Tube-driven series FX loop changes tone?

[jamesmafyew]

I've got a Sunn Model T Reissue (Fender-made), and it has a very standard Fender tube FX loop in it. Everything works as it should in terms of gain, level control, whatever. It's relay-switched from either the footswitch or a switch on the back of the amp. It has separate send/return controls for each channel (helpful!).

Except: it seems like when the loop is switched in, the tone changes pretty significantly. It has a nice broad midrange with the loop out of circuit, then a very "scooped" sound with the loop in. More bass and more treble --- makes me want to stop playing and twiddle knobs. This is true regardless of what effects I use, whether they're bypassed, or even with just a patch cable between send and return jacks. It's also unaffected by what brand/type of tube I put in.

The loop works so well in terms of functionality, I just want to be able to switch effects in and out without the weirdness of a sudden dip in the midrange. Any ideas for how to deal with this?

Am I just being too picky -- is this normal for a series loop? All my other amps have solid-state parallel loops, so my expectations could just be unrealistic.

Preamp schematic attached. The FX loop is down at the bottom, V5A and V5B.

[J M Fahey]

Hi jamesmafyew.
I have been analysing your schematic and got as intrigued as you, because I see no *electronic* cause for that (obviously I don't doubt your ears).
To begin with, not only V5a and V5b take part in this, but V4b too, and in an important way: it splits the signal into direct and loop sends in a very different way.
Direct signal is taken from its cathode, and sent *to another board* to pin 11, K4b through a capacitor and a 22K resistor.
I see a FET mute and two back to back Zeners that will clip any signal over 15V peak , but that would happen on both direct or effect signals, there's no difference there.
Effect signal is taken from V4b's plate, in theory the same level but phase inverted, and sent *to another board* to drive pots, then selector relay K3a, a cathode follower, another relay, and finally out. (at last). In all that path I see nothing that should change frequency response.
Same happens with return path, up to and including V5b, which makes up for the signal presumably lost along that long, tortuous path.
What am I pointing my finger at?
Given that I see not *electronic* components changing frequency response, I guess that the problem is a *Layout* one.
Complicated PCBs, crisscrossing tracks, connectors (never "zero" ohms), ribbon cables (I haven't seen a Model T Reissue innards but I'm sure it has a few of them), etc. , introduce small unintended capacitors at random coupling different parts of the circuit, that should better be kept apart.
If those cross-couplings are so bad as to cause oscillation, they will be corrected, but lower level ones, which might affect sound somewhat, may go unnoticed.
Besides, that "Reissue" is quite more complicated than the original one.
In a nutshell?: I guess you'll have to live with it.
Useless ranting follows: "why, oh God !, do they "enhance" reissues?"
(Besides the security mods: grounded power wiring, no death cap, etc.)

[jamesmafyew]

Yeah that's what I was thinking --- I couldn't see anything in the schematic that would account for it. It's the same design as used in several other recent-production Fender amps. I think the new "Twin Amp" is about 75% the same as this "Sunn" reissue. Of course it's not really a reissue it's a completely different design, and I know that... but I do like the amp quite a lot for what it is.

There is one ribbon cable running from the main board to the FX loop board, it's 8-pin. There's no audio on it, it's just for the switching circuitry, since the FX loop board also has the footswitch jack and the FX loop relays. There are 6 wires running to and from the V5 tube socket, a red and black pair coming from the PS, and the signal wires are shielded. It doesn't look like there's much room to rearrange anything.

Edit: what I don't get is, the signal still goes to and from the FX loop board even when the loop is off. The line from the cathode of V4B still goes to the FX board through a shielded wire, zip-tied to the wire (from V4B's plate) that goes to the loop itself. The bypass line travels the length of the PCB to the Main Amp Out jack, then back to the mainboard from the same relay as the loop before going to the phase inverter. So there's little appreciable difference in wire routing between FX-on vs. FX-off, except for the wires to and from the V5 socket... which aren't very long.

I suppose it could be a design problem with the PCB layout? Or should I blame those six wires from V5?

[pdf64]

Have you tried some different tubes in the relevent slots, ie V4 and V5?
I couldn't get the schematic to open, BTW.

[jamesmafyew]

Yes, I've put all different kinds of tubes in those spots... Chinese, JJs, Sovteks, Svets, and Tung-Sols. I can't hear a difference between any of those tubes in those spots. Whatever that says about me... I definitely can hear a difference between loop on and loop off.

The schematic doesn't seem to be helpful anyway since it's a sound design to begin with.

I'm wondering now if it has anything to do with the fact that I never have a reason to turn the amp up past 2 on either volume knob (and rarely that high... the amp is LOUD), and if maybe V4B/V5A aren't in their linear range or something (??) with the tiny amount of signal they're probably getting. I don't use this amp at volume with my band except as a slave amp to my Marshall, where it does exceptionally well (I play doom metal). I use the FX return for that and it sounds just fine; level-matched to the Marshall with the return pot, the two amps sound very similar.

I use a PODxt in the loop when I'm using it on its own, so the sends are at 10 (roughly at +4dbu) and returns at 1; I set the balance with the POD's output knob (it's about 2 o'clock). I think I can use the POD's parametric EQ to try and balance the sound. It probably just needs some 8kHz+ and some 100Hz shaved off.