Filtering switch-mode ripple?

[Svart]

Simple decoupling should do for any residual noise.

Floating supplies?  I'll assume that they mean the same as "isolated" which means that there is no ground reference from input to output.  If that is the case then yes you can series them up for 48v.

EDIT: yes they datasheet says the outputs are floating.

 

[e.oelberg]

here is a super low noise switcher:
http://funk-tonstudiotechnik.de/SMPS-12-2006.pdf

scroll down for the test results

nicholas

 

[jdbakker]

here is a super low noise switcher:
http://funk-tonstudiotechnik.de/SMPS-12-2006.pdf

scroll down for the test results

Looks nice, on the surface.

Their test results only go up to 20kHz. That's not quite sufficient, as switchers usually have more energy at higher frequencies and this can get demodulated into the audio band, as was discussed in this thread.

(I always wonder why manufacturers only provide limited data. Is it because they don't know it's relevant? Is it because they know it's relevant, but they're afraid it'll confuse their customers? Is it because they can't be bothered to do full measurements? Or do they deliberately hide flaws? While I think the folks at Funk are capable, I'm a bit disappointed.)

JDB.

 

[e.oelberg]

thats true, though I'm sure this guy measured up to anywhere possible, funk's stuff is incredible clean and SMPS he builds are more expensive than his linear supplies
which are also the most clean you can get.

 

[earthsled]

Hi all,

I'd like to pick up this discussion again and ask for some help in a specific situation...

I have two inexpensive Mean Well power supplies, one is +24V at 1.1A, and the other is +48V at 0.313A. I'd like to try these supplies with a Langevin AM-17 program amplifier. The AM-17 requires 1A of +24V in "low power" mode according to the spec sheet. So my first question is, am I allowing enough margin with an SMPS rated for 1.1A?

Next, I am still searching for a procedure to determine a proper filter for the +24V rail.

Svart said...

I'd still like to know what the ripple frequency is.  That will determine how to filter it properly.

But for a simple filter I would go L/C, either ferrite or inductor.

Dan Kennedy said...

For junkbox filters, I've been using the 1 amp size inductors that are pretty much targeted to SMPS
use. They're available in surface mount and leaded styles, in values form 10's to 100's of uH, and are
cheap, in the $1 or so range.

That and some not too fancy caps and you can reduce the output garbage to tolerable levels in a
small space.

So, it sounds like I should go ahead and connect the power supply to the unit without a filter, and see how clean the DC rail appears on the scope. If I see a measurable ripple frequency, then I should be able to calculate the proper inductor-based filter. Is this correct?


With regards to the +48V phantom power, Abbey said...

It does not need to be regulated anymore, but needs to be filtered passively with a cascade of RC's (47R/1000uF)

I assume "cascade" means one RC filter is followed by the next. If this is so, how many repeats are necessary considering the phantom power requirements of a mono mic pre?

Thanks for your help!

 

[joaquins]

How many stages will depend on how many ripple you need to take out, maybe you should test it on the go... You could make a lot of numbers as low pass filters and time constants, but in real world caps are not only caps, series resistance, smearing, etc may let pass more hi freq that numbers said... When you use 1000µF cap is always a good idea to use a bypass ceramic cap, specially when filtering hi freq stuffs... even 2 ceramic caps is a good idea, maybe a 100nF crappy one and 1nF NPO is a good choice, NPO is not so overpriced if you just need to do a couple of devices, they are expensive in massive production, but a a couple of bux for a bunch of caps for a one time project it may worth it.

JS

 

[Rochey]

Ermmm this is more or less exactly whas done on expat audio's CRC and Uber PSU boards... Except we're usually filtering mains noise.

 

[earthsled]

Thanks joaquins and Rochey for the suggestions!

Except we're usually filtering mains noise.

Interesting...
Are switching power supplies more susceptible to mains noise compared to linear types?
If so, I have one of those AC inlet EMI filters to try.

 

[Rochey]

earthsled.

In my limited experience, mains noise tends to be lower frequency than SMPS noise. However, it also depends on what else is sharing that mains ring.

Susceptibility to mains noise varies from power supply to power supply. PSRR is a function of the feedback loop used to regulate the output. A Linear supply with regulators can have excellent PSRR, if used with an external opamp for instance. Many folks use an SMPS followed by a linear regulator to filter some of the noise.

/R

 

[MagnetoSound]

... I have one of those AC inlet EMI filters to try.

Isn't it also about keeping the switching noise off the mains supply?

 

[earthsled]

I finally had some time to experiment this morning, and I'm happy to report the +24V SMPS performed better than expected. I didn't hear anything strange in the noise floor when I listened over nearfields. I haven't hooked up the scope yet, but I'm not sure if this will be necessary.

Looking at the schematics, I may have lucked-out with this particular unit because the AM-17 circuit includes a low-pass filter section on the +24V rail. According to the documentation, this filter is to prevent "television sync-pulse interference," so I imagine it may be helping with any high-frquency crap that might be on the rail.

 

[Rochey]

great to hear you had some success!

Do you know what rate your SMPS is switching at?

 

[earthsled]

Do you know what rate your SMPS is switching at?

The data sheet says 60kHz.

 

[Rochey]

That is surprisingly slow.
Is there any kind of digital (such as an ADC) nearby? Even though it's out of hearing range, there is a danger, if the output goes to an ADC input, that the out of band noise (OBN) can be coupled down in-band. I strongly suggest some low pass filtering at 20kHz or so.