Requesting feedback: attenuator + re-amplifier schematic

[cdemike]

I've been kicking around design ideas for an attenuator for a while now and I'm reaching a point where I think I'll probably go one of two ways: build a JohnH attenuator or build a load box + re-amplifier. I like the idea of the re-amplifier because it allows for continuous volume control rather than in stepped intervals, but I'm feeling out of my depth designing the re-amplifier. It seems to me that it should be as high fidelity as possible since the goal would be to produce the signal coming out of the main amp's OT as accurately as possible, but I'm not sure how successful designs like the Fryette Power Station specifically approach this. I do know that the Power Station uses a phase inverter (not sure what topology) into a push-pull power section based around 4x 6L6.

In that light, the ECC99 power section might seem strange, but I don't need a ton of power since this is mostly for home use; I could just use the Weber attenuator I have now to play with a drummer (the Weber difficult to use for stable apartment-friendly volumes, which is the reason I'm starting this project). In addition to the low demand for output power, I also have an extra ECC99 on hand along with a Hammond 125C and a handful of 12BZ7's, so that helps keep costs down. I chose a 12BZ7 since haven't really found a ton of use for the few I bought, but I do have a few extra 12AX7's if the 12BZ7 is a poor fit for this application.

The load box section is based on a JohnH M4, which uses the LC network to simulate a 4x12's resonant frequency when the attenuator is used as a load box. So I'm reasonably confident the load box side makes sense since it's a well-tested design.

M4-230722.gif

My thinking designing the re-amplifier was to maximize headroom and bandwidth by using plenty of NFB (both global and local via non-bypassed cathodes). I considered also working in local NFB at the driver's grid, but I wasn't sure if that would be unnecessary. I also went with a cathodyne given its linearity over a LTPI. Given how great the input signal magnitude could be, though, I wasn't sure if a driver was even necessary. If not, I was also considering a MOSFET-based split load phase inverter, but it seems like global NFB might help keep the ECC99's operation more linear.

In addition to NFB and linear amplifier design, I was also trying to figure out the best way how to approach grounding. It seems to me that it might be a good idea to keep the load box's ground floating on its side since there's a ton of current flowing through the load box, but most of the attenuator designs I've seen use a chassis ground. Also, between the noisy rectifier and heater grounds and the potential for a ground loop with a separate power supply linked to the main amp's power supply via the chassis ground, I wasn't sure how best to stave off a noisy grounding scheme if keeping the load box's ground floating is a bad idea. Would an isolation transformer be the best solution? In that case, maybe using a transformer phase splitter and working local NFB into the push-pull pair would be the best approach?

This is the schematic as it stands now:

Reamper 50w input.png

[mhuss]

The symbol drawings in that schematic look awfully familiar.

The attenuator seems very complicated, is there no easier way to accomplish what you want?

Is the re-amp a completely separate product/build? Why not combine the two?

[cdemike]

The symbol drawings in that schematic look awfully familiar.

The attenuator seems very complicated, is there no easier way to accomplish what you want?

Is the re-amp a completely separate product/build? Why not combine the two?

They're definitely a lot nicer looking than the very unappealing hand-drawn schematics I make, so kudos to the original artist
In seriousness, my apologies for not asking in advance of using those symbols!

My main goal is more precise control of volumes at quieter settings, since there's about a 1-2 degree area on my Weber's control sweep that's loud enough for me to hear how my amp is reacting to pick attack over my string noise while also not getting me evicted from my apartment. The Weber also has a tendency to spontaneously move itself out of that very narrow sweet spot which causes noise and weird artifacts in the sound. It's only really an issue at these very low settings, but that really all I use the attenuator for anyway.

But I totally agree, and I should have said in my first post that this seems like this is a very complicated and expensive solution to something that might be accomplished by other means. Part of my hang-up that's leading me to try to get the attenuator acting as closely to a real speaker as possible is that I don't get the option to turn my amps up very often anymore, and since I do most of my modifications and builds at home. I'm somewhat concerned that I've been voicing my amps using a misleading representation of what they'd sound like if unattenuated so as to keep up with a drummer and so my thinking was that if I'm going to replace my attenuator anyway, I'd want to do my vey best to get my amp sounding as close as possible to what it would sound like without an attenuator to solve that issue.

In more practical terms, I think that means I need to figure out whether a plain M2 JohnH attenuator (schematic attached) would sound close enough to the un-attenuated amp versus going for something as accurately as possible. Almost all of the research I've done points to the Fryette Power Station as the highest fidelity attenuator which seems to be attributable to how well the reactive load works and the fact that it's a re-amping device, so both the OT and speaker behave similarly as they would without the attenuator. The iteration of the JohnH attenuator with the LC network appears somewhat similar to the Aiken loadbox design, and both the write-ups and the data that's been posted seem very favorable in terms of how they work as loads, so the OT side seems happy. I'm not really aware of a way to keep the speaker side also happy aside from impedance matching, which the simpler M2 attenuator also does, but it seems to me that there's not really a way around there being a significant increase in resistance between the OT secondary winding and the speaker with any kind of attenuation.

So reamplifying the signal seems to be the closest to a no-compromises solution if I'm understanding that right, but I do want to emphasize the "if" in that statement.

John H attenuator M2 (Nov 2022).gif

[mhuss]

Re-amplifying to isolate the actual speaker from the attenuating network makes calculations simpler and avoids having your network depend on load/speaker impedance. You just have to make sure the re-amp stays in its linear range, which doesn't seem to be an issue in your use case. Also, ideally the re-amp should be transparent and not contribute to the sound itself, which means a solid state/chip amp would be more ideal(and some re-amping attenuators do this).

Also, IME you really have to build something like this and try it; there's no theory or calculation that can guarantee it will sound/perform as you want.