Amp with Anode mixer - Signal gets louder when removing one side?

[cboysen]

Hi! I have built new amp, a 12 watt tweed twin style amp with 6v6 output tubes, and reverb.

It uses a common anode mixer to mix the reverb with the dry signal. The reverb recovery uses two triodes in series, the latter being the mixer. The two anode mixer triodes shares a 56k, but when pulling the reverb recovery tube, the signal gets louder / clearer / better. This is odd, considering the other part of the anode mixer now sees 56k plate resistance instead of 112k.

Since the reverb recovery and the dry part of the anode mixer shares the same power supply node of 5k / 10uf (3hz cut-off), I fear that a feedback signal travels through the node to the dry part of the amp in opposite phase. I don't have a scope on hand to test this, but it's the only logical explaination besides capacitive coupling.

If I were to separate the reverb recovery filter supply from the dry amp, how would I go about mixing the signal, now that I cannot use a shared plate resistor?

Hope it makes sense.

Kind regards
Christian

[pdf64]

Consider the anode resistance.
The remaining triode is no longer having to drive into the anode resistance of the valve that’s been removed.

[sluckey]

The reverb recovery uses two triodes in series, the latter being the mixer. The two anode mixer triodes shares a 56k, but when pulling the reverb recovery tube, the signal gets louder / clearer / better. This is odd, considering the other part of the anode mixer now sees 56k plate resistance instead of 112k.

That 56K shared plate resistor does not magically change to 112K when V4 is in circuit.

[cboysen]

Thanks for the quick replies.

Alright, I'm just a little confused. It was my understanding that when using a shared plate load, the load resistance should be halved, for the triode to 'see' the same load. When I pull one half of the load, wouldn't the remaining triode now see half the plate resistance? Also, it seems to be a generally good idea to separate the reverb circuits filtering from the rest of the amp, so I thought I might have messed up something here.

Anyway, should I up the plate load to 100k to simulate what's going on?

Also, if you see any obvious improvements to the schematic, all tips are welcome. I'm still learning.

Kind regards
Christian

[pdf64]

With a shared anode mixer, due to their internal anode resistance, each triode loads the other. Like one car having to tow another.
Removing one of the triodes is like taking away the car that was being towed, ie the dead weight has gone.

Your fx loop has a rather high output impedance. With a decent length of cable plugged in, the sound will go dull.
Plus the high signal level will be liable to overdrive stuff in the loop.

Inside a negative feedback loop, to avoid unexpected performance / instability, it is usual to avoid unnecessary phase shifts within the bandwidth of the OT. Hence the values of C19 and 21 seem a curiously low. Probably not so much as to cause a problem, just seems odd.

[cboysen]

With a shared anode mixer, due to their internal anode resistance, each triode loads the other. Like one car having to tow another.
Removing one of the triodes is like taking away the car that was being towed, ie the dead weight has gone.

Your fx loop has a rather high output impedance. With a decent length of cable plugged in, the sound will go dull.
Plus the high signal level will be liable to overdrive stuff in the loop.

Inside a negative feedback loop, to avoid unexpected performance / instability, it is usual to avoid unnecessary phase shifts within the bandwidth of the OT. Hence the values of C19 and 21 seem a curiously low. Probably not so much as to cause a problem, just seems odd.

It makes a lot of sense when you put it that way. Is there a way to mix the signal without this interference. I chose to use common anode mixing, as apposed to passive resistors, especially to avoid interactions. Maybe the standard fender 3.3M/10pf mixing of reverb is a better way to do it?

C19 and 21 are chosen to avoid excessive bass, and tested on another amp I built a few years ago, it worked out alright. I may decrease the NFB in the future.

The FX loop is solely for the intend of using an external preamp, such as my kemper directly into the FX return, so it should be ok. 

Thanks for the tips.

[pdf64]

…Is there a way to mix the signal without this interference. I chose to use common anode mixing, as apposed to passive resistors, especially to avoid interactions. Maybe the standard fender 3.3M/10pf mixing of reverb is a better way to do it?…

Shared anode mixing is great. Minimises interactions yes, and should be lower noise than a simpler resistive passive mixer.
But it’s still a passive mixer, and they all have an inherent signal loss characteristic.
If you’ve got a spare triode, it’s a toss up between that and an inverting NFB mixer.
See
http://aikenamps.com/index.php/designing-single-stage-inverting-feedback-amplifiers

Every method has its set of benefits and drawbacks 

[PRR]

> my understanding that when using a shared plate load, the load resistance should be halved, for the triode to 'see' the same load.

If both grids get the same signal, yes, maybe.

But with different signals, no. The two plates don't help each other, they fight.

Plate mixing is usually not the best use of the tubes. With triodes, anything more than microphone level at the grids is gonna distort some. It may be difficult for a triode mixer to drive 6V6.

OTOH, guitar amp is not about squeaky clean. But if it gives you "less", you probably want to re-think. Mixing at a grid at least one stage of gain above a pickup (axe or tank) is usually unobjectionable.