Handling power glitches from compressor

[Madscientist267]

Figured this should go in 'controls', least of the offenders... Mods, of course, correct me if I'm wrong...

I'm limited on where I can physically put the new freezer, and like everything else that is doomed to become part of one of my projects, it ended up in the office.

Problem with this is, a single circuit handles the two downstairs 'bedrooms' (one of which is the office), and of course, there's a lot of experimenting (as well as full time equipment) running in here.

With nowhere else to plug it in, it shares an outlet with my grid charger for the RE, and on the first day having the freezer, I didn't notice any problem because I was running the DC system completely off of solar.

Had to rearrange some priorities due to circumstances beyond my control, and so I brought the panels inside for safe keeping, and switched the system over to what I call 'Grid Maintenance' - a couple diodes get bypassed, the grid charger gets turned on, and it floats the batteries at ~13.8V.

But when the compressor kicks in on the freezer, it's actually enough to trigger one (more likely more) of the 7 relays in the charger briefly. In all actuality, it's probably the 339 comparator responding to a spike it's seeing. But it must be induced somehow - The 'brain' board runs entirely from the sense lines that lead directly to the 8D's terminals.

It's not strong enough to trip the 3 UPS units that reside with the freezer circuit-wise religiously; however every once and again I'll hear a UPS jump to battery momentarily along with the charger. Seems to happen during the day, possibly due to heavier loading than at night. I have no idea. The charger however flinches every time, regardless of the time of day.

I've considered things like an EMI filter (got a zillion of them laying around from old PC de-mans), but seem to be in an argument with myself as to where to place it (or them).

Thought the ultimate solution would be to filter the line going to the freezer as much as possible, but these filters are rated for computers, not compressors. While the compressor only uses ~1.6A RLA (which in theory should be fine), the LRA (startup surge) is something like 12A. Well outside the rating of the filter, I'm sure. I'd bet they're no good for more than 5A or so (no labels ). Trick is, is it a short enough surge that it wouldn't bother the filters?

The other thought that came to me was to put a filter on the mains for the grid charger. But then what about the other equipment that's seeing the glitches? Might not even 'protect' them from dropouts in the same way it would smooth out the surge at the compressor by attaching it to the freezer.

Or maybe both? But is that overkill?

Here is the link to the charger, if it helps, and this is the link to the freezer (from Lowes) that is giving me problems.

Steve

[madlabs]

Mad,

An extra cap on the comparator input and supply? Maybe a large cap (~ 100uF) and a decouple cap (.1uF)? Might smooth it out enough. Reverse EMF diodes?

Jonathan

[joestue]

this is on the grid right?

wire the freezer on another circuit breaker.

too many unknowns here.

[dnix71]

Plug your Kill-A-Watt meter into the freezer and watch the PF during startup. I bought a cheap upright freezer from Brandsmart years ago that damaged my desktop computer with the spikes it caused starting and stopping. I could actually hear the pops coming through the speakers of my computer when the compressor kicked off and on.  The power supply caps failed on the computer and I had to throw it away because it was an all-in-one G4 eMac. It turned out there were 2 versions of that compressor, one with start cap and one with a klixon on the start winding. The klixon was cheaper to make so that's why it was sold that way.

After adding a start cap I got the start surge down from 700 watts to 400 watts and upped the PF to about .45. It was about .19 without the cap. The cap from eBay cost me less than $20 including shipping.

[rossw]

Thought the ultimate solution would be to filter the line going to the freezer as much as possible, but these filters are rated for computers, not compressors. While the compressor only uses ~1.6A RLA (which in theory should be fine), the LRA (startup surge) is something like 12A. Well outside the rating of the filter, I'm sure. I'd bet they're no good for more than 5A or so (no labels ). Trick is, is it a short enough surge that it wouldn't bother the filters?

While the brief surge isn't likely to be a problem for the filter in terms of overheating, I think it's far more likely that you'll have a problem with saturating the core(s) of the filter during the worst of the spike, thus substantially limiting its filtering ability.

Past experience has shown that filtering as close to the source as possible (particularly for radiated interference) is the most effective, and as such thats' where I'd start.

If you have heaps of these filters, you could likely wire two or more in parallel - WATCH THE PHASING! - to achive the same result. Otherwise, make or buy one suitable for the peak current you expect.

Once you filter there, however, you still have the problem of the peak current being drawn through the house wiring. That will be a secondary source of interference, and will be much harder to filter.

The other thought that came to me was to put a filter on the mains for the grid charger. But then what about the other equipment that's seeing the glitches? Might not even 'protect' them from dropouts in the same way it would smooth out the surge at the compressor by attaching it to the freezer.

Or maybe both? But is that overkill?

Some additional filtering on the comparator may help reduce its tendancy to fire off on spikes, and if easy would be a worthwhile addition.

One other thing I've done in the past on a particularly noisy piece of industrial equipment, and not sure if its possible with what you have or not, is to get a suitably rated solid-state relay with zero-crossing detector. (Most, but not all, have the ZCD so check, it's critical in this aplication). Many freezers have a thermostat that *directly* controls the compressor - ie, switches the whole x amps. Rewire so that the compressor is controlled by the SSR, and the thermostat now only controls the "coil". This has several effects - all good.
1. Reduces the current through the thermostat, eliminates the arcing that often happens at contact close and more so when opening.
2. the compressor will *always* start at the zero-crossing, where its current flow is zero. This is MUCH kinder on your inverter and produces far less interference
3. the compressor will *always* stop when the current is zero, which also creates minimum interference and spikes.

Good luck though, they can be hard to nail.

[joestue]

But when the compressor kicks in on the freezer, it's actually enough to trigger one (more likely more) of the 7 relays in the charger briefly. In all actuality, it's probably the 339 comparator responding to a spike it's seeing. But it must be induced somehow - The 'brain' board runs entirely from the sense lines that lead directly to the 8D's terminals.

somehow i missed this earlier.
this is the only part that matters.
if a compressor turning on/off can create an induced voltage spike that can flip the relay on your battery sense board, then something a little more powerfull could blow it up.
It wouldn't supprise me if a standard common mode filter could stop it.
Most of them don't have differential filters in them, only common mode. so they don't have a current rating other than I^2R. (and I^2t of course as well)

[Madscientist267]

I have in fact confirmed it is the comparator firing... I disabled 'recovery' mode via the switch on the back of the charger and the nuisance tripping has stopped.

Opening this switch does nothing more than disconnect the relay from the driver transistor (which is driven by the comparator).

common mode filter could stop it

On the DC input (sense lines) right at the board, or at the AC input coming into the charger?

wire the freezer on another circuit breaker.

This is the least feasible of all possible solutions. It would be way more work and considerably more money than I have to throw at it unfortunately...

make or buy one |a filter| suitable for the peak current you expect

Doing this on the AC feed to the compressor, heavy gauge wire on a common-mode wound toroid or independent inductors for hot/neutral? Here we go with the damn inductors again... LOL

get a suitably rated solid-state relay with zero-crossing detector

Hadn't thought of this, the only one I have is rated 6A, and at that, I'm not sure it does zero-cross. I have some heavy duty triacs (15A I believe?)... how hard would it be to make a solid state with the zero cross detect?

extra cap on the comparator input and supply? Maybe a large cap (~ 100uF) and a decouple cap (.1uF)

This is probably a good chunk of the problem. There isn't a whole lot of filtering at all on the brain. Two 47uF/35V caps, one on either end of the 339, tied in parallel, for the Vcc/Gnd rails, and that's about it.

I considered at the time using very small caps at the inputs, but had run across various unwanted side effects with this before, and so opted to leave them out. Guess it bit me.

As you can see now, there's not a whole lot of room to add anything to the board, and even using the back side is out of the question for anything of any physical size. Things are just too tight in the case to try and add anything else. 


Now the trick is, what's going to be the most effective (and cost effective as well) method of fixing this? LOL

Steve

[rossw]

make or buy one (filter) suitable for the peak current you expect

Doing this on the AC feed to the compressor, heavy gauge wire on a common-mode wound toroid or independent inductors for hot/neutral? Here we go with the damn inductors again... LOL

If I had to make it myself (I'd rather not), I'd say a common-mode rejection, along with filter caps to earth from both sides would be preferable.

get a suitably rated solid-state relay with zero-crossing detector

Hadn't thought of this, the only one I have is rated 6A, and at that, I'm not sure it does zero-cross. I have some heavy duty triacs (15A I believe?)... how hard would it be to make a solid state with the zero cross detect?

Triacs will fire at just about ANY part of the cycle - EXCEPT at zero crossing
Making the zero-detector is just way more effort than its worth. You can buy them cheap as chips on ebay, or only a little more from most hobiest electronics places.

Ebay 110651456635 (25A with zero-crossing) $10 buy-now, freight-free.

[joestue]

IRL, 3 capacitors would probably fix your problem.

but like i said earlier, the root of the problem is your dc voltage sense lines aren't protected.