2
\$\begingroup\$

I am running into an issue I don't understand why is happening. I have found how to "fix" it, but I am interested in ideas on what the actual source of the issue is

Short version:

I have a LP gas powered three phase 38 kW Generac generator that needs to sync with a DEYE Hybrid PV system (made up of 3 single phase inverters) designed to use generators to charge battery banks as backup.

When the generator turns on and is running idle, the frequency reading on the generator with no wires or loads, between L1-L2 is 180 Hz or more, frequency reading between L1-N, L2-N,L2-L3, L1-L3, L3-N is always 60 Hz.

Because of this inverters don't sync and generator can't charge batteries.

When I add a big enough resistive load (1500 W) to the generator prior to the inverter, the frequency is regulated (measuring 60 Hz on L1-L2). and inverter syncs. So the load "fixes" the issue

We ran a Fluke 434 energy analyzer and nothing showed up that could explain it, all seemed "fine".

Despite the resistor fix I still want to understand what could cause two different multimeters, A DAWSON DMC220A and a FLUKE 117, and of course an inverter, to read 180 Hz were it should read 60 Hz only on L1-L2 probing... Any ideas?

(Very) Long version:

I have a residential solar PV hybrid three phase system with "Deye" brand inverters with a backup LP power generator in Mexico. Three 8 kW single phase inverters, model SUN-8K-SG01LP1- US/EU, are wired in parallel to form a stand alone three phase grid (220/127 WYE @ 60 Hz) A master inverter controls the other two inverters via RS-485 protocol, per the manufacturer design. These inverters control the charge and discharge of a 36 kWh Pylontech battery bank that is mostly charged by a 7.28 kWp solar PV array on the roof. The system is DC coupled, so batteries are charged by solar pv within the inverter through a controller on the DC side. The loads from the house are driven by the inverters with solar and storage. The three phase load is equally distributed in a distribution block to the three single phase inverters.

These inverters also have a grid/generator input port to allow the system to charge batteries and drive loads through either a utility grid or a back up power generator. (here is where it gets interesting...)

Because the loads are driven by the battery bank and the inverters running only on Solar power, the voltage signal from the inverters is usually the main voltage signal. When the state of charge (SoC) of the battery bank goes below a certain percentage, the inverter sends a signal to the gas-powered generator to turn on. The power generator turns on and creates its own voltage signal, also 220/127 WYE @ 60 Hz. Inside the inverter at one point there are two voltage signals: the one created by the inverter and the batteries and the one from the generator. IF the inverter determines the backup ac signal is within operating range it switches or syncs, like a transfer or UPS switch, to the backup generator driving the loads with the AC from the generator and then charging the battery bank with excess power. Herein lies the problem:

Part a) Because the power generator is only there to help charge the battery bank in case of storm or many cloudy days, it has no other loads, it is switched and "managed" by the inverter system. When it turns on it runs on idle mode, generating a voltage signal until the inverter switches the loads to the generator port. Only then does the generator has a load applied to it

Part b) As stated before the three phase system is composed of three single phase inverters in parallel. The generator is "true" three phase, so three wires run from it + neutral and P.E. to the distribution block. The three wires are then split into three sets of two wires. Inverter A gets L1-L2, inverter B gets L2-L3 and inverter C gets L3-L1 wires to "go back" to a single phase system.

When looking at the system, inverters B and C had a proper voltage and frequency reading from the generator input, so 220 V line to line, 60 Hz line to line and 127 V line to neutral. When I looked at inverter A the frequency reading on the inverter was 0 Hz with a yellow color to signify error. This struck me as quite unusual. So I took out a Dawson DCM220A digital multimeter and checked the frequency reading on the input:

When probing line 1 to line 2, the frequency reading on the multimeter was 180 Hz or more.

Basically my master inverter was sensing a wrong frequency reading and thus determined that the voltage signal was unfit, and never switched (synced) the load and the battery charging to the generator port.

Just to be clear, the same reading, i.e., probing line 1 to line 3 and line 3 to line 2, the frequency readings were in range of 60 Hz. Line 1 to neutral, line 2 to neutral and line 3 to neutral, also had proper 60 Hz readings.

Before finishing the installation of the battery bank and PV system, the house was run by the power generator, so this let me know that the actual frequency was not 180 Hz, otherwise appliances would have been ruined or would not have turned on.

At this point I thought there were some electronic signals messing up the line or something. So I went out and measured the frequency on line 1 to line 2 directly on the generator without any load, any wires, nothing. All the wires were disconnected, even the P.E. wire. The issue persisted. 180 Hz on line to line voltage reading between L1 and L2, 60 Hz on all other lines. and references.

Because I knew the generator could actually power the load from the house, I figured something had to do with idle mode. I took a resistive load, basically a 220 V (line to line) electric heater, and had it wired directly to the generator output before the inverter (one L1 and L2) so the heater was run by the generator itself, not the inverter system.

When I did this, I realized that the load ran properly on only the generator and after a while, since the inverter was still wired to the generator, the batteries were charging from the generator! So I took a reading from L1-L2 Frequency and to my delight, frequency was now 60 Hz.

In short, a big enough resistive load (about 1500 W) corrected the measurement/frequency issue on L1-L2 and thus the inverter deemed the signal proper to sync with it and charge the batteries from the generator.

To re-test the issue, we unplugged the resistive load, and the issue persisted (the inverter didn't sync) when the generator had no load. We retested with a single phase load, so 1500 W on line 1 to neutral, and this ALSO "fixed" the issue.

The thing is that we figured we can't just leave a house electric heater connected directly to the generator with a makeshift outlet, to turn on every time the generator turns on, just to "adjust" the frequency reading. This is an off grid house so it's mainly a holiday home. Not very safe to have a heater turning on by itself with no supervision.

I proceeded to call Generac (the generator manufacturer). They didn't believe me. This is the Mexican subsidiary obviously, and were reluctant to come out and see for themselves. It took some convincing but they finally agreed:

First time around, they came out, unplugged everything from the generator and tried to measure frequency with a FLUKE 373 meter. They had a hard time but eventually "managed" to read frequency and had a 60 Hz reading. I took out my own multimeter (Dawson DCM220A) and showed them this was not correct. My meter was showing 180 Hz and I showed them that the resistive load fixed the connection issue and without it the inverter had a wrong frequency reading and wouldn't sync. They believed me eventually that the issue persisted.

So they came back with a power analyzer, a FLUKE 434. This was today. We took measurements on harmonics and phase displacement and all sorts of energy quality measurements and nothing was showing that could hint as to why with a multimeter probing L1 and L2 we get a 180 Hz reading (and also the inverter has that reading). In other words, to the power analyzer, "nothing was wrong". THD% was around 3%, phase vectors were fine, phase angles were 120º, etc. All measurements were made with all three phases "plugged in" to the analyzer at all times. We tried a WYE measurement, and also took out the N clamp and measured in "DELTA". Nothing changed. I realized later after I left the site that maybe the best reading would have been only on that single phase with the problem. Tough luck :-(

Additionally, today there was a second Generac engineer present with an 117 FLUKE multimeter, one without the clamp and this meter DID show the 180 Hz reading on L1-L2, like mine. So they both concluded that the issue was not my multimeter but it was also "not the generator's fault". It is a "measurement problem".

As I said before, the sync issue between the inverters and the power generator can be "fixed" with a driver load, consisting of a resistor wired directly to the generator output. Generac agreed to provide a generator pre-heater or something similar, left inside the generator housing, so as not have a house heater with a makeshift outlet. So a better solution all around.

That could be that, but for me, there is still the dilemma as to WHY in the world, this is happening. WHAT could be causing three different devices to show a 180 Hz reading only on one of three lines. The issue is only on the L1-L2 reading, not L1-N or L2-N or L1-L3 or L2-L3.

If anybody has any ideas and has read up to this point your help would be greatly appreciated.

Cheers

Single line diagram Connection diagram generator Connection diagram loads

\$\endgroup\$
4
  • 3
    \$\begingroup\$ Did you capture any waveform data? I would want to see that. Synchronous generators naturally create a 3rd harmonic component (some designs like 2/3rd pitch suppress it). \$\endgroup\$ Apr 2, 2022 at 4:10
  • 2
    \$\begingroup\$ Welcome! That’s a wall of text for sure. Can you draw a block diagram of what you have and how it’s connected? \$\endgroup\$
    – winny
    Apr 2, 2022 at 9:07
  • \$\begingroup\$ @relayman357 We did capture waveform data with an energy analizer, but as I said in my (very long) text. We had all three phases connected at once and the analyzer only graphed the Phase to Neutral wave form. I don't have the data, Generac does. I looked at the harmonic readings and they were "around normal" but again, I don't think it was reading the L1-L2 wave. Only the full WYE three phase sytem. "Synchronous generators naturally create a 3rd harmonic component" this is of great help! Makes tons of sense. I will ask for the data and report back \$\endgroup\$ Apr 4, 2022 at 1:38
  • \$\begingroup\$ @winny Thanks! I know, I'm sorry for the long text. I uploaded a single line diagram and a connection diagram for the distribution box. I hope this helps make sense of what we are doing. (Text is in spanish but should make sense notheless) Most words are cognates, except for maybe "Carga" which means "Load". \$\endgroup\$ Apr 4, 2022 at 1:42

1 Answer 1

3
\$\begingroup\$

The waveform on L1-L2 is distorted enough that your frequency meter picks up the 3rd harmonic instead of the fundamental. For whatever reason, loading up that phase drops the 3rd harmonic down enough that the meter detects the fundamental.

Most likely the PV system inverters have a simplistic phase detector that trips on the same 3rd harmonic and refuses to sync. IMHO those inverters suck. They shouldn't be confused by the 3rd harmonic, especially that the fundamental is surely present. I'm not amused at all. This shouldn't be a problem with well-designed inverters.

Why is the waveform so deformed? I'm not sure. It's possible that some of the generator's internal circuitry runs off L1-L2 and has some nasty load that causes this sort of distortion. Perhaps one of the rectifier diodes in that load is shorted. That unbalances the rectifier and the time average of its input current is not zero anymore. That would cause a DC current to flow through the generator windings, making them act quite nonlinear and produce rather distorted waveforms at light loads.

You need to look at the waveforms between all those phases. Make sure you're using an isolated oscilloscope, like a Fluke scopemeter etc.

\$\endgroup\$
3
  • \$\begingroup\$ Thanks a lot! This makes tons of sense. I wonder if the power/energy analyzer doesn't work exactly the same as an isolated Oscillosope, even though it has the "SCOPE" function. As I stated in a comment above, I looked at the harmonics in real time, but with all three phases connected at the same time. Maybe that was the issue. I figure both Multimeters and the inverters all have the same "simplistic phase detector" that can't get past the 3rd harmonic... Do you reckon this is a design issue with generators, or a flaw on a particular one. All the same thanks again! \$\endgroup\$ Apr 4, 2022 at 1:52
  • \$\begingroup\$ I'm not sure to what extent the issue is with the generator, if any at all. I'd say that whatever you're using to measure the frequency is not particularly clever, and the inverter phase detectors are outright stupid. The inverters should have a fairly narrow lock frequency range, say 40-90 Hz at most, and their phase detectors should have a bandpass filter that captures fundamentals in that range, and rejects harmonics. Due to the low frequencies involved, this could be done in digital domain using an Arduino Uno, so the fact they don't do it is absolutely baffling. \$\endgroup\$ Apr 4, 2022 at 5:39
  • \$\begingroup\$ The fundamental frequency is obviously still present between L1-L2, just not enough of it relative to the 3rd harmonic. And it's perhaps true that 3rd harmonic is not present L1-N and L2-N. I bet it's present there, just to a smaller extent, and it doesn't confuse the frequency meters / power analyzers. \$\endgroup\$ Apr 4, 2022 at 5:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.