Are smoothing capacitors necessary in my case?

[exclamationmark]

Kind of an afterthought, as my existing pre-amp is working fine, with just a single huge smoothing capacitor supplying 4(!) triode halves of 12au7 in a rail arrangement. Though funnily enough, I've adopted a multi star ground system in contrast to the apalling power scheme... B+ is only 48v supplied from a ridiculously filtered SMPS laptop supply. The ripple is actually measured in microvolts. So I'm wondering, are dropping resistors/more smoothing capacitors actually beneficial in my case? (Excuse the paint diagram... The lines going directly to ground are meant to represent the generic triode stage)

While were at it, how does this 'low frequency instability' actually propogate within the B+ rail?

[Chuck H]

Not much to see in your diagram... I would have additional filters. It's true enough that you don't really need them for "smoothing" the DC but there is "decoupling" to consider. The large value aluminum capacitors have considerable resistance at certain frequency/impedances and at the very high amplification ratios of instrument amps even the resistance of the ground leads and chassis can have an effect. A very small amount of ripple appearent in the preamp will be VERY audible at the output whn it's been amplified between 100 and 10,000 times. Sharing power amp and preamp filtering without decoupling is an invitation for ripple induced on the early pramp stages IMHO. If a great big capacitance were all that was needed I think the big MFG's would have caught on by now (cost efficiency/parts count/etc...). You can try it for yourself but I would suggest leaving chassis and board space for additional filtering in case you decide you need it. One tip... No one else is doing it this way. Maybe because it doesn't work and additional decoupling is needed.

[exclamationmark]

This design in particular is just a rack mount pre-amp. No power amp section. It works fine as it is but I can't help but wonder, "what if I had done things the proper way?". I don't want to drop the B+ any lower, hence the lack of dropping resistors. Could I perhaps include something like a 47uF capacitor in between every 2 stages to de-couple any noise? Since I don't have any dropping resistors, I understand this won't form a filter of any sort, but should still help with the stability, right?

EDIT: I just realized because the number of stages was even, it wasn't oscillating by some sort of dumb luck miracle. Back to the drawing board I guess. Is there any other way I can preserve my B+ while attaining sufficient decoupling?

[exclamationmark]

Would something like this work? Basically stolen from the SLO100 power supply. Would also be handy, because I can keep my rail supply thing I have going, and just jumper extra wires across to the respective plate resistors. The voltage drop should be minimized this way.

[es345]

You mention to use a Laptop supply. I would add 100nF in parallel to each electrolyt capacitor to filter high frequency spikes

[exclamationmark]

You mention to use a Laptop supply. I would add 100nF in parallel to each electrolyt capacitor to filter high frequency spikes

I actually learned this the hard way - Now I have a CLC filter at the input so the high frequency hash is pretty much demolished. Thanks for the suggestion though!

[Chuck H]

It's not just ripple that gets decoupled, it's also whatever signal is riding on the power rail. If you don't decouple it you can have problems, including oscillating. Your idea in post 4 should work fine, though you may only need two branches if all four triodes are in series (one for each tube rather than each triode). I know you don't wan to drop anymore volts, but if the amp needs more decoupling you can't simply wish the problem away because it's inconvenient. You NEED to decouple between stages of the same phase. At that point, if you NEED more voltage you'll have to deal with that.

[exclamationmark]

It's not just ripple that gets decoupled, it's also whatever signal is riding on the power rail. If you don't decouple it you can have problems, including oscillating. Your idea in post 4 should work fine, though you may only need two branches if all four triodes are in series (one for each tube rather than each triode). I know you don't wan to drop anymore volts, but if the amp needs more decoupling you can't simply wish the problem away because it's inconvenient. You NEED to decouple between stages of the same phase. At that point, if you NEED more voltage you'll have to deal with that.

Electronics seems to be very much about compromise... While we're on the topic of phases in respect to decoupling, could I get away with running a dc coupled cathode follower and a pentode off the same node? I have something in mind, but am running out of board space! I still don't really understand the do's and don'ts in regards to inverting/non inverting stages.

[diagrammatiks]

if you're only running 48v the ripple voltage isn't going to be that high and the tubes aren't going to be working hard enough to go into hard blocking/motorboating.

[kg]

if it ain't broke don't fix it!