Power conditioning

[ezalys]

I'm in a room with pretty crappy power and grounding and am trying to improve common mode noise best I can.

What I'd ideally like is to just run everything off a battery, but I run a lot of experiments that go for several days and I don't want to think about a battery. I know there are these double-conversion UPSes, but I don't know if that means I'm truly ever galvanically isolated from the wall. I would personally design something with two batteries, a hold-over cap bank, and a set of relays that exchange the battery that is being charged and the battery powering the electronics. Is this how they work? I have maybe $2k to play with, so things like the Eaton 9130 are definitely an option, I just don't know how they work.

There's also the option of a ferroresonant transformer, but I don't have the impression that these do much for common mode noise. Maybe they do? What's the typical shunting capacitance from input to output? It's not so much that the incoming sine wave is dirty -- plus, all my equipment is reasonably new and has high performance LDOs. I don't imagine much differential noise is making its way into my data.

Maybe there's some fundamental misunderstanding in what I'm asking for here, if so, I'd always appreciate help asking the right question!

[ezalys]

I should say -- it seems all these UPSes and power conditioners bond grounds together, so I don't really see what they'd end up doing for lifting ground loops -- which, sorry, I don't mean to confuse ground loops and cm noise, I'm still a bit new to low frequency sensitive measurements. My experience is mostly in the microwave.

I'm afraid the root issue is that I have a measurement which requires DC to 1 GHz coupling, so breaking ground loops in the signal path is somewhat nontrivial. I would much rather lift my digitizer, which is a little 12-volt powered thing in a Hammond box with a Virtex-7 in it and optical fiber connection. It's just begging to be battery powered. I just don't want to think about batteries -- what can I say.

[ejeffrey]

A floating power supply will break the ground loop at DC.  A common mode choke will reduce common mode noise at higher frequency -- as long as you have blocked the DC ground loop so that DC imbalance doesn't saturate the CMC.  Low frequency CM noise (60 Hz) is hard to attenuate with a choke but can be reduced by using medical grade power supplies that have low primary to secondary capacitance.

After that moving your signals to balanced differential signals so you can use amplifiers with high common mode rejection.  There are fully differential opamps that can operate DC to 1 GHz with good CM rejection.

[ajb]

No UPS system is going to isolate the equipment ground for safety reasons.  The double conversion or 'online' UPS systems just always run the the output from the inverter rather than passing the input straight through, otherwise they're no different from the regular type in principle.

If your acquisition system can easily run off of a battery and already has an isolated data connection, that seems like a fine solution.  How much power do you need?  Sealed lead acid batteries are cheap, easy to deal with, and offer reasonable capacity--maybe up to 24Ah in reasonable sizes?  Maybe that would run your whole experiment with no need for a battery changeover?  If that's not enough, something like a deep discharge wet cell (like are used in golf carts and battery powered lifts) offers a ton of capacity, although is a bit more of a pain (in the back) to deal with.  Either way the chargers are cheap and readily available.

[james_s]

Would it be possible to improve the grounding? A good ground has multiple benefits, not the least of which is safety.

[NiHaoMike]

What about a motor spinning a generator through a plastic rod?

[james_s]

Wouldn't an isolation transformer accomplish a similar goal without the maintenance issues, noise and low efficiency of a mechanical solution? An isolation transformer with a UPS would provide a similar level of isolation. This all still seems like a band-aid though, working around rather than fixing what sounds like a serious problem with the electrical system.

[NiHaoMike]

Wouldn't an isolation transformer accomplish a similar goal without the maintenance issues, noise and low efficiency of a mechanical solution? An isolation transformer with a UPS would provide a similar level of isolation. This all still seems like a band-aid though, working around rather than fixing what sounds like a serious problem with the electrical system.

The point is to have substantially less stray capacitance than a transformer. The capacitance of a motor spinning a generator can be extremely small by using longer plastic rods.