I have an inverter/atx compatibility problem.

Thanks in advance for reading this post!!

The inverter is a modified sine wave Coleman Power mate 2000w/4000w model with GFCI. Oh, BTW, Coleman's technical helpdesk was not helpful. Using 6 x 6v batteris in a series/parallel config to yield 12vdc. wired two batteries in series and had three of these banks parallelled.

Frequently during the day, when commercial power is transfered from local gas plant to the nuclear facility, bumps occur in the power. Linux really loves these unplanned interuptions yes, I could use a vanilla UPS, but my plans are to put my home entirely off the grid.

The inverter ran two of my cisco 4000 routers for load testing of the batteries. I was using a desulfator to reclaim some older batteries. the batteries will hold up two cisco's for several days. Runs a table saw with no delay or drag, although it does suck a lot of juice from the batteries.

I initially had trouble setting everything up. the inverter kept detecting ground fault when everthing was wired per NEC (National Electrical Code). I do know how to wire this stuff up. After all, I was an electrician for years.

I tried a sacrificial lamb power supply to start performing diagnostics. plugging the power supply into the inverter removed the ground fault so long as the power supply was completely isolated from earth and only used the ground from the powersupply cord.

the inverter ran the ciscos on and off for more than two months, but I am sure this cannot be a good thing to continue with this practice.

The diagnostics revealed voltage on the power suppy case to ground, tried two different ones. The ciscos have the same voltage on the case to ground when tested later. I tried another cheaper inverter and have the same results. Friends tested their inverters and discovered the same. All of the ones tested are transformerless inverters. each show 67vac hot to ground 67 vac neutral to ground & 120 hot to neutral.

I do realize that the neutral wire in this case will be a load carring conductor instead of a phase balancing grounded conductor as it is with true single/three phase. but, I did not expect to have voltage floating around on both power supply cases.

I tried the power supplies on commercial power to test for changes and they did not have voltage from the case to ground which is what I had expected.

Thus, I summized that all inverters will operate with a bastardized version of single phase and that ATX power supplies are designed for 3wire single/three phase. I consulted with a guru of sorts about what I had observed and has summized. He did not know what to make of the results either or why they would be behaving in this manner. He tended to agree with my assement for lack of any better explaination.

The two solutions that I have thought about are:

1) Replace the inverter with two that are stackable that would give a true 3wire + ground single phase environment

2) or add a center tapped transformer to the existing inverter to step the voltage to 220 3wire + ground single phase environment.

3) try some type of power/line conditioner

Any thoughts?

Is your inverter grounded or floating? The inverters I've used have a ground lug that should be connected to true ground. If it's not grounded, then that might explain the case voltage.

You may or may not know, but a modified sine-wave looks like a choppy square wave. In the audio world, they call this audio clipping. If you connect the output to the oscilliscope you will see this. If you compared it to commerical power, it will look very different. A modified sine-wave is easier to make because one digital chip and a mosfet can be used with out producing a lot of heat. Use a Class-D amplifier, a sine wave generator set at about 60 hertz, and a transformer to output an AC voltage instead of using that inverter. For a mini-power plant, 24 to 36 volts for battery power is better because low gauge wire does not need to be use which can create problems like over heating.

Did you use an RMS multimeter or did you use a none-RMS multimeter.

Until you find a good inverter, number 3 will be your best bet to use for your commerical power lines.

On my power lines I get a 117 volts for common and hot. From common and ground, I get 0 volts. For hot and ground, I get 117 volts. A device from either Radio Shack, Home Depot, Lewis, and other hard or electronic stores that detects mis-wirings and good wirings. It tells me the wiring of my AC power is correctly wired.

Thanks for the prompt responses.

macemoneta: the inverter has an earth ground lug & when all is wired per the NEC (National Electric Code) ie. all grounds connected, it detects ground fault only when the computer loads are connected. everything works if the load is isolated to the inverter & the inverter is grounded to earth. it also works well per NEC if connected to other loads like a motor or lighting load. but motor & lighting loads are isolated from ground unless a fault occurs.

Electro: I am unsure if the digital meter I have is RMS. It is an emco. It has a switch to change from reading dc to ac. I suspect this may change it from non-rms to rms. it has connectors to test npn & pnp transitors, capacitors, diodes etc. plugging the tester you mentioned into the inverter causes an immediate ground fault even though I have not hit the GFCI test button. BTW, when the load is isolated directly to the inverter, everything works ok. However, this does violate the NEC and is not safe.

I do understand why you recommend the 24-48 vdc system, but this was one phase of several phases to come. Each is a step towards pushing my home off the commercial grid. With a computer power supply, does one need to employ a true single phase 3wire (solution 1) or will one of the true sine wave inverters produce different results? I understand that sine wave is better for the equipment, but fail to see or understand how the wave form will change the loaded nuetral condition. I have no reference or equipment by which to test and compare.

I suppose that I can use my current inverter I have for lighting & motor loads and employ (solution 1) for the computer toys.

If the inverter is grounded, and the high-voltage side ground is not equal to the inverter ground, then there is a split ground in the inverter. There should be no potential between the inverter ground and the ground on the AC side, otherwise you have a potentially hazardous condition. While downstream equipment may operate normally, you have an electrocution and fire hazard. Even using it with a double-insulated load like the motors you described is not advisable.

macemoneta:

That is exactly my point about this condition is unsafe. This unit violates the NEC for I must bond the grounds to earth. all the ones that I have tested which are of the transformerless type demonstrate these same characteristics.

I am at a complete loss on how to resolve this. I would hate to by stackable inverters only to find out I am in the same situation.

This type of inverter is generally for mobile use, not to be wired to existing circuits (or mixed with wired equipment).

What you can do is electrically isolate the inverter and batteries, and tie the AC-side ground to true ground to create a safe AC circuit. All grounds on the high-voltage AC side will then be at the same potential, and all voltages will be to the same reference.

The risk is that if the inverter or battery is inadvertantly connected to a ground, the inverter will be blown (if you're lucky; the batteries could explode as well).

Forget your Coleman inverter because you will not be able to make it completely safe to city standards. Use a true sine wave inverter instead. I found several at http://www.4lots.com/.

As an alernative option have you considered taking the 12vdc from the batteries and converting it directly to the voltages that the computer power supply outputs? It would be more efficient than the 120vac -> 12vdc ->120vac -> power supply output that you are running using the inverter, of course for true power backup you will need an inverter for the monitor.....unless of course you have a LCD monitor that takes a dc input from an external power supply. My 17" envision lcd takes 12vdc from a small power supply, i dont know what other monitors have similar voltage requirements as this is my only lcd.


Mike

Thanks to all for your input.

macemoneta: Thanks for the advice.

Electro: Thanks for the link and advice. I am checking things out on the link.

mike105105: I have thought about doing just that, but have found very little in the way of schematics to creat the pos/neg voltages needed by motherboards. I do think you are right that this would be the most efficient way to run the computers. if you have links to a good schematic for ~500 watts that would be great. most of the schematics I have found are for 200-300w. Also, will be for the time being running a 12vdc system