Looking for a well filtered, Quiet SMPS design

[goldstache]

I've been tweaking an SMPS design with a 555.  Wondering about ripple and what experiences you all have with filtering to get the best B+!
Thanks!

[R.G.]

(1) don't use a 555; they can be perverted to SMPS use, but there are much better controllers
(2) if you MUST use a 555, use a CMOS one; the bipolar 555's pull huge spikes of current right at the switching times
(3) what is your switching frequency? If it's over 20kHz, layout has as big or bigger part in making it quiet as component selection
(4) if your switching frequency is over 50kHz, layout is even more critical
(5) what are your input and output voltages? This makes a difference in the magnetics if you are designing a magnetic step up
(6) ARE you using magnetics, or is this a diode-capacitor multiplier?
(7) learn about current loops

[goldstache]

(1) don't use a 555; they can be perverted to SMPS use, but there are much better controllers
(2) if you MUST use a 555, use a CMOS one; the bipolar 555's pull huge spikes of current right at the switching times
(3) what is your switching frequency? If it's over 20kHz, layout has as big or bigger part in making it quiet as component selection
(4) if your switching frequency is over 50kHz, layout is even more critical
(5) what are your input and output voltages? This makes a difference in the magnetics if you are designing a magnetic step up
(6) ARE you using magnetics, or is this a diode-capacitor multiplier?
(7) learn about current loops

Thanks R.G.
I'm reverse engineering an already existing nixie design and am trying to grasp SMPS design concepts.  Newb.
Not sure how to determine the switching frequency
5) 12VDC in 300VDCout
6) diode/cap
And current loops will be my next search. 
In essence I have a (1) don't use a 555; they can be perverted to SMPS use, but there are much better controllers
(2) if you MUST use a 555, use a CMOS one; the bipolar 555's pull huge spikes of current right at the switching times
(3) what is your switching frequency? If it's over 20kHz, layout has as big or bigger part in making it quiet as component selection
(4) if your switching frequency is over 50kHz, layout is even more critical
(5) what are your input and output voltages? This makes a difference in the magnetics if you are designing a magnetic step up
(6) ARE you using magnetics, or is this a diode-capacitor multiplier?
(7) learn about current loopsguitar preamp/bax EQ circuit with 6021 subminis and I would like to design an SMPS design instead of my XFMR based power supply for space reasons.

[R.G.]

I'm reverse engineering an already existing nixie design and am trying to grasp SMPS design concepts.  Newb.
Not sure how to determine the switching frequency

OK. We all start somewhere. I'll aim comments at that level.

5) 12VDC in 300VDCout
6) diode/cap

OK, that helps. You're trying to get 300/12 = 25 times more voltage out. If you're using a diode/cap multiplier, that means you're going to lose about one diode drop per stage, so you're also losing 25*0.7V = 17.5V to diodes, so you're tying to get 318/12V = 26.5 times - call it 27 times as much voltage out.

It's fundamental to SMPS design that the power input has to be equal to the power output plus losses. So you're going to be putting more than 27 times as much current in by the 12V input as you take out in DC on the output. You're going to need to know the output currents to do a good job of estimating.

If you use a 555, it can do about 300ma pulled down and up on its output, and it does do both a pull up and down, which helps in certain setups like this. So a 555 by itself can't do much more than 11ma output if the rectifiers and filters are 100% efficient. Chance are you'll get decent output to about 6ma, and see it sag a lot after than. There are ways to finesse this, but just doing rules of thumb, that's where I'd start.

The bipolar 555 pulls pulses of a couple of amperes (!) for a very short time every time it switches. This spike is pulled from power and shunted through the device to ground. It is quite difficult to suppress this spike unless your grounding is very, very well done.

And current loops will be my next search. 

The item about current loops is based on the fact that all "conductors" are really resistors. That amperes-big spike through the 555 and the amperes-big spikes it will pull into and out of the rectifier chain to charge the chain, have to come from somewhere. If it comes through a piece of copper wire or traces that also supply your preamp's signal ground reference, you're wiggling the preamp's signal up and down with the spikes, and that will get amplified just like your signal.

If it's ultrasonic, you can't hear the actual spikes, but they will interact with the audio to make for audible down-converted whine in some cases.

The way to do this is to wire power and ground out to the SMPS on separate wires from the circuit ground, and put both a large-capacitance bypass cap and some small, high-quality ceramic bypass caps as close to the power stage feeding the the multiplier chain (that is, the 555) as you can get them. Something like 470-2200uF bypassed by 0.1uF and 0.01uF ceramic caps as close to the 555 as you can get them to minimize the resistance from filter caps to the switching element. The idea is that the wires coming to these bypasses carry mostly DC, and the caps act like a nearby bucket of charge so that the current needs of the power switching come out of the local bucket, and only a heavily filtered DC current goes into the bucket. The reason I mentioned loops is that the main power loop - local filter cap, through the 555, through the rectifiers and charge pump caps, back to a "ground" return, and back to the filter cap - stays as small and local as possible.

sguitar preamp/bax EQ circuit with 6021 subminis and I would like to design an SMPS design instead of my XFMR based power supply for space reasons.
Makes sense. I don't know what the plate current on 6021 tubes are or how many you're using. If it's only a couple of gain stages at about 1-2ma of plate current, you're probably good within that power limitation above.

It might work. Remember that you're making BIG voltage transitions, FAST edges, and you're trying to put this very close to your preamp circuits which amplify up small analog signals. The closer these get together, the more stray capacitance exists and the more of your power supply switching noise that's coupled into the signal path. Getting both power and amplifiers in a small box without noise problems may be tough.

So if you do run into these issues, don't be too hard on yourself. You're setting yourself a tough challenge, and don't have a lot of background for the problem. But it'll be a GREAT learning experience.