Harmonic trem in Princeton reverb

[rootz]

The current involved in the LFO are very low. I do not expect much problems, but better safe than sorry I guess. Just need to order some more parts and an 290bex and I'm good to go.

And yes, a source follower would work very well in this setup. I personally don't see anything wrong with a solid state LFO, follower and phase splitter. I've seen that concept somewhere around on ampgarage. I would leave the mixer tube intact as it is in the signal chain.

[rootz]

And... We've got a working prototype and I think this is a massively cool one. I mean, we all know the standard terms, but this really is something different.

But it is a prototype so it has a plethora of issues. More humm than I'd like, some big thumping when the master is high and trem on. Probably some grounding issues. Popping trem on/off switch.

Anyway, let me first share a short vid of the amp in action and tackle the issues later.

[Colossal]

Great sounding amp!

[martin manning]

Nice! Great sounding trem and reverb!

[rootz]

Thanks for the compliments! Youtube and iPhone compression don't do the amps sound any good, but for now it's the best I can do.

Now on to the issues - and I hope to get some help here - first my grounding scheme. Attached is the current layout, apart from some details. The grounding scheme is a bit of a mash up, but the JJ 40/20/20/20 leaves me with limited options. So I worked with a ground buss. All pots go there. The V1 preamp cathodes with reverb, the trem and the bias all go to the main ground at the can cap. PI goes to the loop jacks. Reverb jacks non isolated, but that never gave me hum issues. Input jack grounded at the input. Output tubes grounded near the sockets. I'd like a grounding scheme like in an ODS, but yeah, can cap.
I think the problem is the ground buss connects the preamp and reverb controls also to the trem controls. The currents involved are tiny though...
So what do you think?

[Guy77]

Thanks for the compliments! Youtube and iPhone compression don't do the amps sound any good, but for now it's the best I can do.

Now on to the issues - and I hope to get some help here - first my grounding scheme. Attached is the current layout, apart from some details. The grounding scheme is a bit of a mash up, but the JJ 40/20/20/20 leaves me with limited options. So I worked with a ground buss. All pots go there. The V1 preamp cathodes with reverb, the trem and the bias all go to the main ground at the can cap. PI goes to the loop jacks. Reverb jacks non isolated, but that never gave me hum issues. Input jack grounded at the input. Output tubes grounded near the sockets. I'd like a grounding scheme like in an ODS, but yeah, can cap.
I think the problem is the ground buss connects the preamp and reverb controls also to the trem controls. The currents involved are tiny though...
So what do you think?

That trem sounds absolutely fantastic Bas! Congrats on another great amp.
Whenever I have a bit of unwanted hum in a new build ( a model I dont build on a regular basis) I like to elevate the AC heater voltages with some DC voltage. I find this makes the amp quieter. I am sure you already know about this but here is the link just in case ...

https://ampgarage.com/forum/viewtopic.p ... 5&start=30

Cheers and Happy New Year!

[rootz]

Happy new year to you too Guy, and of course all others here.

I'm afraid elevated heaters are not going to cut it as it doesn't appear to be mains frequency hum, but 100Hz. But the most obtrusive noise is the thumping of the trem, especially at higher master volume settings. Will try to splitting the ground bus tomorrow and see what that gets me.

[rootz]

So, some progress here. I got rid of the clicking in the trem circuit. It turned out to not be a layout or grounding problem. The problem is really in the basic idea of the amp itself IMHO. The mixing stage mixes both in phase and the out of phase LFO signals and should more or less cancel them. But it does not do a very good job at it, either in sim or real circuit. With a scope some 9V p-p residual LFO signal was visible after the mixers. Mind you, the mixer tube was pretty well balanced on my tube tester. The residual LFO signal had double the frequency of the LFO itself,
so that means the problem was in the mixing stage itself.

Now I had a 22n cap after the mixing resistors, leading to a 3.3meg resistor and 470k resistor//1meg master pot to ground in the dry path of the reverb. The -3dB point was at 2Hz that way. No wonder all residual LFO signal got amplified!

So to get rid of the LFO signal, I chose a -3dB point of 40Hz. A 1nF cap would be just right for that point. And voila, no more ticking of the trem.

Then there was that hum. That was a grounding problem. It was cured by grounding the bus bar at the input and the rest of the board at the can cap. There was a bit of an oopsie with the input jack. I totally forgot the 1 meg grid leak of V1a.

I also changed the plate resistors on V1b. IT was a 100k/10k split load. Now it is 100k/22k. This gives a bit more gain, which suits the Princeton reverb like reverb and power amp better.

[sluckey]

Not saying you should do this, but Vox used a multi-pole filter to get rid of residual LFO signal. Look at the top right portion of this schematic...

[rootz]

Thanks Sluckey. It was actually the Vox solution that have me the Eureka moment! Vox made a very drastic high pass filter that way. I simulated that and could see the amp would lose most of its low end with the filter. Is that observable when playing the amp?

I guess I could use only part of the Vox solution or with some adjusted cap values, to preserve more low end. If I had any space left on the eyelet board that is…

[martin manning]

You might consider a balance trimmer in the phase splitter Rk to minimize the LFO residual.

An alternative to cascaded RC's like Vox used would be a Twin-T notch filter. For example, using 470k resistors and 47n caps (three of each) will knock 12 dB off over a narrow range centered about 7 Hz, and be 3 dB down at ~28 Hz, same as a 5xRC shifted down to get the same attenuation at 7 Hz.

[rootz]

Please, elaborate on the balancing of the PI helping. How would that help if the LFO signal is bleeding in the circuit before the PI? Doesn't it always get amplified by the PI then, either balanced properly or not?

That T notch filter is a neat idea! Hadn't thought of that. Speaking of which, that means a band stop filter would also work? Lemme break out LTspice and cobble your idea and a band stop together. Surely better than a single small cap, although the amp still has more than enough low end.

[rootz]

Prototyping got a bit out of hand. Luckily, it is close to my favourite part of amp building. The gut shot is of the Princeton with harmonic trem, before various mods and shielded wiring were done. Looks like some more mods to come.

[martin manning]

Please, elaborate on the balancing of the PI helping. How would that help if the LFO signal is bleeding in the circuit before the PI? Doesn't it always get amplified by the PI then, either balanced properly or not?

I’m referring to the phase splitter in the harmonic trem.

[rootz]

Yes, off course, that makes more sense

I played with balancing the mixer stages and it didn't seem to get me the results you'd expect. The inverter triode has quite some local feedback around it, so my thinking was: why not experiment with that. So I altered the value of one of those 1 meg resistors from 680k to the stock 1 meg. I tested this first in sim, so I knew that by doing so that the LFO signal on the lower half of the mixer would increase, thus bringing it in line with the top half. I noticed no improvement to speak of, so I wasn't too inclined to break out the scope and measure it there was an improvement to speak of. But a pot in line with the 1meg resistor I altered would do the job of balancing just fine.

So I checked the principle of balancing again in LTspice. This is with mathematically perfectly balanced triodes all through the circuit. Even if I balance the LFO and inverter to be within millivolts of each other, I get residual LFO signal. But it almost looks like a saw tooth. So I was thinking the signals might not be a perfect match, so can never cancel each other out completely. So I inverted the LFO signal going to the lower mixer half and could see them being not perfectly 180 degrees out of phase and the wave form was slightly different. No wonder balancing did not help all that much. And again, this was with perfectly matched triodes.

Attached are screen shots where you can see the signal (one inverted to make it easy to compare!) and the other the residue when I add the LFO signal on the top half of the mixer to that on the lower half. There is my sawtooth!

So at this point I can either improve the LFO signal symmetry, so it will cancel out better, or filter with/after the mixer stage. The latter seems more logical to me as it is pretty much independent from tube condition. Also, the dropping resistor before the LFO/inverter node seems to have a big impact on residue. Cross talk via the power supply I guess?