Pye Black Box amplifier PCL83 version

[Michael Maurice]

Hi All,

Last year I serviced one of these, replacing resistors and capacitors as required, that is all the Hunts and the cathode bypass capacitor.

A couple of weeks ago it developed a fault and basicly fried the O/P transformer. I sent that off for rewinding and was told that it had shorted to the frame.

I've now re-installed it in the amplifier along with new PCL83's expecting the amplifier to work but the amplifier is oscilating at around 250KHz (3-4 divisions with the scope set to 1.0μS) and around 4V p/p with the speaker connected.

There apeared to be some smoke coming from the rewound transformer. HT is around 230V.

Obviously I switched off immediately and checked my work.

The pentode grids are at zero though one of them went to around -15V before settling around zero. All other resistors check out OK. There is no discernable ripple on the smoothing caps.

I did try reversing the phase of the speaker connections which made it howl, so its not that.

Does anyone know whats going wrong?

[ukcol]

If memory serves correctly, I read that this particular version of the BB rather than having the usual single negative feedback path has multiple paths (3?) at least one of which is positive. Perhaps your example has a fault condition that has caused the loop gain of the positive feedback path to exceed 1.

[Biggles]

It may be worth adding a 100pF to say 1nF capacitor from each pentode grid to earth to calm it down, or maybe try adding stopper resistors in anode connections. Possibly the new transformer windings have different characteristics to the original. You may just have to be content with slight modifications to the circuit to get it to behave. Of course it is possible that the oscillations may have caused the original transformer to fry. Power supply going high impedance may be the cause. Try decoupling it with another capacitor in parallel to the smoother, or you could try the original valves back in.

[Ed_Dinning]

Hi Mike, I had one of these that was oscillating, not as badly as yours but had caused things to burn out over time, we scrapped it in the end as it could be tamed, but did not then sound good.
I believe it is this model that has the C/R networks shown connected anode to screen, not anode to HT in the company schematic. The consensus is the they should be anode to HT.

Ed

[Michael Maurice]

I've always seen these with the RC network connected anode to HT.

The question is why does it oscillate in the first place? It did work when I first serviced it some 6 months ago then the transformer failed.

Is it possible that the transformer has be rewound incorrectly?

[Focus Diode]

Not impossible for the rewound transformer to be faulty but it appears the original fault condition that caused the initial failure persists.

One of the replacement capacitors failed I wonder?

[Radio Wrangler]

It is possible that this variant doesn't have much stability margin in its feedback loop. If so it wouldn't take much additional phase lag or increase in gain to let it oscillate. One of the factors which has always been the final limitation on how much feedback can be applied in a valve amplifier is the leakage inductance of the output transformer. Has the rewind been done true to whatever arrangement the original had?

But as Focus Diode says, there may well be a different cause still in place. A slightly inductive resistor where there once was a carbon comp? 250kHz sounds about right for the full loop to be unstable rather than a localised instability of one device.

I don't have a copy of the circuit, but if there are multiple feedback paths then I'd get a bit concerned about just how well the original design was done. If some of the feedback is positive, then that's sometimes done to bootstrap load impedances to push the gain up. Quite common in basic transistor designs, unusual with valves.

David

[Edward Huggins]

The original 5 stage circuit was fairly stable, but the Boy's at the back of the room of Radio Development in St. Andrews Road kept on tweaking. Sometimes what appears before you on the chassis may not conform to the circuit!

[Michael Maurice]

The resistors I've used come from RS and are Vishay metal film type 2 watts and 500V
The capacitors are Vishay axial polyester 400V. 47nF replace the original 40nF.

Most of the resistors were within spec and were left alone.

[Steve_P]

Check the voltages on the PCL83 anodes - if they are different - even by a small amount _ then I would suspect C13, C14, R27 and R28. (Pye Manual),

Try taking off the feedback loops.

It is possible that the Transformer has been wound wrong or is still faulty. But check R30 and the speakers too.

Cheers.

Steve P.

[ukcol]

Originally Posted by Radio Wrangler

I don't have a copy of the circuit, but if there are multiple feedback paths then I'd get a bit concerned about just how well the original design was done.

Here is the circuit.



It looks like there are 3 feedback paths.

1] One from the top end of the output transformer secondary. (Negative).

2] One from across the low value resistor (R30) in the bottom end of the output transformer secondary. (Positive).

3] One from the control grid of the upper output valve V2B. (Negative).

The polarities of the paths are my interpretation and need to be check by someone intelligent.

[Ed_Dinning]

Hi Gents, note that this circuit shows the C/R networks from anode to screen, not HT.
Note that on some of the amps I've seen R30 is a short length of Nichrome wire.

Ed

[Radio Wrangler]

The anode to screen dampers on the output valves probably are there to act as dampers on the output transformer, and are done as a pair because of leakage flux effects in the transformer.

More interesting is R22 feeding back the anode of V2B to the grid of V3B. It looks like a neutralisation scheme in an RF PA!

R10 looks to be the main feedback path to the cathode of V1A from the speaker output

R29 looks to sample the output CURRENT to the speaker and feed it back in the opposite phase to the cathode of V1A

Then there is another feedback path from the grid of V2B back to the tone controlery leading into the grid of V1A and also to the cathode of poor old V1A.

I'm sorry. This circuit defies straight-forward hand analysis. It needs reasonable models of the valves (including screen transition effects of the pentodes) and throwing at spice. I do not believe this was created scientifically back in the fifties. It looks like a kludged dog's breakfast with added in tone controls and then layers of sticking plasters to try to stabilise the monster.

This doesn't help Michael, I'm afraid.

I think I would convert it to a simpler circuit with simpler feedback and lower loop gain. The tone controls would become a passive section ahead of the amplifier.

No wonder no two seemed quite the same and they did so many versions.

David

[Synchrodyne]

It looks as if the PCL83 version of the Pye Black Box embodied the same idea as was found in the Pye HF25 Proctor and HF10 Mozart hi-fi amplifiers. These included variable positive feedback (current feedback) as well as customary negative feedback in order that the damping factor could be varied to suit the speaker in use. I think this was an idea that was current in the USA in the mid-1950s, and it was mentioned by G.A. Briggs in his “Loudspeakers” book. In the case of the PCL83 BB, the feedback ratio was presumably fixed given that the fitted loudspeaker was a constant. That facility looks to have been dropped for the stereo version of the Mozart, the HFS20, but whether that reflected perofrmance problems or cost-shaving I don’t know. (Perhaps Edward H. Might have some insights there.) The PCL83 BB appears to have added a tone control feedback loop in addition to the negative and positive feedback loops, perhaps a carryover from the earlier EL42 BB, although the latter did not have the positive feedback loop. I haven’t tried, but maybe one could figure out the various PCL83 BB feedback loop functions by comparing it with the HF25 and HF10 on the one hand, and the EL42 BB on the other.


Cheers,

[Michael Maurice]

At the moment, I'm repairing just the amplifier on the bench and using the bench speaker to test it with. I have done this before with no problems.

Is anyone suggesting that the speakers might be the cause of this?

I have never seen R30 in this amplifier, I could find a small choke and wire it in, say 10 or 20mH and see what happens.

I might when I've got the time, substitute a transformer from another Black Box and see whats happening, in the meantime, I'll do some voltage checks as Steve P suggested and respond when I can.

Regards

Michael

[Radio Wrangler]

I missed one bit, that current sensing means you need a model for the speakers as well.

David

[Heatercathodeshort]

Don't forget that the UCL83 and PCL83 in particular are very prone to instability and it only takes a slight variation in circuit design or layout to instigate this.
They were a very temperamental valve when employed as frame osc /output and audio amp/output in mid 50's television receivers and were soon replaced by the PCL82. Not a good move by Pye to use up their stocks of the PCL83 in a version of the Black Box.J.

[Steve_P]

I wonder how much the circuit has to be modified to take a PCL82?

The Bases need to be rewired but apart from that...? On Second thoughts the heater voltages....

Just a thought

Cheers,

Steve P.

[ukcol]

Originally Posted by Michael Maurice

I have never seen R30 in this amplifier........l

According to the manual R30 comprises four and three quarter inches of 22 S.W.G. Eureka resistance wire. I imagine it must be wound on a former and may look like a small choke. The actual value of resistance is not given and so anyone attempting to model the circuit will need access to a Eureka resistance wire data table.

[peter_sol]

Hi Michael. Sorry you have problems with this.
Don't forget that those units have 2 x 3 ohm speakers in parallel plus the tweeter so normally have a load of l.5 ohms.