Need help filtering power generator output

[sprok]

.

[bitslice]

I don't really know, but I was wondering if you used a 3 phase Thyristor controlled rectifier bridge and then shifted the phase slightly, then it would avoid triggering while the input was glitching.

[uncle_bob]

Hi

What you have sounds like an automotive alternator. The common way to quiet them down is to attach a lead acid battery to them.

If that is not an alternative, can you isolate the ground on the DC supply from the ground on your other "stuff"? Put another way, can the alternator plus rectifier float and not mess things up? If so, you can choke both leads to the device. That will go a long way to solving your problem.

Bob

[edpalmer42]

How much current are you dealing with?  A solution that works fine for 1 amp might be horrible for 100 amps.

Ed

[SeanB]

Have you tied a 1000uF or larger capacitor, this should definitely reduce that pulse. If the black box is a battery charger it will have pretty much zero filtering, relying on the battery to do the filtering, so it really needs a lot of output capacitance to emulate the battery.

[uncle_bob]

Hi

How much do you have invested in this solution? (Is this simply a junk box experiment)

Why are you using it over another approach? (Is it running off of a turbine ...)

What does the system need to do once it is up and running?

How much can you afford to put into a solution? (Is $10 to much money?)

Lots of things that could impact the answer.

Bob

[uncle_bob]

Hi

Ok, if you already have a battery in the circuit *and* the battery is ok ... there is a ground issue. Try pulling the power supply feed directly off the battery terminals and see what happens. If you still see the spikes, the battery is likely shot. If they go away ... problem solved.

Bob

[CT2718]

It sounds like you are tapping the alternator output of one of those "Inverter" type of generators (e.g. Honda EU1000i), and it looks as though the spikes are probably the result of circulating currents in an inadvertent ground loop.

One thing that is not entirely clear is whether or not those spikes are the result of some other phenomenon and not related to the inverter.  Several questions to ask include:

1 - Any obvious ground loops in the set up of the scope?  If the scope is being powered from the mains voltage "black box" inverter on the generator, all bets are off.  (Is the original "black box" mains converter still present?  Is its input "floating" or "ground referenced"?

2 - Placement of the connection of the scope leads to make the readings.  If the ground lead of the scope was connected on the "common" ground at the input of the switching converter rather than the output, that could affect the magnitude of a pulse seen on the scope.

3 - Is the spike being conducted from the input to the output (perhaps via the ground lead) by input capacitance of the switcher?  You didn't mention whether or not the "switching converter" providing the 5 volts was under load or not.  If it is under a small/minimal load, place something in series with the voltage input of the switcher (a few ohms resistance, a large chunk of wire in a coil to provide resistance and inductance, etc.) to see if this dramatically effects the size of the pulse.  (This may also be a "ground loop" issue - but if the switcher itself is a no-name EvilBay type, all bets are off!)

4 - Is this an artifact of the switcher itself in responding to the spike in the input?  Stray reactance could cause the spike on the input to propagate through or cause the switcher itself to do some sort of odd glitching of its own.  Again, no-name switchers can be suspect - particularly for buck-type.

I'd bet on #2 or #3 being the most likely (but definitely not rule out #1.)

The fix for this would be to place a significant amount of inductance in series with the input to the switcher - several TENS of microHenries at the very least, preferably a few hundred.  Any capacitors to be added - which must be placed "downstream" of any inductance - must absolutely be of the "Low ESR" types intended for switching supplies ("normal" capacitors cannot be expected to work well) and the smallest value that you even want to start messing with would be 470uF - but I suspect that much more will be needed.  (Again, if adding capacitance doesn't change anything, suspect a ground loop!)

You did not mention if this switching supply and its loads would be completely isolated (e.g. floating) from the generator in all ways (e.g. not connected in any way - directly or indirectly to its chassis or to the mains voltage converter onboard) but if it IS going to be referenced back somewhere, you will probably want to put such a choke in both the positive and negative input leads of the converter to reduce the probability of circulating currents.

Best of luck!

CT

[Circlotron]

Some outboard motors with permanent magnet alternators have an SCR that simply shorts out the stator winding partway into the half cycle as a means of voltage regulation. Relies entirely on the inherent current limiting of the alternator. The gradual rising and sudden fall of the waveform makes me think that is also what is going on inside that black box. If all else fails and nothing else is being driven off the alternator maybe it would be easier to make another regulator, but only big enough for the intended load, not the full alternator capacity as it is now.

[Rerouter]

Add an input LC to your device?

The fact your seeing the spikes passed along to your output implies 1 of a few things, either your regulator has a horrible PSRR, you have introduced a way for this signal to couple in, or your not measuring from the correct reference point,

Your spikes are coming in at 1.5KHz, so use a 1mH or larger common mode choke and a 220uF input cap, and you will knock down the worst of it (The actual frequency components of the spikes are much higher so will be attenuated quite heavily)

[JPortici]

Ferrite beads! they do wonders in removing spikes from switching regulators.
And an LC filter for the lower frequency components of the ripple

[CT2718]

The fact that the choke in the ground made a difference strongly implies the presence of circulating currents/ground loops.

Make sure that the choke(s) that you use have hundreds of microhenries inductance:  Much less and the spike depicted - which has significant low-frequency components - need a significant amount of reactance in series.  (Its worth noting that simple ferrite beads - unless you can get a few dozen turns of wire in them - simply won't provide enough series reactance to do much good.)

If your current is quite low (less than 5 amps or so) you may be able to use one of those mains-type EMI filters such as the Corcom units.  Many of those made by reputable manufacturers have on them a schematic depiction of the internals on them and many have have millihenries of inductance.

Whatever you do, make sure that you add bulk capacitance with a low-ESR capacitor (1000uF+) only "after" (e.g. downstream) of the alternator or else this capacitance will reinforce circulating currents in the connecting leads/ground.

Good luck.

CT