Noise on PSU rails with own design, how can it be filtered best.

[Ice-9]

I've just got back to my PSU that i have been working on and i am at the stage where i thought it was all finished, but i am getting noise on the rails. Its a ticking noise  i'm getting.

It's exactly as in the schematic below, and its working as it should so i built an opamp buffer to test it with (also in the schematic) When the op amp is bypassed through the footswitch there is no noise at all so i assume  the noise is being put on the +/-15v rails from the inductor / HV part of the PSU. (not RFI from inductor or from being near the audio signal as no noise in bypass)

To test this I cut the power to the inductor which i have circled in red on the schematic and the circuit becomes quiet again.

The ideas i have to filter this is,
1. If i double up on the input/smoothing caps (470uf) at the input with extra caps and or higher values will this be the answer.
2. I thought i could use low value resistors at in the +/- 15v supply rails before connecting to the opamps supply.
3. I was thinking of more/higher value filter caps on the regulator output as well.

Any ideas what else i could try are very welcome, or if what i'm suggesting is right or wrong.

Times like these I wish i had an oscilliscope.

Thanks

[Ice-9]

I opted for option 1 and have added a couple extra 470uf caps in there to filter, That has managed to quieten the ticking down near to nothing now.

[grapefruit]

Hi Mick,

What DC-DC Converter chip are you using. I can't quite read the number. I wasn't aware of any that could output 230V from 12V.

It's good that you got the ticking down. I reckon if you add another 100uH (you'll have to choose a value for maximum filtering with minimum voltage drop) inductor where your red cross is and increase the 100uF capacitor to 470uF you should get a big improvement.

Also, make sure the ground wires from the converter circuit go straight back to the main filter caps. I would probably star ground them to the 100uF capacitor (but increase this value) and then take that directly back to the main filter caps.

Cheers,
Stew.

[Ice-9]

The chip for the DC-DC is the maxim max1771, i chose this for it very low external component count and i er ! had one in my parts box. It would be quite easy to get 300v from it as well with a 2M resistor in the feedback circuit instead of 1M5 but i'm happy with the 230v i have.

You suggest using another 100uf inductor in front of the one already there, what if i just increased the inductor that i have, to say 200uh or 300uh along with increasing the 100uf cap to 470uf, would that get the same results as you suggest .

the grounds are all near the main input filter caps (ive used a solid ground plane as well)

Tonight i hooked up a 12ax7 common cathode 2 stage circuit and it seems to perform well, I wanted to get the psu working well before moving on to the audio design of the effect im getting closer now.

[Cliff Schecht]

The chip for the DC-DC is the maxim max1771, i chose this for it very low external component count and i er ! had one in my parts box. It would be quite easy to get 300v from it as well with a 2M resistor in the feedback circuit instead of 1M5 but i'm happy with the 230v i have.

You suggest using another 100uf inductor in front of the one already there, what if i just increased the inductor that i have, to say 200uh or 300uh along with increasing the 100uf cap to 470uf, would that get the same results as you suggest .

the grounds are all near the main input filter caps (ive used a solid ground plane as well)

Tonight i hooked up a 12ax7 common cathode 2 stage circuit and it seems to perform well, I wanted to get the psu working well before moving on to the audio design of the effect im getting closer now.

He's telling you to put in a choke. Essentially, it buys you a good low-pass filter that will help reduce ripple before the switcher circuit. It should also help to completely kill that ticking you're dealing with.

[earthtonesaudio]

I wouldn't add an inductor in series with your inductor.  It would kinda mess with the switcher.

[grapefruit]

Yes, what Cliff said. I assume the 100uH inductor has been chosen for that switcher circuit for a good reason so don't change that.

You already have a primary and secondary filter cap, the only thing missing is the inductor / choke in between them. If you increase the 100uF to 470uF you should then have a filtered supply and the inductance of the choke shouldn't affect your switching circuit because you have the big filter cap there. The key here is to keep all the noisy crap at the switcher, and not on your main power supply line, hence the choke - not just a bigger filter capacitor. This way you are preventing the crap from getting into your unregulated supply rather than trying to get rid of it. As Cliff said this should kill the annoying ticking.

Regards,
Stewart.

[Ice-9]

Ahh ! i see, that makes sense. In this case i would think the choke will not need to be such a high current inductor as the one i use in the switching circuit, is this correct ?
Thanks

[Rob Strand]

When you see the ticking, have you got a load on the output?

[grapefruit]

Without looking at the datasheet for the max1771 I'd say if you use an inductor  (for the added choke inductor) with at least twice the current rating of your maximum load it should be OK. Beware that this is pure hand waving but it seems reasonable to me. See if the datasheet can shed more light...


Cheers,
Stew.

[Ice-9]

When you see the ticking, have you got a load on the output?

When the ticking was present, there was no load on the HV output, only load was the op amp buffer on the Low voltage regulated +/-15 . Strange thing is now with load or no load on the HV the ticking has dissapeared. I did change a capacitor, maybe original one was bad. I'll do some more testing over the weekend and will be trying a choke as suggested earlier.

[grapefruit]

The routing of your ground (0V) wires and filtering between your low voltage unregulated supply and switching supply will be the things that make a difference. You should do the best you can with both even if you don't have a problem *now*. The op amp circuit you have in the schematic doesn't have much gain. If you put in a circuit with more gain you may have more problems.

I've never had to sort out a noise on the PSU problem without an oscilloscope, but if you can get the noise back you could try using a preamp with a capacitor coupled probe to check the power supply lines. It won't help if the frequency of the noise is above the audio spectrum but it may be useful.

Cheers,
Stew.

[earthtonesaudio]

A couple thoughts...
-What's the frequency of your switcher?  I would assume it's high-audio range if not above audio.  If the noise is coming from there you'd hear a whine, not a tick.  If that's correct, this points more toward the linear supply as the noise source.
-Are you sure you have enough of a input/output difference for the linear regulators to maintain regulation?
-I still think a choke in front of the switcher's inductor is a bad idea, unless you really know what you're doing.  Think about what the switcher is doing, and ask whether or not that 100uF cap between the 100uH inductor and a choke is going to provide a steady low impedance source of voltage at DC and the switching frequency.  Personally, I wouldn't put that much trust in a cap.

Check out this one:
http://www.calex.com/pdf/3power_impedance.pdf
And this one:
http://www.designers-guide.org/Design/bypassing.pdf

[grapefruit]

If the noise is coming from there you'd hear a whine, not a tick.  If that's correct, this points more toward the linear supply as the noise source.
-Are you sure you have enough of a input/output difference for the linear regulators to maintain regulation?

That's a point. Measure the input voltage to the 7815 to make sure you have plenty of headroom.

Stew.

[George Giblet]

If you run that HV regulator with no load it will tick (as impled by Rob Strand).

There is a minimum amount of energy the switcher can dump into the cap each switch cycle.  With no load that amount of energy causes the voltage to rise a little above the set point, the regulation feedback loop sees the output being too high and shuts off the switcher.  Because the output load is to small > 1.5MEG it takes a long time for the output voltage to bleed off the cap to the point where the feedback loop bothers to kick the switcher into action.  In other words under no load the 1.5Meg and 4.7u cause the switcher to act as a low frequency oscillator.

It is common to place a dummy load say 5% to 10% of rated output on the output.  That means the switcher can work with its normal feedback behaviour and switch frequency/duty cycle and avoids the undesirable ticking (or erratic switching) you are seeing.

[Cliff Schecht]

If you run that HV regulator with no load it will tick (as impled by Rob Strand).

There is a minimum amount of energy the switcher can dump into the cap each switch cycle.  With no load that amount of energy causes the voltage to rise a little above the set point, the regulation feedback loop sees the output being too high and shuts off the switcher.  Because the output load is to small > 1.5MEG it takes a long time for the output voltage to bleed off the cap to the point where the feedback loop bothers to kick the switcher into action.  In other words under no load the 1.5Meg and 4.7u cause the switcher to act as a low frequency oscillator.

It is common to place a dummy load say 5% to 10% of rated output on the output.  That means the switcher can work with its normal feedback behaviour and switch frequency/duty cycle and avoids the undesirable ticking (or erratic switching) you are seeing.

Spot on! A dummy load is standard practice for just about any switcher, unless it's already permanently connected to a load (at which point a "dummy" load is moot).

[Ice-9]

Thanks everyone,
There is 24v and -24v on the regulator inputs so there is plenty headroom there at 9v difference for the 15v regs The input on these ones are rated at max 30v and -30v as well so thee not over volted.

nice links eathtonesaudio, had a quick look and printed them off for full reading

If the noise is coming from there you'd hear a whine, not a tick.  If that's correct, this points more toward the linear supply as the noise source.
-Are you sure you have enough of a input/output difference for the linear regulators to maintain regulation?

That's a point. Measure the input voltage to the 7815 to make sure you have plenty of headroom.

Stew.

There is 24v and -24v on the regulator inputs so there is plenty headroom there at 9v difference for the 15v regs The input on these ones are rated at max 30v and -30v as well so thee not over volted.

I did a test with and without load on the switching circuit as advised, and NO Load = ticking , LOAD= no ticking 
so thanks to all you, I now have a quiet PSU.