book but various tricks/practices are to be found within, well worth a read:

http://www.paleoelectronics.com/RDH4/CHAPTR23.PDF

Lawrence.

[peter_scott]

Originally Posted by Jim McLay

In TRFs like the Music Magnet, a series variable capacitor was used in the aerial connection, perhaps to attempt to match whatever aerial was being used to the front-end impedance of the 1st RF stage.

That series capacitor was just the Music Magnet's volume control.

Peter

[Jim McLay]

Hi Lawrence / Pete,

Thanks for the latest tips! The paleoelectronics link is full of stuff I used to dream about.

The more I see, the worse it gets! The article in Lawrences link tends to say that you can adjust that "volume control" aerial capacitor to maximise the signal by "tuning" the aerial. I suspect that when the cap is reduced to a few pF on MW / LW, the volume would indeed fall to very little, hence the name in the MM4.

I've found a TRF by Robert Thornton (PW,k May 1980) which goes further with aerial tuning in providing taps on the primary of the aerial transformer as well as a series variable C to allow almost any aerial to be matched to the RF amplifier grid circuit, and I think I'll do this, and do volume control at AF after the detector.

It's beginning to look difficult to get the three stages to track "properly" because things you do with the aerial coupling will tend to affect the tuning of that stage relative to the other two. It could be that too high a "Q" in a tuned circuit could actually work against me similarly. However, I think that "suck it and see" is going to be the guiding principle, in the lack of concrete data about impedances, coupling factors & almost anything else.

Jim

[Jim McLay]

Light has dawned. Eureka? or maybe a senior moment.

Anyway, I was looking at the aerial circuit for the Philips 634A, hunting for ways to use the 3-gang condenser, and suddenly realised that it is a band-pass filter identical to circuit 27a in The Boys Book of Crystal Sets.

So.

My strategy is now to build the crystal set, get it working, and then work on a subsequent valve tuned RF stage using the third section of the 3-gang. (I mention the 3-gang repeatedly since this is the original reason for the thread. I am scared of moderators!).

[Jim McLay]

For those of you following the fate of the 3-gand condenser, here is a stab at a do-able circuit diagram, which I've pulled from various sources. The front-end is crystal set, and the rest is basically Osram Music Magnet 4.

If I've got it right, the first 2 gangs form a tuned band-pass filter with L6/ L11 for MW, with L8 and L13 switched in for LW tuning. The centre-tap on L11 may have been the crystal set designers effort at matching into a pair of 2,000 ohm earphones.

The third gang tunes the secondary of the third coupling transformer primary L9.

The three transformers are to be identical.

V2 is being operated as a detector with reaction being provided via L14, loosely coupled to the third transformer. AF is passed via HFC L15 to an audio transformer T1 whose turns ratio I have yet to unearth. V3 might drive a high impedance set of phones, or use another transformer with a low impedance load.

[Jim McLay]

Intervalve 1:6 step-up and output transformer 60:1 step-down for 3 - 8 ohm speaker) suggested (R Thornton PW May 1980)

[G6Tanuki]

I'm a bit confused about the wiring of the resistor/capacitor to earth in the front end tuned-circuit: this will essentially appear in series with the antenna at 'hum' frequencies and because the capacitor will be of relatively high impedance to such frequencies it could couple the hum into the grid of the first stage.

Better to miss these components out, allowing the bottom end of the coils to be connected directly to earth.

You may want a grid-leak R/C combination [100pF and 1MOhm in parallel] in series with the feed to the first-stage grid.

I'm also confused about R4 - it's in parallel with the coil and so will b essentially shorted-out at DC/bias voltages. Is it needed?

I'd connect the 1MOhm resistor between the grid of V2 and the filament -ve rather than to the centre-tap? of the cool.

Your screen-grid resistors - at 1K)hm - seem rather low values too - I'd be expecting something in the tens of KOhms here.

[Jim McLay]

Hi G6,

Many thanks for the feedback. You've made some good points here.

That R1-C4 is (apparently) what makes the front-end into a band-pass filter. The coil assembly L4 is screened from L6, and I think that L4 - C4 - L6 form a T-section filter. R1 appears in the crystal set circuit at 1000 ohms and the value chosen may have to do with the passage of AF signals to the headphones. I imagine a normal grid leak of 1M should be put there.

The Philips 634A uses a similar band-pass filter, with some megohms in the R1 position being used to provide AGC to the grid. It also placed R4 across the LW coil to dampen the RF response; I suspect the set could howl under certain circumstances without it.

C8 and R5 and the "centre tap" on the coil, again seen on the 634A, mystify me. The 634A doesn't give any coil details but shows a tap. The actual centre tap comes from the crystal set design, and may have been the crystal sets way of connecting the tuned circuit to his germanium diode detector / headphones without loading the tuned circuit too much. The crystal set book states that headphones must not be less than 2,000 ohms impedance.

Sorry about the 1k resistors, the CAD program defaults to 1k when you add in resistors. I'm glad you were able to see the thumbnail, it looked a bit poor on my screen.

I've no doubt I'll be making some changes before the set is finished!

Jim