Pygmy Club four band

[Mike Watterson]

There is an earlier 1954 Pygmy Club with LW, MW and 3x SW, that was revised in 1956. It has 1AC6=DK92, 1T4(=DF91), 1S5 (=DAF91) and 3Q4 (=DL95). It has the option for an auto-transformer based mains PSU that makes the chassis at live or neutral. It has a schematic, though the wiring of the 5-way rotary wavechange and coils etc isn't shown. It's 3V battery, 8V/6.25 mains LT, 67V battery, 75V mains HT. Cunningly folds the series chain in half, but then the  8V to 6.25V ballast resistor (33R) is then across the batteries! A B7G plug/socket is used to swap mains/battery.

My wreck is the 1957 model (may have come out in 1956). There is no schematic on the usual places like Radio Museum, Netherlands Vintage Radio or Doc TSF.

The main differences
DK96, DF96, DAF96 and DL96.
4 x push buttons on top instead of rotary switch at the back. So one less SW band. The DK96, unlike the DK92, isn't great for the shorter end of SW.
Similar case and scale, but changed for push buttons at top front instead of rear knob.
2 x D cells in parallel for 1.4V LT. It uses a different PSU obviously, though I've not seen one. However it looks like a similar arrangement.

Faults on first day
1) The on/off switch on volume control is broken inside, no HT on/off.
2) The 8uF 150V HT decoupler is bad, replaced with 22uF 200V, which is smaller!
3) Bad corrosion on battery power connection to main chassis and wire off. Fixed.
4) While there are many tubular ceramic caps, it has a load of crumbling French equivalent to "Hunts modseal" high leakage caps. The 40nF HT decoupler on an IFT was measuring 0.2mA and rising at 60V test. A 350V rated part.
5) A wire from LW coil on ferrite rod is adrift.
6) A wire from MW coil on ferrite rod is adrift.
7) Replaced switched volume control with a standard one, which amazingly fits though the shaft is a little fatter. Knob grub screw / knob seized on old one, but pot mounted in a slot so pot was able to be removed.
8) All valve sockets & valves with some corrosion. Used a very fine Dremmel tip/cone on the sockets. Also switch cleaner/
9) All the wavechange switches erratic.
10) A Silver mica cap lead was snapped off about 2mm from the tag in the Coil/Switch pack. Had to take out loudspeaker to get at it. How did that happen?
11) Speaker voice coil is seized. I have one the same size, though original seems like 3 Ohms. Mine is 8 Ohms.
12) LO tuning cap shorts over most of the range. Adjusting the locking grub screw at rear while using 2V 3A current limited PSU and wire disconnected made it mostly work. One bad point.
13) Both trimmers found  to be fully tightened on the tuning cap, well, they'll not get lost!

Cosmetic Issues
The scale outer is abraded. Maybe try toothpaste.
Only lowest section of SW whip still in place. I've a box of aerials, but none fit.
Some fine cracks. Might look OK cleaned.

Day two
Decided it's slightly similar to the Philips G77 model. Same valves, 67V HT, 2 x D cells in parallel and simple 96 series MW/LW. So reverse engineered Club and edited the Philips schematic:
   

The B7G power connector is viewed from the top, like an IC and shown twice.
The loose MW wire goes to same point as LW wire still attached, as that point is earthed on MW. The other end of MW coil is direct to tuning cap on the switch / coil pack unit.
There was a soldered tag with nothing on it. On LW that point is earthed, otherwise open. So I put the other LW wire there.
The ferrite rod picks up sig gen with one loop on MW and LW with those connections.

The usual HT feed of LO coils is via 33K. A 33K goes to the coil/switch pack and seems to be common to at least one "on" switch position for each band. The coil/switch pack has four coils, obviously the LO, one behind each switch. There are five silver mica caps. One appears to be in series with DK96 pin 6, the RF in grid. There ought be a 10nF to 100nF (the older Club has 40nF) decoupling the 33K, R2, but I don't see it. There should also be one or two extra coils for the two SW bands, I don't see them.

The IFTs have 455 KHz stamped on the base. It was closer to 456. But the final IFT wouldn't peak (L8), it was fully in. DF96 anode level was high and feed to volume control very low. C15 (220pF) seemed faulty. Level came up a lot with replacement 220pF. Strangely the DAF96 to DL96 coupling was the SAME type 220 pF ceramic tubular. It measured 240pF, close enough. This is usually 2nF to 10nF as it's audio.

I was suspicious that the internal cap on L8, shown as C14 on my circuit was either missing, open circuit or not soldered. I'd already had intermittent HT due to a 0V connection (R10 100 Ohms) never having seen solder! The R10 was over 450 ohms and R11 330 Ohms was close to 400, so both replaced. It would have put DL96 close to cut off. I suspect the mains adaptor shorts out the 330Ω or the 100Ω as the 330 Ohm is wired to unused pins 2 & 3 of the power B7G socket.
A 47pF had a slight improvement. 100pF more, but core still fully in. A 240 pF was too much, core almost out. A 180 pF across the L8 IFT terminals was just right on core position. So the original cap inside the can isn't connected or is faulty. The Sensitivity now about x10 better!

I replaced the 220pF with 2n2F. The older Pygmy has 5nF and 1nF to 10nF is common.

Current Status
It now works with the signal generator on MW & LW (one turn loop on ferrite) and picks up local interference if very close. Before it didn't at receive anything other than IF.
Both SW bands are picking up when signal generator is connected to whip input. But at about 1/2 the frequency!

It's very deaf. No stations, not even the local 252 kHz.

So I'm wondering what to try next. I've swapped the valves with "good" ones. No difference.

[Murphyv310]

Hi.
Nice write up.
Yes the 220pf AF coupling is way wrong. Has the set been damp stored? In the past I have had issues with battery valve sets with the oscillator coils and even the caps due to moisture ingress. It obviously tunes but is the LO way off?

[Mike Watterson]

The SW that is 50m to 40m (about 6 MHz to 7 MHz) is about 4MHz max. The 38 m to 20 m OC1 band is similarly about 1/2. (About 7 MHz to 14Mhz)

The MW & LW seem close to reality on LO, just deaf.

But first why no LW 252? There should even be BBC R4L, even if noisy. We'll see if anything on MW at dusk. I just sorted a 1935 McMichael 235 "Short Superhet". Hand capacitance about 10cm from the giant RF coil tower starts to bring in BBC R4LW. It's got a 2 m long wire as aerial now in the kitchen. It's one of these mad RF circuits MEANT for a short wire, it's rubbish on even a big loop. I did have stability issues till I bent a piece of coffee tin as a giant flat spring against the IF amp valve and final IFT. I guess the metallisation isn't connecting to the thin wire at the base, which I can't actually see. Massive air-cored 128.5 kHz IFTs . The double diode output pentode looks like a transmitter valve and is quite bright.

Maybe I'll try measuring resistances of RF paths in LW & MW tomorrow, except the tags of the switches are pretty inaccessible.

This is what one should look like: https://www.radiomuseum.org/r/pygmy_club_2.html

The older model is obvious from lack of any visible wavechange, it's on the rear. https://www.radiomuseum.org/r/pigmy_club.html

The "black" background is simply the hardboard everything is riveted to. I'd want to polish and varnish the brass scale cursor.

However unless something I've forgotten, or it's just the band switches, then it's for display only. I've always had trouble with band switches on Roberts like R600 and R700. I've one that's still intermittent.

Anything steel is rusted on the Pygmy Club. Also all the B7G connectors (power and 4 x valves) had the sort of corrosion you see only on battery sets, I guess vapour from dying cells.

[Murphyv310]

That'll keep you busy Michael.

[Mike Watterson]

Some photos:

The mysteriously snapped SM cap lead. But not AT the connection!
   
You can see the odd value 220pF cap between DAF96 anode and DL96 grid. Got to be a manufacturing error, like the never soldered 0V wire at the corner of transformer, which is an odd place to join 3 x wires and a 100 Ohm resistor.

The awkward to access switches and 4 x LO coils:
   
   
The coil/switch unit is riveted to the hardboard panel that screws on inside front half of case.

The two mounts for the ferrite rod are separately riveted to the hardboard panel.
The main chassis is riveted to the hardboard panel.
The speaker is bolted on!
The B7G power socket is riveted to the hardboard panel.
The audio output transformer is curiously mounted on the hardboard.

Chassis views
   
   
   

The HT+ is connected from the B7G power socket direct to 1st IFT. This then feeds pin 5 DK96 screen grid where the main HT electrolytic (replaced) is on pin 5 to chassis. Earth tags are generally not used, just a very powerful iron and a pool of solder on the chassis, so when replacing parts it's best to leave enough old lead to solder on.

[Mike Watterson]

Tried it from dusk. It does pick up a few UK MW stations. Very sharp tuning. Still nothing on either SW band or LW.

So I measured the few resistors. Apart from the coil/switch unit it's a very minimalist battery valve superhet

Using the designations on the circuit I edited:
(code@measured)
R1: 27K @ 33K,  DK96 g1 LO grid stopper
R -: 150  @ 180, DK96 g1 LO grid stopper
R2: 33K @ 28K, DK96 LO HT via coil to g2 (pseudo triode anode). No decoupler!
R3:  none, it's HT filter for DF96 on the Philips G77. The DF96 g2 is often fed from the R2, LO HT dropper/filter. Here g2 and IFT HT go to main HT.
R4: 3M3 @ 3M7, AGC filter feed. The AGC only feeds V2, there is no AGC to the DK96.
R5: 47K @ 76K, IF stop to volume pot. Now replaced. Little if any change.
R6: 1M pot. A new one was fitted due to faulty switch on the old one.
R7: 10M @ 11M3, DAF96 grid leak / bias.
R8: 2M2 @ 3M1, DAF96 screen grid / g2. The leaky "hunts modseal style" 40nF cap was replaced on the first day, that would have killed gain.
R9: 1M @ 967K, DAF96 anode load. C18 maybe should be checked as audio is muffled.
R10:  2M2 @ 3M1, DL96 g1 bias. Soldered between B7G socket and power switch HT- as it is switched. A wire returns to the B7G power socket where the R11 and R12 are fitted.
R11: 330. Was replaced early as it was about 400Ω. Part of DL96 bias via HT-. Mounted between  pins 2 & 3 of B7G power socket, but no plug connection.
R12: 100. Was replaced on first day as was about 450Ω. R11 and R10 giving about -3.5V for bias. Mounted between  pin 2  of B7G power socket and earth / OV common on the corner of audio transformer surround.

Still no LW, though works at about right frequencies on sig gen with 1 or 2 turns around ferrite rod.
MW reception was poor / quiet.
Buzz when touching signal points on volume control as loud as would be expected on 62V HT.

I've double checked this "minimalist superhet"
No RF / Mixer AGC (DK96)
No grid leak resistor on RF grid, pin 6 DK96. The pin 6 to ground:
OC1:  1.6 to 1.7Ω
OC2:  1.6 to 1.7Ω
MW:  2.4Ω
LW:  10Ω
All buttons out: 2.4Ω.

There are four coils at the four switches. The one at the LW switch looks to have too few turns for LW. It also seems to have alloy slugs. Perhaps this is the SW RF in coil for both SW bands.
The coil at the 2nd switch, OC2 (what I called SW1 earlier, 50m to 40m) looks more like MW or LW. The coil at the MW switch is similar. The coil at the 1st switch, OC1, does look like a SW LO coil.  A theory to test. Check signal on g2 of DK96 (even at 10:1 the scope probe on g1/LO tuning cap pulls the frequency down to much. See are there only three LO coils (the LW and MW could share a coil) and if coil at LW switch position, 4th, is actually a SW RF coil. It's not even impossible that there are only two LO coils, one for all of SW and one for LW and MW, as there are five silver mica caps at the switch/coil unit.

The valve sockets seem OK apart from the DAF96, which is slightly intermittent. The checking of LO at g2 DK96 and feeding sig gen via a 10:1 scope probe to RF section of tuning should establish  if the switch bank is working and what the four coils do.

So I won't give up just yet as it looks like it's now close to working?

[Murphyv310]

What a cheap looking radio, access looks poor too.
I've full confidence you'll get it going well again.
Keep us up to date.

[Mike Watterson]

Minimalist is the word!

Even if you removed the IFTs and drilled out the rivets to have more access, the switch tags and wiring are near inaccessible. However that's not uncommon. All the main chassis parts are easy. The worst I ever worked on for access were three Eddystones for a friend. Loads of waxies UNDER the giant switch wafers and under coils. I removed some coils and cut some waxies out with points of scissors closed by pliers. Tweezers and skinny iron to resolder.

The idea of hardboard rather than the steel of a 1950s Vidor or Ever Ready looks cheap, but actually a good idea apart from the weight. Some UK sets did use aluminium, but an amazing number of 1950s UK portables had steel panels, though the Sky Baby is plastic. The Sky Casket is pretty minimalist but a strange cabinet and expensive speaker.

[Mike Watterson]

As expected from visually examining the four coils:

Rear view of switches.
OC1 OC2 MW LW

The OC2 is maybe about 5.5MHz to 7.5MHz, perhaps a little less, I'd call that SW1!
The OC2 is about 7MHz to 14MHz, I'd call that SW2!

Rear view, 2nd coil from left is indeed MW LO as suspected, even though it's at the OC2 switch. Loads of LO on pin 3, g2, the pseudo anode of the Local Oscillator section. Nothing on the other switch positions. However driving input hard revealed:
1st Coil affects OC1 and OC2, as suspected.
2nd Coil is MW
3rd Coil affects LW.
4th does nothing.

So I'm thinking there HAS to be an RF in coil for SW. It's the 4th coil at the LW switch, because it needs to be as far as possible from the SW LO coil.

Next I make a discovery by accident. I had a 2.5m wire on the wire that originally went to the telescopic whip. I removed as I wanted to connect signal generator there. OC2 LO came to life. Even holding the insulation kills the LO. I connect signal generator by 47K to it and that works. As expected the 4th coil peaks the RF in OK at about 6.5MHz with tuning set open and first coil, at OC1 switch set for 6.5 MHz.

The coils behind the switch now make sense. Padding the two coils with SM caps can certainly change 40m to 20m (OC1 approximate coverage) to about 49m to 40m.

Now, is there a shorted cap, a short, open or faulty switch preventing OC1 and LW running properly and causing aerial wire to "kill" LO? It's progress, but the symptoms can't be fully explained by faulty switches?

[Mike Watterson]

I found a SM with a snapped lead going from R2 33K LO HT feed to 0V. So I removed the decoupler I'd added (usually that point has 20nF to 100nF decoupling and also is DF96 / DF91 screen grid supply. I reconnected that and the OC2 LO died, MW LO died and the part of band where OC1 LO was running died. But the LW LO came to life!

Something is wrong. The R2, 33K, has full length leads and they are bent round to put it on the coil pack terminal. Is it actually at the right place?

I decided to proper map out the coils and switches, so I took off the first IFT unit. I removed some of the SM caps noting value and location. All tested close enough.
Then I decided that was easy and I should take off the second IFT and also open it to find out what was wrong, because it absolutely shouldn't need 180pF added across the secondary. The secondary coil was open circuit! Duh. I should have checked continuity earlier, I was warned about that in about 1967. Nice little pot cores. The mica tuning caps are part of the coil supports. I removed the wire and there seemed to be about 230 turns of 3 or 4 core enamel Litz. My Litz is fatter, so only 120 turns fitted. I found a spool of enamel solid very fine and randomly wound on 240 turns. I didn't cut the wire, but just tinned a small patch. The primary is 1.34mH and 85pF. I unsoldered one coil wire to measure. The secondary capacitor is 56 pF and for a given frequency L x C of one tuned circuit = L x C of another. The inductance was 1.74mH, which seems a bit low for 56 pF so I added a load more turns.

Just random wound by hand. Didn't take long.

I verified by tracing the switches that my assumption where the LW and MW coil wires had come off was correct and I can see the first coil at OC1 position is shared with OC2. Also the RF switch contacts of OC1 and OC2 go to the fourth coil ar LW switch position. The coil at OC2 does indeed connect to MW switch and the coil at MW switch does go to LW switch.

Next time I'll finish tracing out the coil/switch pack and figure out what's wrong. At least the switches now seem OK. The OC1 and OC2 are three pole change over . The MW and LW are two pole change over. I guess the extra poles are to do with the SW coil at the LW switch and using it instead of the ferrite rod aerial.