Circuit for smoothing power supply output - help please

ags

I am designing a board that will use the Propeller and Wiznet module powered with 3.3v and other ICs at 5v. I am using the Propeller Proto Board USB design as the basis for my board power supply. I'll have a LM1086-3.3 providing the regulated 3.3v supply, but instead of that being supplied by the output of a LM1086-5 on-board, which itself is supplied by 6-9v DC from off-board, I will be using an off-board regulated 5v supply, which means I won't be using the LM1086-5 in my board design. This all sounds good so far. Unfortunately, the 5v supply is a 100A 5V supply, and when I looked at the output with an o-scope it looks like there is about 200mV PtP noise on the output. I'm concerned that if I use that 5v source without further smoothing, it will cause me problems.

Question: can anyone point me towards a circuit design different than the typical (such as the LM1086) regulator, but more like a filter that will smooth the output of my 5V supply. My first thought was to just put a big (22000uF) electrolytic capacitor at the 5V connector; then I read a bit about a Pi network. What shall I use? Thanks for any help.

Phil Pilgrim (PhiPi)

What's the frequency of the noise?

-Phil

ags

I was guessing someone would ask that. I'll have to look again. It appeared to be random noise, not periodic, but I should be able to get some more details. It's a brand-new (to me) oscope and I've not yet read the manual; I wonder if there's a way to capture and upload a trace of the screen? I'll have to look.

kwinn

A 100A 5V power supply is almost certainly a switcher so the noise is most likely at the switching frequency. A small inductor followed by a 10-22uF and a 0.01uF capacitor in parallel should take care of the noise.

ags

That's correct. It is a switching supply, and the specs read 100mV p-p noise & ripple.

Will you please quantify what a small inductor value is?
Would the 10-22uF cap be a specific type? Aluminum electolytic, tantalum?
Is the 0.01 uF cap a ceramic?

Thanks.

kwinn

A tantalum would be better but an Al electrolytic would also work. The 0.01 would be ceramic, and it would bypass most of the high frequency noise. These two are for the input to the 3.3V regulator. You will also need capacitors on the output. A 50 - 100uH inductor and/or a ferrite bead could be placed on the input of the +5V (in front of the capacitors) if you were powering a low noise analog circuit. For powering a digital circuit this would not be necessary.

LoopyByteloose

A 100 amp 5 volt switcher? Doesn't that require something like 10% minimum load (10amps) in order for regulation to work?

ags

I just checked the specifications (here: http://www.mpja.com/download/1840ps.pdf) and it reads "Output Current Range: 0-100A". That doesn't say it will be regulated at 1A, though.

This will be driving a LM1086-3.3 (which shouldn't be a problem, the 3.3v out should be clean (with proper capacitors as specified in manufacturer datasheet) and 74HCT595 shift registers - which is where I'm concerned the noise might be a problem, even with 0.1uF ceramic decoupling capacitors close to the VCC/GND pins.

kwinn

A 100A power supply to drive an LM1086 seems like a bit of overkill. I just assumed the LM1086 and whatever circuitry it powered was an addition to what was already connected to the supply. Some switchers will work and regulate with no load, some won't. The LM1086 is essentially no load as far as this power supply goes.

Good luck with it, and please let us know how it works out.

Phil Pilgrim (PhiPi)

What else does that supply drive that needs so much current? If you're really using that much current in other circuitry, chances are that load transients will be a much bigger issue than noise. If you really don't need that much current, but are simply using it because "it's there", don't be penny-wise and pound foolish. Quit fooling around! Get an appropriately-rated supply that fulfills your noise requirements, and move on.

-Phil

ags

You assumed correctly: the 100A supply is primarily for driving the load that the propeller is controlling, which is (depending on the particular configuration) anywhere from 3A to 96A.

Phil Pilgrim (PhiPi)

I'd use an entirely different supply for the Prop. By using the same supply, you're only making life harder than it needs to be.

-Phil

kwinn

I am reluctant to disagree with you Phil, but I do what ags is doing quite often. I use the prop to replace an obsolete board or module and usually power it from the existing 5V power supply by shorting out the protoboard 5V regulator. In cases where I use a dip prop chip on a replacement board I use a 3.3V regulator (LM317) with a 22uF and 0.01uF on the input and output of the regulator. Never had a problem with the power.

LoopyByteloose

I guess the same filtering would work on a 3amp 5volt switcher, or something else, like an ATX power supply. But I have been pretty much using separate power as it is hard to find clarity of what is a minimum load for a switcher.

kwinn

@Loopy Byteloose, yes, it works equally well with an ATX or AT power supply. The only caveat which I should have mentioned in my previous post is that the LM317's are not low dropout regulators so the nominal +5V supply should be slightly more than 5V. This has almost always been the case for me. Typically the voltage at the 5V regulator has been 5.2 - 5.3V to compensate for the voltage drop across the connectors and wiring. For an LDO regulator like the LM1086 this is not a concern.

ags

Just to clarify, I'm not trying to be lazy or save money by using the 100A 5V supply for this project, or re-purpose it from something else. It is an inherent part of the overall system and the +5V output will already be in the same enclosure as the Prop controller board, so I thought it good engineering to see if it might work as a supply for the controller board as well. (I guess I am attempting to avoid duplication of parts/functions).

What I've done is to give myself an option on the board design: with one jumper setting I can bring the +5V in from the 100A supply, with a 10uF and 0.01uF ceramic (the LM1086 data sheet calls for 10uF tantalum on both input and output, so I thought I'd use that value - if the forum wisdom is to go even larger than the manufacturer's recommendation I can do that) to ground, then out to the 74HCT595 shift registers and to Vin of the LM1076-3.3.

In case that is too noisy, another jumper setting will allow me to bring in 6-9V from another supply (wall-wart most likely) and drive a LN1086-5, which will then drive the '595s and the LM1086-3.3.

Is there any problem with the first setting, where I would be putting +5V on Vout of the LM1086-5, and leaving Vin floating?

Here's a schematic of what I mean. Thanks for the advice & help.

kwinn

Should not be a problem. You are doing almost exactly the same thing I have done when I use a propeller protoboard. I bring the existing 5V in to the protoboard power jack and short out the on board regulator. I don't think leaving Vin floating will be a problem but I am not 100% sure. What I have always done is to bring the +5V or unregulated voltage to the Vin of the +5V regulator and jumper the regulator Vin to Vout for the 5V input. Layout wise simpler and leaves the Vin capacitor in the circuit.