1176 Motorboating/Oscillating

[radardoug]

You don't need RL1 on your board at all. Just link all the contacts together. Have you got another layer on that board? No connection between the quench diodes and the relays. It would pay to put ground plane over the tracks to the output transformer. And if they run close to the input circuitry, a ground track on the same side as they are between them and the input circuitry. Or if it is a long run, do it in shielded cable.

 

[sircletus]

Does this mean the board itself is possibly not grounded to the chassis?

All grounds are on a copper pour, which I hid for purposes of the screen shot that both layers could be seen.  That's what you're not seeing.

I generally try to keep my signal traces on one side (in this case, the bottom) with power traces on the other side.  Faux ground plane is same side as power (so, ground plane would be red here) on this layout.  There are some exceptions here and there (several in this little screen shot) for simplicity's sake.

I understand with the relays.  Good call.  For whatever reason, visualization of switching has always been my downfall.

 

[sircletus]

For those interested and/or encountering something similar in other designs, the problem is solved, while retaining relay true bypass.  For starters, I moved the output XLR connector about six inches away from the input on the rear panel.  I also placed the second bypass relay (DPDT for the +/- output) on a small daughterboard (pad-per-hole vector board) directly at the XLR output, connected to PCB via Mogami twisted pair.  All is well.

Radardoug, if you're reading this, hear me out and please correct me if my logic is flawed!  Here's why I decided to stick with two DPDTs for true bypass:  I was thinking that if I used one DPDT switching at the output only, the output of the previous device in the signal chain will be seeing the 600Ω load of the attenuator in parallel with whatever device comes after the compressor.

Same with employing a DPDT only at the input: I would imagine any preceding device's output would see the secondary of the output transformer as a load in parallel with the input of the next device in the chain.

Either way, I don't necessarily want the device to present any kind of load when in "true bypass."

 

[SSLtech]

REAL 1176's never even had bypass.

You could always bypass them at the console in may instances, but nobody bothers to remove the load of the 1176 input... they just switch the 'listen' from the input to the output. The insert send still fed the device.

I'd say this is overcomplicated... I've never had a problem just switching the 'sniff'. Yes, one could imagine that there would be a difference, but if this really were a problem, I'd have expected more manufacturers and users to have noticed. -But of course this is yours do do as you wish.