Bypassing brief voltage drop due to sudden 120A*12V draw

I have a 12V lithium battery bank capable of handling over 300A surge for 30 seconds (according to mfg spec).
Using an inverter that's capable of 2000W continuous and 4000W surge, I'm trying to power a small window AC unit.
The continuous power consumption on the AC unit is 600W, but the momentary surge when the compressor kicks on is somewhere closer to 1200-1400W. It's enough to drop the voltage from the battery to the inverter by 2 volts for about 2 seconds which shuts down the inverter output until the voltage recovers.

I have 10 * 56,000uF capacitors I was thinking of putting in parallel across the inverter input terminals, hoping that might help decouple the momentary voltage drop.  Since I already have the capacitors on hand, it would be much cheaper than buying another lithium battery bank.

Any thoughts or better ideas?
Thanks!

You can do the math for how much current you can hold in capacitors, but without dirtying the back of an envelope a bunch of amps for a couple entire seconds seems like a lot of charge for just caps. 

1 amp will discharge 1 Farad roughly 70% in one second. Reservoir caps generally work over mSec time scales and small fractions of an entire time constant.

[oops] the time constant thing is for RxC discharge... e^-t/rc. Cap discharge is simple straight line  1A will discharge 1F, 1 volt per second [/oops]

In a raise the bridge/low the water suggestion, is there any way to trick the inverter to ignore or tolerate the momentary voltage dip? I suspect the air conditioner is fine with voltage sag during start up.

JR

JohnRoberts said:

You can do the math for how much current you can hold in capacitors, but without dirtying the back of an envelope a bunch of amps for a couple entire seconds seems like a lot of charge for just caps. 
1 amp will discharge 1 Farad roughly 70% in one second. Reservoir caps generally work over mSec time scales and small fractions of an entire time constant.

Click to expand...

It's certainly an ugly brute force thought, but I have about 0.5F on hand which until 15minutes ago were sitting on the shelf collecting dust. So 0.5F might buy me around 200ms before the caps discharge below 12.4V?
At full charge the battery bank sits at 14.4V. The inverter isn't supposed to shut down until 10.5V but in this case I think the sudden current draw makes the inverter's low-volt sensor kick on briefly.

In a raise the bridge/low the water suggestion, is there any way to trick the inverter to ignore or tolerate the momentary voltage dip? I suspect the air conditioner is fine with voltage sag during start up.

Click to expand...

I kind of like this idea better--but I'd rather not crack open the hood and void a warranty on a new inverter (that I might have to return :-X).

Perhaps a big honking power resistor in series with the AC unit?
After the compressor is up and running maybe bypass it with a relay...or not.
Assuming the AC unit is fairly tolerant of voltage sag, maybe I could switch a big MOSFET around 80% duty at startup to reduce the inrush current to the AC--maybe $10 in parts.

Additional thoughts and ideas always appreciated!

*SHORT* fat wires from battery to inverter? The battery rating suggests car starter-motor wires, actually truck starter wires, and preferably shorter than a typical engine-start application.

I have similar trouble with a deep well pump. The start-current clearly assumes I am connected to a megawatt dam generator. On my too-long wire to the street, my lamps dim when the pump starts. I have started it on a portable generator but it is clearly risky. (Even the running current is 84% of the generator's 1-phase rating.)

PRR said:

*SHORT* fat wires from battery to inverter? The battery rating suggests car starter-motor wires, actually truck starter wires, and preferably shorter than a typical engine-start application.

I have similar trouble with a deep well pump. The start-current clearly assumes I am connected to a megawatt dam generator. On my too-long wire to the street, my lamps dim when the pump starts. I have started it on a portable generator but it is clearly risky. (Even the running current is 84% of the generator's 1-phase rating.)

Click to expand...

I was hoping to hear something like that because my first knee-jerk thought was maybe I need fatter/shorter copper. From battery-breaker-inverter-breaker-load I've got 1/0 awg @ ~12ft total. The RV guys might balk saying I should be using at least 3/0.  Measuring the V drop under normal conditions from battery-breaker-inverter input terminals, I've got ~10mV difference.

I'm going to try hanging 0.5F on the input to the inverter tomorrow, see what happens and report back.
How does a lifepo4 battery typically react to big transient loads? Is it supposed to sag? Or is the V-drop mostly due to skinny wires, or is the inverter just shutdown-trigger happy?

Fun, fun! ;D

Is it a true sine wave inverter? Motors really hate that stepped square wave crap.. And the stepped square type inverters hate motors, the inductance drives them bat-shit crazy, they just don't work and play well together. I tried a stepped square wave inverter on a gantry crane motor in a boxtruck, it loudly protested. It wouldn't even move at all with no load,  that crane motor would not be all that different of a load than an A/C compressor.

If a sine wave inverter, it could be that the inverter has output current sampling/limiting, which dumps it off-line to protect itself at the start-up surge This is likely.

Gene

JohnRoberts said:

raise the bridge/low the water

Click to expand...

Or move to a more temperate climate?

Ethan

Will the surge current damage the battery? 
What cell chemistry?
Are the cells in a metal  containment enclosure?

Two people I know that use high discharge rate Li cells with model helicopters store them in metal boxes one uses sand around the pack in a metal box.

@Gene, yes it is a true sine wave inverter.
@Gus, according to the mfg it should be ok handling 300A surge for about 30seconds. It's a LiFePo4 . From my understanding it is slightly lower energy density than the more familiar "lithium-ion", but doesn't suffer from the same kind of thermal issues the former is known for...that would explain why I haven't felt the battery even get warm under moderate load (can't say the same for my laptop).
It's just the sudden startup current thats causing trouble.  Today I tried a beefier inverter, pretty much the same result. Maybe I just need more battery. :-\ ($$$).

#0 Copper
0.1 Ohms @ 1,000ft
12ft = 0.0012 Ohms
120 Amp surge
0.144V drop

BUT: <<0.0012 Ohm connections are big and perfect.

Ethan said:

The continuous power consumption on the AC unit is 600W, but the momentary surge when the compressor kicks on is somewhere closer to 1200-1400W.

Click to expand...

I missed this part when I posted last night, I think you are greatly underestimating the initial start-up surge, not to be confused with the longer coming-up-to-full-speed current that you likely measured.

The initial current draw of a motor starts at the LRA, or Locked Rotor Amps, since when power is applied, the rotor is sitting still. This surge may only last a few AC cycles until a rotary magnetic field is established and rotation starts, the AC coupling between stator and rotor drops down from 60hz to about 2hz, and the current tapers down. How long that all takes depends on the flywheel rotational mass/inertia of the rotor and anything heavy (scroll compressor and counterweight) connected to the shaft.

A rule of thumb with AC motors is that LRA is generally 5-6 times the full load amps. Your A/C compressor at 600 watts is not really loaded to full power, so the LRA may well exceed the 4000W max of the inverter. I would guess that the shutdown circuits in the inverter are really quick to respond to output overload, trying to protect semiconductors within.

If you want to try something seemingly wacky, add some ballast to tame that initial surge, in the form of a cheap 50' or 100', 16 gauge extension cord between the A/C unit and inverter. This should tame that LRA surge, but might be a bit harder on the motor. If you try this, be ready to shut it down if the motor doesn't come up to speed. Motors all have thermal cutouts, but why test that circuit.

If it is shutting down from low battery voltage, adding this ballast would help on that end, too.

Gene

+1 to long extension cord, but if it works it will be a source of heat when the air conditioner is running.

Is there a similar current peak issue with compressor cycling off and back on due to thermostat?

Maybe a smaller air conditioner? You could make one from peltier devices without a compressor, but expensive and not effective for large cooling (better for cooling tiny stuff).

JR

JohnRoberts said:

...but if it works it will be a source of heat when the air conditioner is running.

Click to expand...

A minor issue.

Numbers show with 120V, 600W = 5 amps, with an assumed PF of 1. Probably close enough, in these days of green engineering.

#16 at100' both ways is 0.8 ohms, 5 A means 4V drop, 20W dissipation.  20W * 3.41214 (W to BTU conversion) ~~ 68 BTU generated by the cord. Half that, if a 50' cord works out OK.

Assuming an EER efficiency of 10, 600W draw indicates 6,000 BTU and with an additional 68 BTU cooling load in the room,  total cooling capacity drops by less than 1.2%. Nothing to sweat about. 

[silly] Or you could just throw the bulk of the 100' extension cord out the window, and duct-tape off the leaks.[/silly]

I'd stay away from Peltiers, they are one of the most inefficient quirks of physics known to man. A bank of Peltiers in a highly insulated container, drawing an almost fuse clearing amount of current from an automotive cigarette lighter socket, can't even keep a six-pack of beer ice cold, just "cool". Peltiers at best consume 20 times the energy of what they can pump on a good day, a mechanical refrigeration unit can pump 2 or more times the heat energy that it consumes in power to do so.

Gene

Gene Pink said:

I'd stay away from Peltiers, they are one of the most inefficient quirks of physics known to man. A bank of Peltiers in a highly insulated container, drawing an almost fuse clearing amount of current from an automotive cigarette lighter socket, can't even keep a six-pack of beer ice cold, just "cool". Peltiers at best consume 20 times the energy of what they can pump on a good day, a mechanical refrigeration unit can pump 2 or more times the heat energy that it consumes in power to do so.

Gene

Click to expand...

Yup, I just mentioned the peltiers in passing... I wasted a bunch of time and money making a peltier powered slow cooker (crock pot). My design thesis was cooling my kitchen while heating (cooking) food, but it was way too much work to get usable heat while simultaneously cooling the ambient. The energy used to pump the heat from cold side to hot side was actually useful when captured for my cooking application, but a final limitation turned out to be that peltier cells have a max temperature limit where their solder melts or whatever. Perhaps fine for a warming plate to modest temperature, but completely inadequate for even a slow cooker.  I ultimately bailed on the PT units and bolted some power resistors under my hot plate, so while I got it to work, it was too much drama and effort. I eventually just bought a nice commercial unit.

About the only thing peltier appears to shine for is cooling modest sized masses to very low temperatures (like testing electronics for space conditions) with no concern for efficiency.

FWIW I have a small peltier based dehumidifier. I purchased it hoping it would runs silently enough to use in my bedroom overnight but A) it has a noisy fan, and B) doesn't extract much moisture from the air. I have it running 24x7 in a typical small bathroom and it pulls maybe a quart of water a week...  The real dehumidifier in my main room can grab a gallon of water during one over-night session. 

JR

Gene Pink said:

If you want to try something seemingly wacky, add some ballast to tame that initial surge, in the form of a cheap 50' or 100', 16 gauge extension cord between the A/C unit and inverter. This should tame that LRA surge, but might be a bit harder on the motor. If you try this, be ready to shut it down if the motor doesn't come up to speed. Motors all have thermal cutouts, but why test that circuit.

Click to expand...

Of course!
I read this yesterday while wondering how I might add some resistance in series with the AC unit plug--I imagined a 1ohm, 5kW resistor would be pretty hard to come by ;D ;D
The extension cord was a much better idea. I had a cord lying around that looked about 25ft-ish, measured 0.18R from end to end.  IT WORKS!  The AC runs without tripping the inverter!  Although, I didn't check for heat on the cord.

One thing always leads to another. As pretty as the bright orange loops of extension cord may be, I'm wondering how I might make this solution more aesthetically acceptable and smaller.  I'm toying with the idea of coiling some stainless steel guitar strings to get 0.18R and make a little adapter box. Maybe the G-string.  ;D

Thanks for the help!

Glad it worked out for you

Ethan said:

One thing always leads to another. As pretty as the bright orange loops of extension cord may be, I'm wondering how I might make this solution more aesthetically acceptable and smaller.  I'm toying with the idea of coiling some stainless steel guitar strings to get 0.18R and make a little adapter box. Maybe the G-string.  ;D

Click to expand...

Did you ever mention in this thread where this A/C is located? Somehow, I was picturing some sort of travel trailer with a living space.

[silly] Perhaps put two horizontal rows of screws into the wall one row of them 3' above the other, and zig-zag the orange extension cord between them, so that it looks like a giant schematic drawing of a resistor. Tell your non electronic friends that it is "A really cool wall art sculpture".[/silly]

You don't need to dissipate 4KW except for a few milliseconds, you would probably be fine with a 100 watt resistor  Your extension cord measures 0.18? is that one way or round trip?

An aside, do power resistors have rating curves for pulse current/duration, to figure what hefty spike would blow them like a fuse?

For homemade resistors, I made a 50A car battery load tester, about 0.24 ohm. One of those "I need something right now" quick builds from decades ago. #14 galvanized steel wire wrapped around something 7/8". Just an idea. Note that when galvanized anything gets really hot the first few times, it smells like cat piss, if this is an issue, perhaps a different material, but it shouldn't get that hot in your use.

Really old car battery load tester jpeg attached, if it helps.

Gene

Gene Pink said:

An aside, do power resistors have rating curves for pulse current/duration, to figure what hefty spike would blow them like a fuse?

Click to expand...

Yes... they rate them as power overload X for time Y (shorter time higher power overload). 

I have fried lots of small 1/4 resistors and my favorite were ROHM that would glow red hot and still work after they cooled down (while resistance was a little off)  :.  Most other brands of cheaper resistors would open before burning off their paint. They do make "flameproof" resistors that are pretty much designed to not cause flames for 100X overload, presumably by opening up first.

Prudent amp designers use robust oversized emitter resistors in power stages to keep down repair costs (while oversized for normal amp dissipation, still undersized for full fault power if continuous). The amp fuse generally fails, before the emitter resistors do. I've seen some cheap value amps start using smaller SMD emitter resistors and they burn nice holes in the PCB when the transistors short. 

For homemade resistors, I made a 50A car battery load tester, about 0.24 ohm. One of those "I need something right now" quick builds from decades ago. #14 galvanized steel wire wrapped around something 7/8". Just an idea. Note that when galvanized anything gets really hot the first few times, it smells like cat piss, if this is an issue, perhaps a different material, but it shouldn't get that hot in your use.

Click to expand...

I couldn't find galv steel wire on my lookup table but 14 ga "iron" wire shows a fuse current of 51A. #14 copper wire fuses at 166A so resistance is the important factor.  http://www.powerstream.com/wire-fusing-currents.htm

I may remember that cat piss smell from working with galvanized sheet metal.

Really old car battery load tester jpeg attached, if it helps.

Gene

Click to expand...

I recently had to make a DIY fuse when the small amp driving my surround speakers, stopped playing. I have replacement fuses on order, but in the meantime I rolled my own from a single strand of 30ga copper wire tacked across the open fuse.  That DIY 10A fuse is a bit much for my puny amp but JIC there was something actually wrong, I swapped a good 1.5A fuse into the amp that blew, and my DIY 10A fuse into one that didn't (working 3 days so far with no new problems).

Kids don't try this at home... ;D

JR

@Gene: The cord measures 0.18R from 1 conductor end-to-end. So, effectively 2*0.18R 'round trip' if you mean from from hot to neutral.
You and @JR now have me strangely intrigued about this hot galvanized steel == cat piss smell. ;D

I'm not the outdoorsy type (the wife is) but I inherited a small camper/trailer with 4 solar panels on the roof that generate ~1kW/h. The lead-acid battery bank that was in there was pretty much DOA, so I replaced it with a 150Ah LiFePo4 pack. Ideally, I'd like to get another pack in parallel, but at $1k/each it isn't quite in the "practical column" of the family budget.

Yes... they rate them as power overload X for time Y (shorter time higher power overload). 

Click to expand...

I've never seen that before on a data sheet--then again, before now I can't remember the last time I felt compelled to read a datasheet for a resistor. Good to know!

> -I imagined a 1ohm, 5kW resistor would be pretty hard to come by

Use the internet!! Google "1ohm, 5kW"

One of the first hits is "TE CONNECTIVITY  TE1500B1R0J" which is actually 1.5kW, but you could get four. Newark has 98 in stock for $55/piece. We also know you could use a much smaller resistor for this short surge.

> they rate them as power overload X for time Y

The TE series is rated "3x Overload for 5 Seconds". So that's practically your 5kW right there. But I don't think we are anywhere near that power. Taking 20 Amps (at the 120V side) and the empirical 0.36 (say 0.4) Ohms we have 672 Watts peak, so a 3X smaller TE resistor is OK, there is a 250W model. Nearest fit seems to be "TE CONNECTIVITY  HSC250R47J  Resistor, Solder Lug, 0.47 ohm, 250 W, 2.2 kV, ± 5%, HSC Series, Wirewound", only $25 but only 1 in stock at Newark.

You can get look-alike resistors in eBay and I was very disappointed when I got some. (TE is the old Ohmite.)

Hmmm. The big resistors are getting scarce in distribution. Fence-wire may be your Plan B.

> galv steel wire ... "iron" wire

As I suspect you know, all "iron" today will be some form of steel. However steel is iron with "other stuff" in it, and this (also drawing and annealing) can change the conductivity over a wide range. For initial guesswork, assume "iron" has 7X more Ohms than copper the same size. But if it turns out 6 or 10 do not be surprised.

Aside from the smell, burning the galvanize off is toxic. Don't try this at home. Take it in the neighbor's yard.

Thanks PRR!

PRR said:

Use the internet!! Google "1ohm, 5kW"

Click to expand...

Who is this internet person you speak of? .  But had I consulted her, I might not have learned about the connection between cats and galvanized steel!

The big resistors are getting scarce in distribution. Fence-wire may be your Plan B.

Click to expand...

I have yet to take a multimeter to my steel guitar strings to see how practical that might be but I'm kind of hoping that works out since I've already gotten more than the $10 value out of those.

Aside from the smell, burning the galvanize off is toxic. Don't try this at home. Take it in the neighbor's yard.

Click to expand...

Ha! ;D