Very annoying Invicta "Duovox" model 62 Amplifier.

[mole42uk]

I'm trying to get this Invicta stereo amplifier going properly for Roger. I've had it far too long because every time I get it on the bench I can't find a fault, but it doesn't sound right at all. It's a stereo single-ended EL84 amplifier with half an ECC83 as a splitter and another half ECC83 driver.

Today I've found that, if the volume control is advanced more than half-way, one channel oscillates at high audio frequency. Turn the volume down again, it stops. Feedback loop, thinks I, but there's nothing I can find wrong there.

All the DC supply voltages are within reason, all the resistors around the output and the splitter stages are within tolerance. That capacitor has been changed. The only voltage oddity that I can find is the EL84 cathode voltage is about 50% higher than the data sheet, but the cathode resistors fitted are 270Ω rather than the 220Ω on the data sheet.

Sometime in the past it has eaten an output transformer (on the channel that squeaks), which has been replaced with a RadioSpares universal type, but the voltages around there seem pretty much the same as the other channel so I don't immediately suspect that.

Any ideas gratefully received!

[Edward Huggins]

Could be self-oscillation via the feedback loop? Wrong tapping on OPTX? Quite a simple 2 stage circuit. You say it's a single-ended circuit, but you then mention a (Phase?) "Splitter" which is confusing.

[ukcol]

Hi Richard,

Perhaps the characteristics of the R.S. transformer are sufficiently different from the original to allow the amplifier to go into RF instability.

This problem is usually dealt with by using grid and/or anode stopper resistors in the output stage. Finding the right values for the resistors and deciding whether to use grid or anode stoppers or both is a bit of a fingers crossed art rather than a science, see discussion here.

It may be worth adding stoppers to your amp or increasing their values if some are already fitted.

I am assuming you have ruled out the feedback from the transformer being the wrong phase.

[peter_sol]

Output stages are built around the output transformer so it is not a surprise that the side which had been replaced is screeching
Try a cap across the secondary perhaps a 0.001uf? or if there is already a cap increase it.

[mole42uk]

Originally Posted by Edward Huggins

You say it's a single-ended circuit, but you then mention a (Phase?) "Splitter" which is confusing.

Sorry! It hasn't been a good day :-(

Of course it doesn't have a splitter...... it uses the two halves of an ECC83 as a pre-amp before the Baxandall tone circuit and as a driver after it.

[mole42uk]

Thanks guys, it's good to work with you!

Colin asked if I was sure the 'new' output was wired in the correct phase, so I connected both outputs to the two channel 'scope and fed the same signal to both inputs.

The phase is correct, so I know the feedback loop is in the correct phase, but the squeaky channel clips top and bottom much, much earlier than the other. Since the voltages match channel to channel, and swapping the valves doesn't change the results, I'm beginning to think the output transformer might be the suspect. I will have to look for the RadioSpares data sheet and see if I can use another tap to improve the results.

[mole42uk]

This evening I changd the primary tapping to something more appropriate, according to calculations for a single EL84 at 36mA anode current. The squeal is gone, but that channel is much quieter than the other and clips earlier. I'm now suspecting the secondary may be using the wrong tapping (an 8Ω speaker connected to a 2.5Ω tapping) as well so I'm planning a series of trial-and-error tests to see if I can get a near-enough combination.

[stevehertz]

Hmm. Who is Roger?

[Ed_Dinning]

Hi Richard, is the new op trans designed for single ended or push-pull operation?
Some of these were basically PP types with lams interleaved (no air gap) and if used on single ended (where the lams are butted with a thin paper air gap) can saturate and give all sorts of problems.

Ed

[John10b]

RS have added a "new" output transformer to their range, designed specifically for EL84, it's made by OEM, trouble is it is twice the price of the other one i.e. £15 and £28 respectively.
I have had problems with unstable amps which was cured by replacing the o/p transformer, without modifying the circuit.
John

[DangerMan]

Multi-tapped transformers can have some nasty leakage inductances and capacitances (and therefore resonances) when compared to bespoke ones of the correct ratio. It is quite possible that these are causing instability when the feedback loop is closed around the amplifier, especially if there is heavy negative feedback applied by the design.
Using different taps may reduce the loop gain and other characteristics, hence the improvement in stability when tried.

A better transformer may indeed be the best solution in the end though.

[ronbryan]

As part of your investigation, have you checked any HT decoupling capacitors? If they are past their best, unwanted feedback to low level stages via the HT line can occur at higher volume settings.

Ron

[RojDW48]

Originally Posted by stevehertz

Hmm. Who is Roger?

It's only me!

[mole42uk]

Originally Posted by Ed_Dinning

Hi Richard, is the new op trans designed for single ended or push-pull operation?

Hi Ed,

I gave a clue - it's a "RadioSpares" branded transformer, so dates back to sometime in the early '70s I'd think!! It has so many taps that it can be used in single-ended or push-pull as required. In those days ISTR that RadioSpares components were designed to a specification rather than to a price.

Although labelled as the "Standard", the labelling and tappings seem to match the "Universal" for which I have a data sheet. It had been wired inappropriately, in that the primary was presenting 12kΩ to the output valve, when something more like 3kΩ would be more acceptable.

It's still not right, but a lot better than it was so I'm hopeful that another session with signal generator and oscilloscope, and a carefully selection of tappings might bring us to a closer approximation of the original output.

[peter_sol]

5K is optimum for EL84

[Ed_Dinning]

Hi Richard, if the inductance on this transformer are different from the original then you may get some improvement by carefully adjusting the feedback capacitor.
(assuming it is the standard R, C parallel circuit). Feed a 1KHz sq wave in and adjust the cap for best output response. This is normally a cap of a few 100 pF only.

Ed

[ronbryan]

I still don't understand why the negative feedback loop from the OT is under suspicion when the instability is provoked by increasing the volume control setting (see post #1). Surely any negative feedback loop on this amplifier would not include the stage with the volume control in it, as the amount of NFB would then depend on the control setting.

Ron

[Edward Huggins]

I think on this model the Tone control may be in the NFB loop. As this is likely to be a Pye design, what's the model number?

[mole42uk]

Ron is correct, the feedback loop includes only the driver and output stages. It's a very simple feedback - one 15k resistor from the speaker connection back to the ECC83 cathode.

Edward, the tone controls are a typical "James" network between the pre-amplifier and the volume control.

It's an Invica "Duovox" model 62.

I may get time to look at it again today, but my work computer died last Friday and it's a bit of a priority!

[Edward Huggins]

Thanks for advising the Invicta model number, it's not the unit I thought it was. Now I can see it, don't compromise on that OPTX.