My next project - the first French transistor radio

[Vintage_RC]

I've had a chance to do a few more measurements, these were all taken with the 2uF capacitor C25 removed.

Q3 b-e = +1.25V
agc line = -5.19V
Q3 e-gnd = -6.37V
Q3 b-gnd = -5.18V
Q3 c-gnd = -12.57V

I have attached the complete circuit but with the limit on jpg size readability might not be great.

Q3 b-e was +1.56V suggesting C25 had some leakage but clearly there are other factors causing the positive voltage on Q3 base. Any further suggestions welcomed.

[PJL]

Q5 is the next to be looked at. What are the ebc voltages?

[Vintage_RC]

Here are some results around Q5. In all case the -ve lead was on the left item so for example Q5 b-c means the voltage from base to collector with the meter -ve lead placed on the base. C25 was back in circuit but replaced with a 2.2uF Siemens film capacitor tested for value and leakage (non measurable).

Q5 e-b = -28.5mV
Q5 b-c = -4.13V
Q5 e-c = -4.17V
Q5 gnd-b = -38.3mV
Q5 gnd-e = -9.8mV
Q5 gnd-c = -4.2V
gnd-agc = -5.4V

I am still curious about R10 which makes the emitter of Q3 considerably more negative relative to the base than the Q4 stage which is near identical but without such a resistor. As an experiment I lifted one lead of R10 and the e-b went from +1.25V to -100mV. Encouragingly the radio then started working with moderate sensitivity on the MW and poor sensitivity on LW. This does not of course prove that R10 is the cause of the problem as presumably the set once worked OK with it in place although my research indicates that of the around 620 sets sold in 1956 about 160 had to be returned with problems. My set was made in 1957.

[Vintage_RC]

Subsequent to the above tests, with R10 removed so I could receive stations I monitored the agc line as I tuned through the strongest station I could find and there wasn't much change - about 40mV less negative.

[Herald1360]

The biassing of Q3 seems to depend on the AVC line voltage (approximately Q5 collector voltage) and the voltage imposed on its emitter by R10/R9 potentiometer between -12V and -4V. Q4 just gets AVC volts direct via R19. Everything is a bit odd thanks to the rails for Q3 and Q4 being the -12V and -4V lines, presumably because they're low voltage rated devices?


The whole shooting match seems to be rather unsatisfactory and likely depends more on leakage in the devices than any well defined base voltages and emitter resistor drops. Removing R10 on the face of it applies a far more realistic bias to Q3 and the set starting to work kind of verifies that. I wonder if the whole design wasn't a bit empirical with Rs tweaked until it worked and function is now upset by drifting values and device characteristics.


Presumably the AVC line will go more positive (less negative) as Q5 collector current increases with signal which would bias Q1, Q3 and Q4 back to reduce their gain.


I wonder if leaving out R10 and tweaking the values of R8 and R11 to get reasonable emitter currents in Q3 and Q4 (1mA?) would be a good starting point to go with?


If it's not too difficult/dangerous to the set I'd check every resistor involved in the IF, AVC and detector circuits for value against what marked on them. It could just be a build up of errors that is causing the problem.

[Vintage_RC]

I am currently creating the detector stage (Q5) in LTspice with a view to trying to trying to getting a better understanding of its operation and the voltage levels to expect.

[Herald1360]

I'd be very interested if you've got any decent Spice models for germanium transistors!

The only commonly found ones fall over on basics like VBE!

[Vintage_RC]

I have got some, I think they came from someone in Russia. I'll report back with the results.

[PJL]

I would expect q5 to have a near zero colector current without any signal.

[Vintage_RC]

I injected a modulated 130KHz signal at low level into the start of the IF stages and peaked the IF transformers which were a little off. Having done that the set is now performing well (R10 one leg lifted). The tracks on the PCB are fragile and this set is rare and an interesting example of a first generation transistor radio. I am minded (for now at least) to leave things as they are, it is now working more than well enough to demonstrate which is all I want from it. I have my Decca TPW70 as a daily user set.

[Vintage_RC]

Originally Posted by Herald1360

I'd be very interested if you've got any decent Spice models for germanium transistors!

The only commonly found ones fall over on basics like VBE!

Here are models for the AC128 and AC151. I have not done extensive tests but the Vbe is about 0.13V when I tried them in a simple amplifier circuit with a collector current of 1mA. Try them and see what you think.

[Herald1360]

Thanks for those models. A quick play with the AC128 one looks promising- at any rate the Vbe is plausible and it worked as a simple audio amplifier in my old version of Simetrix.

Got any Ge RF device models? I guess playing with parameters in the AF models might give plausible results.

[Herald1360]

Had a play with one of the models tweaked for higher frequency response in the detector circuit of the French set. It worked pretty well as a detector when fed with some AM RF at 1MHz. With the model used, though, the quiescent Vc on the detector was around -8.5V, rather higher than your measurements gave for the actual set. I suspect the detector circuit may be quite device dependent!


The circuit simulated is attached.

[Vintage_RC]

Originally Posted by Herald1360

Got any Ge RF device models? I guess playing with parameters in the AF models might give plausible results.

Attached is a model of the AF124, at 1mA Ic Vbe was 0.27V and in a simple LTspice amplifier it worked at RF frequencies.

[Vintage_RC]

I have played around a lot with my simulation of the Solistor detector, note that the IF frequency is 130KHz . I have attached the circuit as of the last run. The agc voltage tied in with my actual radio fairly well and it does act as a detector. AM modulation was achieved with three Spice oscillators; one for carrier, one for mod and then combined with a VCO.

[Vintage_RC]

The Solistor 8 is all back together again (with R10 mod and realignment) and is working rather well. I have attached a photo of the re-assembled set and you can see/hear a short video of it tuned to Radio 5 here:-
https://vimeo.com/user528768/review/...669/b9852ce9cd

[Vintage_RC]

Better to use this link for the video of the Solistor 8 in operation:-
https://vimeo.com/389087669

[PJL]

R10 and the similar R1 set the bias point so that the earlier stage gains are reduced first when the AGC is operating. I can't reliably make out the values of R1(10K)/R2(22K) and R9(15K)/R10(22K) so the calculations may not be accurate. Also, the battery voltages are critical particularly the 4v rail.

The total current through R17 is around 300uA.
Q5 is taking around 20uA.
R1/R2 bias around 44uA.
R9/R10 bias around 38uA.

That leaves the combined base currents of Q1/Q3 and Q4 as 200uA which sounds high.

[Vintage_RC]

The set is back together so I can't check actual component values or make any measurements at the moment. Looking at the full size copy of the circuit (which is not particularly clear), R1=10K, R2=150K (I think), R9=5K6, R10=22K. The 4V rail is actually 4.5V and the 12V rail is actually 13.5V

[PJL]

Not convinced about your 150K but the higher voltages means the base currents are closer to 300uA. What you are left with is a base leakage problem on one or more of the transistors so probably not worth pursuing if they are original.