Sessionette - highly sensitive to EMI

[Mick Bailey]

This is a late model (but early production) reissue Sessionette guitar amp that's come to me with noise issues. It's really sensitive to any interference from mobile phones and in particular my DECT cordless phone, but the owner says it's only recently been doing this. Otherwise the amp is working fine.

I've isolated the problem down to one area; The output of IC6a is coupled via C44 and if I ground the negative side of the cap all is fine - complete noise immunity. If I ground the positive side of the cap the noise is back in full. I changed the cap. What the schematic doesn't show is that the signal path from that cap is via lengthy PCB tracks and then unshielded wiring to the rear PCB 'send' and then back the to the 'return' opamp buffer.

What I can't comprehend is how this worked OK previously. I can't see any failure. IC6 is fine, no DC offset. No unusual voltages and the amp is quiet apart from the interference. As a test I removed C44, detached the leads to the rear PCB and wired IC6a pin 1 directly to C15 with screened cable. A slight improvement, but not much.

This is driving me nuts. Does anyone have any ideas? Maybe I'm missing something really obvious. Just like a pantomime character where the audience is shouting 'He's behind you!"

[Helmholtz]

I would wonder how/where the high frequency EMI gets inside. Is the amp well shielded? If yes, the EMI might still enter via the mains lead and spread/travel along the power supply wiring.
If e.g. the HF carrier rides on the +/- 16V supply, it might get demodulated/detected by D2/D4.

[pdf64]

I don’t trust an opamp (potentiality) feeding the outside world without a series resistor, even just 100 ohms. That at least helps to maintain stability if there was a much of a capacitive load on the send jack, eg a cable
I’d consider replacing IC6, the output could have shorted, overheated and been left walking wounded.

[Mick Bailey]

The opamp is removed right now for testing purposes so I'll install a socket and replace it. There's also a convenient 0R jumper that can be replaced with a series resistor. The amp is well screened with grounded foil to the top and sides. There are a few cables exiting - speaker leads, reverb send/return and a three-core plus screen for channel switching. I've eliminate the reverb and channel switching cabling but the amp is certainly more sensitive where the leads exit. I did even wonder if the speaker leads were radiating noise.

There are a few unrelated deficiencies with this amp; the elevated ground comprises a 100v/1u cap in parallel with a 1/4W carbon comp resistor. No anti-parallel diodes and therefore no current capability. There's also no Zobel network, though the schematic shows this as a 100n cap in series with a 4R7 3W resistor (this is fitted to earlier amps). Later on 100n film caps were added to the main PSU filter caps and these are absent in this version.

[Helmholtz]

Did you try a folding ferrite on the mains or speaker cables?

[Helmholtz]

The amp is well screened with grounded foil to the top and sides.

If it's aluminum you surely know that it self-passivates (oxidizes) and not necessary makes good contacts.

[g1]

Since you mentioned 'something obvious', any chance the Aux. in jack is not grounding?

edit: never mind, C44 test should have eliminated that.

[Enzo]

Idle thinking. Look other side of C15 to R7. D2,4 clamps protect IC1. Any chance one of the diodes is open or leaky and allowing either some DC to IC1 or allowing a diode to "detect" - in the radio sense - the local EMI. ALternatively, you grounded neg end C14 and it kills noise, presumably grounding neg end of C15 would as well? But then does grounding the POS end of C15 do anything?

[Mick Bailey]

If it's aluminum you surely know that it self-passivates (oxidizes) and not necessary makes good contacts.

It's all been replaced and meters through. I was thinking of adding ferrites, mains filters etc. But the owner says it only recently started to do this, so I'm looking for a component/circuit fault rather than a modification at this stage. The recent development of this problem was confirmed today with another band member. They're pretty reliable, pro musicians I do a lot of work for so I think it really is the case that it used to be OK. It's getting late but with a fresh look tomorrow I'll Replace IC6 and take a look at the checks Enzo suggests. I've tried my DECT phone with every other amp I have here (including my own MK1 Sessionette) and there's next to no noise with any even with the phone held right against the amps.

It's interesting that even with grounding the + end of C44 the signal is so damn loud. Not just some background noise.

[Helmholtz]

I was thinking of adding ferrites

My proposal referring to folding ferrites was primarily meant as a means to identify the main "gateway" for EMI. This might be helpful for troubleshooting.

I completely understand that you are looking for something that changed, failed, deteriorated.

[Enzo]

My proposal referring to folding ferrites was primarily meant as a means to identify the main "gateway" for EMI. This might be helpful for troubleshooting.

That right there is the key to the whole process. We need to do anything we can to isolate the problem. Once we know WHERE the problem is, we can determine the cause.

[Mick Bailey]

I don't have any clip-on ferrites but made up a replacement speaker cable wrapped around a ferrite ring. It took a few experiments and needs 8 turns of hookup wire but this almost kills off the noise. I have the ring right inside the chassis at the entry point and minimal external wiring. I did also discover that the reverb send cable was contributing some noise (though oddly, not the return) and replacing the cables with high-quality cables has fixed that.

Revisiting the board; grounding either side of C15 kills the noise and lifting the diode clamp has no effect. I've been able to more accurately pinpoint the most sensitive part of the circuit and it's from C44 onwards to C15. It looks like the speaker leads radiate noise into the can of C44 and the PCB tracks that run the length of the board from C44 to C15.

[Helmholtz]

You might try to ground the speaker frame.

As the carrier frequency is very high, a few pF wired to ground in the "right" place might suffice to attenuate it. That is before demodulation by some semiconductor.

[Mick Bailey]

I rerouted the speaker wires away from the preamp and this has the same effect in reducing the noise as the toroid. The noise is certainly being radiated internally by the speaker leads and also being picked up via the plastic phono plugs on the reverb tray.

All of the solder joints between C44 and the power amp input have been reflowed just to make sure i don't have a rectifying joint. I experimented with some 47pf caps at various points but no improvement. It;s almost there, but not quite good enough and I still have not identified a clear fault. Certainly grounding anywhere from the negative end of C44 onwards kills the noise, but wiring a separate cap and a screened lead from IC6a directly to IC1b doesn't fix it. I'm wondering if IC1b is the problem?

[Mick Bailey]

I replaced IC1 and no change. I got the amp to where I thought it was good enough and at gigging volume there was just a little bit of noise with a phone held right against the amp, but nothing noticeable when playing and with my phone in my pocket. So, I had the amp set up and the owner called by to collect and my heart sank - he didn't even get through the door and the amp was picking up his phone super-loud from 30 feet away.