Tremolux vs. Vibrolux tremolo?

[Wombaticus]

2.2uF Ck

Re: the wiring/layout - can you post any gutshots?

Here's the gutshot in it's current hacked-up but sounding-good state:

http://www.wombatamps.com/Wombat_5G9.JPG

I'm not too proud of how this looks, but believe me, it looked a *lot* prettier before I started futzing around inside trying different things.

It's kinda hard to make out what's what here, so here is an sketch of at least the board layout used:

http://www.wombatamps.com/5G9_final.png

In the main pic, the grid wires in question are the thicker white ones going to the power tube sockets. Note that the shield ground runs for these (thinner white wires) are still attached, but with the ground end floating out in space towards the camera and unconnected. Reconnecting these to chassis ground is what screws up the bass.

Note that with the amp shown, it really sounds quite good, I'm liking it a lot. I just can't understand why grounding the grid shields has such a horrible affect on the bass.

Oh, and Tubeswell - thanks for your time, it's much appreciated!

[Groover]

That doesn't look so bad compared to some I've seen pics of!

[tubeswell]

The bass freq alteration could be the particular cable you're using maybe. Where did you have the signal shield grounded? (How are your grounds organised?)

[Wombaticus]

The bass freq alteration could be the particular cable you're using maybe. Where did you have the signal shield grounded? (How are your grounds organised?)

Maybe? I dunno. But it's good quality milspec shielded cable. The shields were grounded to a lug on the chassis. The overall amp ground scheme features a single chassis attachment point for the circuit ground near the input, and a bussed-stars arrangement that follows Merlin's guidelines pretty closely in terms of where things return. PT CT is grounded to the reservoir cap. I'm pretty careful about my ground schemes, so I don't think that's the issue... but in any case, the the shield grounds are attached at one end only, are therefore really just part of the chassis, and it *shouldn't* matter where they are attached.... Whatever the case, the problem depends on the shields actually being grounded, as when I detach the shield grounds it sounds fine.

Latest news is that the trem instability problem is not yet *quite* licked.... now it's stable and good sounding until the amp gets very warm... it'll be rock solid for 30-45 minutes or so from cold, but once things are cooking it starts to lose stability and will sometimes helicopter. I've got a couple of theories I can test when I have a chance, but it's proving to be a surprisingly persistent nuisance.... Next time I do a trem, it will for sure be going on it's own dedicated PSU node!

[tubeswell]

The reason I asked about the shielded cable is that long shielding can act like a capacitor, but maybe poor quality shielding can also have this effect.

Re: the grounding, in my 5G9 but I grounded the trem circuit at the trem footswitch socket ground. I have the pre-amp end grounds (incl the pre-amp filter cap ground) grounded at the input socket ground, and the power amp grounds and other filter caps grounded at one of the PT bolts. My amps is silent and has no problems. Maybe you need to revisit the grounding altogether. I see you appear to have some kind of ground bus connection on one side of the board. I use separate wires for all my ground returns, so that I minimse any possible ground-return interaction that could be otherwise caused by micro-rises in ground potential from different parts of the circuit wreaking havoc by inducing 'back-wobble' through the filter caps.

[Wombaticus]

The reason I asked about the shielded cable is that long shielding can act like a capacitor, but maybe poor quality shielding can also have this effect.

Re: the grounding, in my 5G9 but I grounded the trem circuit at the trem footswitch socket ground. I have the pre-amp end grounds (incl the pre-amp filter cap ground) grounded at the input socket ground, and the power amp grounds and other filter caps grounded at one of the PT bolts. My amps is silent and has no problems. Maybe you need to revisit the grounding altogether. I see you appear to have some kind of ground bus connection on one side of the board. I use separate wires for all my ground returns, so that I minimse any possible ground-return interaction that could be otherwise caused by micro-rises in ground potential from different parts of the circuit wreaking havoc by inducing 'back-wobble' through the filter caps.

I don't mean to reopen the grounding wars, but the ground scheme I use isn't really a bus, even though it looks like one. Rather, it's a daisy-chained star-grounding scheme, as everything is returned to the negative side of the cap that was feeding it, and it follows the "creek, stream, river" paradigm that should preclude the "back-wobble" you speak of. I have used this basic scheme in everything from 100W 1959SLP-type builds to little 5E3s with great success. Nevertheless, I'm willing to consider that the LFO might have it's own special grounding issues, and will certainly try out some variations if my other ideas wash out....

In particular, I've developed an alternative theory related to what is going on -- on the AC side. I admit I was lazy following the old 5G9 schematic and didn't really properly crunch some numbers related to the PSU filters myself. But when I went back and did so, the numbers are revealing.

Let's take a look, for instance, at the low-frequency filtering from the choke filter. The reactance of the choke is frequency dependent, so it is worthwhile to calculate the overall impedance of the choke at frequencies associated with the LFO.... let's say the lowest/slowest frequency generated by the LFO is maybe 5 Hz. My choke is a 5H, 105R DCR choke (Hammond 158Q). Let's plug that into a handy dandy inductor impedance calculator: Inductor Impedance Calculator

With 5 Hz as the frequency, the impedance of this choke is only going to be 189 Ohms! Therefore, using a passive filter calculator, the 189 Ohm/47uF RC filter that is formed gives a -3dB frequency of only 18Hz - well above our actual 5Hz! In other words, this thing is most assuredly failing to decouple the lowest frequencies from the LFO. So I think I picked my choke poorly. I needed one with a higher induction rating, and in this case, a higher DC resistance would actually be beneficial....at the low frequencies associated with an LFO, the overall impedance of the choke starts to be driven largely by the DC resistance! Still, I think that these things must only be marginally stable in the best of cases, but evidently it is enough.... For instance, if I had chosen the Hammond 158M with 10H and 262R DCR, I get 5Hz inductor impedance of 409 Ohms, and an RC filter with an f = 8.2 Hz

For me, it would be a huge pain to change the choke, so I think the thing for me to do next is simply add a small additional amount of DC resistance in series with the choke (i.e. a hybrid choke/resistor Pi filter) and see where that takes me......

Thoughts? What BTW, was the choke you used in your 5G9?

[tubeswell]

Choke 40mA 4H Fender Deluxe / Champ

Hmmm, could ~40R make a difference?

[Wombaticus]

Choke 40mA 4H Fender Deluxe / Champ

Hmmm, could ~40R make a difference?

Seems unlikely, dunnit? Especially since it has a lower inductance, even with the very slightly higher DCR, and you are presumably using the original 20uF screen/trem supply cap. I even got to wondering if "inappropriate" coupling of the LFO with the plate supply might even be an integral but unappreciated part of the bias-wiggle trem sound....?!?

Nevertheless, I'll try some additional series resistance, see what happens, and report back. The fact that higher capacitance at that node significantly stabilized things for me leads me to believe that I'm at least on the right track... although the fact that it now seems to depend on some component significantly heating up before helicoptering kicks in maybe argues to the contrary... End of the day, I'm guessing it's gonna be a combination of factors.... current flow through the trem, grounding, node capacitance, total impedance of the choke(+resistor), etc. This, I guess, is the part of amp building known as "paying the dues".

[tubeswell]

I'd venture that different grounding may make a difference. The 'creek-stream-river' system might suffice for some amps, but to my mind it would still have the potential such that the busses between different 'rivers' may cause unwanted accentuation of differences in ground potential - esp with an LFO circuit pulling on the screen node cap. Not sure about how much of a role those FETs play in current draw compared to tubes, but that could be a part of it as well.

And yep - my 5G9 has the fairly bog standard 20uF filter caps all round

[Wombaticus]

I'd venture that different grounding may make a difference. The 'creek-stream-river' system might suffice for some amps, but to my mind it would still have the potential such that the busses between different 'rivers' may cause unwanted accentuation of differences in ground potential - esp with an LFO circuit pulling on the screen node cap. Not sure about how much of a role those FETs play in current draw compared to tubes, but that could be a part of it as well.

And yep - my 5G9 has the fairly bog standard 20uF filter caps all round

Tubeswell,

I happy to report that the problem now seems to be completely resolved... I played it for several hours tonight with no sign of helicoptering. Adding 470R of resistance in series with the choke on the B+ rail did the trick.

But while the problem is evidently fixed and the amp is sounding great, I have a nagging uneasiness that I never quite understood the exact nature of the problem itself.

Certainly, the fix was at least consistent with the insufficient low frequency filtering theory. But it doesn't explain why the similar or lower levels of filtering were sufficient for the Fender original and your 5G9. The FETS aren't sucking back a great deal of current -- no more than the valve would have -- so maybe there's something yet to your grounding idea. It also doesn't explain the fact that after my changes prior to the 470R resistor addition, that it would take 30-45 minutes from amp power up before manifesting itself. That *has* to be heat-related, doesn't it? But, ah well, it's working -- I just wish I could draw some more conclusive lessons from this.

So, many thanks for all your advice and help with this project -- it was your 5G9 clip that really convinced me to go for a Tremolux-ish build, and at the end of the day, I too am quite pleased with the result!