# DC Voltage of three phase rectifier greater?

• posted

I have been trying to make my 10 HP VFD accept single phase input. (it works GREAT with three phase input) It gives me a phase fault. According to the schematic, the only way it detects phase fault is by drops in bus voltage.

I added a big DC capacitor and it is not helping.

I measured bus voltage.

Single phase input: 314 volts Three phase input: 342 volts

I also tried to connect L2 to L3 and it did not fix the phase fault.

** ALL OF THIS IS IN COMPLETE ABSENCE OF LOAD **

So, assuming that phase fault is detected by bus voltage:

1) Is that really true that three phase rectifier would give a higher bus voltage than single phase (given same input voltage)

2) What can I practically do to cheat this device and make it accept DC generated by single phase.

I do have a 10 KVA ACME transformed that I could use to boost the input voltage from, say, 230v to 250v or some such.

• posted

What make and model of VFD?

What does the VFD manual say?

How is it connected to the single-phase feed?

How big is the big capacitor?

Joe Gwinn

• posted

Just out of curiosity -- was the single-phase measurement taken by simply removing one of the three three-phase wires to the VFD, or by plugging it into a single-phase outlet?

And -- IIRC, your three phase comes from a rotary converter, so the balance may be a bit off. Try measuring between each pair of terminals. The voltage in the capacitor under low or no load will be purely determined by whichever phase pair has the highest voltage assuming a standard three phase bridge rectifier.

Are you sure that all three phases measure the same voltage? Check that out before you assume that you are providing truly the same input voltage.

If it is truly sensing only the voltage in the capacitor, try a Variac (with the overvoltage wiring option) to see what voltage you need to reach to turn off the fault indication. Once you find the needed voltage, subtract the line voltage from it. Look for a filament transformer or something similar which produces a bit more than that difference, and wire the primary across the 240 VAC line, and the secondary in series with the 240 VAC line to boost the voltage by that much. Obviously, you need the secondary current capability to be high enough to match what the VFD will draw at full load.

That would probably work, too. It it an autotransformer or an isolation transformer?

Good Luck, DoN.

• posted

Omegapak 10 HP VFD , Omegapak VFD 8803 Type P00CO2H.

Manual is here:

It says that Phase Fault detection is done by looking at the DC bus. At least how I interpret page 11 of the manual above, and the measurement board 7.

Through the rectifier.

3,500 uF.
• posted

Both.

Yes, I sort of agree. But I tried everything, specifically I tried connecting the phase with highest voltage to the L1 and L2 inputs.

Slightly different indeed.

After a lot of thinking, I am beginning to suspect that the VFD does more than just monitor bus voltage. I will try to ascertain facts a little better before going on the assumption that only bus voltage is the issue.

Isolation.

• posted

240 V (single phase) AC full-wave rectified and filtered, should be around 340 V.

sqrt(2) * 240

Dropping to 314 V (I presume only nder load) seems to suggest either a lot of resistance in the feed or too little capacitance on the DC bus.

Oh, now this makes no sense at all! Check your rectifiers, I'll bet you have a bad one, and you are only getting half-wave rectification. Hmm, maybe your line voltage is just very low. You'd get 314 V DC from mains at 222 VAC (RMS). I suppose your mains could get that low and still be considered within tolerance. Your VFD obviously was not designed to operate on

208 V 3-phase.

If the problem is truly your mains voltage, a little boost should work, but watch out when the mains go to the high side. Do you have a line reactor on the rectifier? It may have too much inductance and is lowering your voltage. What IS you mains voltage when the problem is observed?

Jon

• posted

Ugh! If it uses an algorithm to detect the ripple frequency in the DC bus, you have a big problem. If it sees 360 Hz ripple, all is OK, if it sees

120 Hz ripple, give phase fault indication.

That would be bad news, and if it is in the microcontroller's firmware, really hard to get around.

But, you ALSO see a large dip in DC voltage, which I find really hard to understand, especially at no load. Given the thing is probably drawing just a few watts to drive the electronics, (and I think most of that was supposed to come from the control supply, powered separately) there should be extremely little ripple and the DC voltage should go all the way up to 1.414 * RMS line voltage. That has to be the key, and until you know why it is well below this, the rest is speculation.

Jon

• posted

Special functions, Phase Failure Fault Enable, p.133

Jim Wilkins

• posted

I did a very extensive investigation of this. I opened this drive and basically took it apart, board by board (while keeping notes as to what is plugged to what).

What I found out is that there is a separate (from DC bus) phase loss detection circuit. At least that is what I think.

The three wires connected to L1, L2, and L3 directly, are connected to a tiny separate three phase rectifier (made of six little diodes), and the output of that rectifier is fed into another board. There is no capacitor at all in this circuit. I cannot think of any purpose of this circuit, other than phase loss detection.

Therefore, simplistically, it would seem that adding a little capacitor to this circuit may stabilize voltage there and fool the board into thinking that all three phases are present.

Any thoughts on this?

i
• posted

Jim, this is very cool, but this function requires a programmer interface for it? (separate programmer module) Is that right?

• posted

I think so. Single phase for one cycle looks like mcdonalds' arches. Three phase is a lot smoother.

Go down to the bottom. Not a lot of area above the curve.

Wes

• posted

I'm not going to study the whole manual for two days and don't know what you have available for tools anyway. I only speed-read it as it scrolled down the screen. There is a 9600 baud serial connection, maybe you can write a program to send command strings or read out the flash memory settings.

I connect to a Radio Shack multimeter in QBasic with OPEN "COM2:600,N,7,2,RS,CS,DS,CD" FOR RANDOM AS 2

600 Baud, No_parity, 7_data_bits, 2_stop_bits, the rest shut off control line timeouts.

SELECT CASE makes a quick parser you can easily add to. I usually check for user input with INKEY\$ and exit with CASE CHR\$(27):EXIT DO. This limits input to single keys but allows ESCape, doesn't require ENTER although it sees it as CHR\$(13), and the program is free to poll the device instead of looping on the keyboard. I haven't figured out how to poll the serial port's I/O memory for data yet and set BREAK so I can exit when the data string doesn't satisfy INPUT.

Whenever I have to figure out some complicated computerized device whose designer and programmer have quit I experiment on something harmless and obvious like toggling an unused or advisory function before risking serious changes. The big first step is getting a response from a ping.

Wes, most meters read Average and scale it to the equivalent RMS value. If they read RMS they'll say so.

Jim Wilkins

• posted

I read it the same way. Others have suggested that the phase fault detection might be keyed on presence of 360 Hz ripple, but even rectified single phase has significant 360 Hz ripple, so I'm not sure that loss of a single phase of three phase would reduce the 360 Hz signal enough to be a reliable way to tell, so the detection algorithm is probably different than simple 360 Hz detection.

Constraints on the algorithm include the following:

The scheme cannot depend too much on exact power frequency.

It must be immune to variations in supply voltage over the nameplate range of input voltages.

The signal being thresholded to declare a fault or no fault must clearly differ if any single phase is lost.

It must be possible to low-pass filter the signal, to eliminate nuisance tripping due to noise.

Ripple voltage as a ratio to DC voltage is simple, and may well be reliable enough.

More reliable but more complex would be the ratio of powers in DC and harmonics.

My question was how S1 and S2 (incoming one-phase line) are connected to L1, L2, L3.

Others have speculated that there might be a blown rectifier. If only one rectifier is gone, changing which terminals are driven from single phase may solve the problem.

You should be achieving almost peak voltage at low 3-phase load.

Joe Gwinn

• posted

Maybe. Many meters don't; other claim they do, but only if the waveform is... where the 'if' is signifiant.

Diode rectification of sine waves is not a difficult case. Switching power supplies, however.....

For a long time, RMS meters measured same by basics -- heating value. They used a resistive load and measured its temperature.

• posted

• posted

I don't know that it's looking at voltage; it may be pulse rate counting. It's hard to tell from afar.

• posted

If this is true, that there's no filtering on the output of this rectifier, you would see 360Hz ripple from it. It's entirely possible that the ripple frequency is monitored. Take a look at the input circuit on the next board. Chances are that there is a voltage divider that brings the voltage down to something like 5 volts. Ig that's the case, maybe you could disconnect this rectifier and replace it with a

360Hz oscillator. That would cost something less than \$0.50. Or, if this is going into the input of a missing pulse detector, maybe you could just fake the output of the detector.

HTH

• posted

• posted

I've always owned Flukes. No idea about the harbor fright stuff.

Wes

-- "Additionally as a security officer, I carry a gun to protect government officials but my life isn't worth protecting at home in their eyes." Dick Anthony Heller

• posted

The issue is whether or not the measured waveform is sinusoidal or something else. The ratio between RMS and avg value of rectified AC is known for sine waves. True RMS for non-sinusoidal waveforms requires either using heating value or circuitry to do the math.

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.