120V to 120/240V. Transformer or other ideas?

[gar]

190821-2220 EDT

fastline:

Good.

Borrow a second Honda, and connect both Ns together. Run both generators with no other connection other than the Ns. Measure L1 to L1 with a Simpson 260 in either AC or DC. With the Simpson you can see 1 Hz or lower variations. With a Fluke 27 it is not very distinguishable what is hapening until you get to lower difference frequencies. I used a function generator to generate a sine wave for the Simpson and Fluke tests.

I am guessing that a small autotransformer may provide sync between the two generators when unloaded. You might want to fuse the autotransformer to protect it in the test.

.

 

[gar]

190822-0900 EST

I have suggested two approaches to get 240 V. I will classify these as "parallel" which requires a "full" size autotransformer, and "series" which requires a "half" size autotransformer.

In post #36 I referenced an ARRL (amateur radio organization) discussion. Here they indicated that Honda was not providing much information on parallel operation.

Of my two suggestions the series approach may not work for stability reasons. Effectively some inductance is placed between the two L1s. Almost certainly the parallel method will work with all sorts of loading.

When you place two Honda's in parallel how do the generators know how to share the load. Frequency synchronization can be had from zero crossing information. Is master vs slave automatically determined by which generator starts first? Or is some switch or parameter set? Can more than two generators be paralleled?

In an ideal circuit with two identical voltage sources, and identical internal impedances, power sharing is automatic. But we don't live in an ideal world. Possibly the generators can be built very close, and thus approach ideal.

​​​​​​If there is no communication between the two generators, other than terminal voltage, then I suspect there is unbalanced loading of the two generators. With an electronic inverter there will be built in current limiting (power). Current limiting means the internal impedance changes (increases) at some current level. Ultimately you are power limited by the engine, and you don't want to stall the engine. Once one generator reaches its maximum power, then the other will pickup load until it reaches its full power. Then voltage might drop with added load. Hopefully when voltage drops to possibly 105 V everything shuts down. It is possible shutdown occurs at full power rather than some voltage.

fastline experiments you can perform will tell us much more, and possibly some of this information is in the manual.

.

 

[fastline]

No, inverter generators will kick out before they stall due to input parameters to the inverter getting out of range.

You can chain many of these in parallel. The main parameter that dictates load share between generators is frequency. However, synchronization between them is very good so load matching is actually extremely close. However, for this reason, using generators that are not the same size is not ideal but I bet inrush is much more stable!

One thing I need to do so put two together and connect each to a channel of the Oscope so I can see how close they really ride.

 

[ggunn]

I would think you'd connect one genny's L1 to the other genny's L2, and use that point as the neutral.

He is talking about two 120V generators; both have only L1 and neutral. I believe he is talking about connecting one's L1 to the other's neutral and making the connection point the neutral for both and the neutral of the second genny L2. It might work but it sounds a little screwy to me; I don't know how you would get them to stay synched up.

 

[synchro]

My understanding is that gar is proposing to have the neutral of both 120V inverters connected together. Then the output of one would be L1, and the output of the other would be L2. The inverters would have to be synchronized 180 degrees apart so that there is 240V RMS between their two outputs. The autotransformer would provide the synchronization. Conceptually you could "stack" the inverters in series with each having the same phasing and with the 120V output of one tied to the neutral terminal of the other. However, this seems unnecessary and without knowing the internal design of the inverter generator it might cause problems when the neutral terminal and the generator housing have a large voltage difference (assuming both housings/equipment grounds are tied together, which they should be).

 

[gar]

190822-1627 EDT

Started this comment earlier. Now back to it after several others have posted.

If there is no communication channel separate from the two power wires, then there is no meaning to L1 and N and some other independent generator's L1 and N. These two independent generators don't know anything about each other. That is what independent means. Simply connecting both neutrals together still will not affect the separate voltages, waveshapes, frequencies, and phase relationships.

N (neutral) means nothing in a two wire system. It is just a useful label. It may have some relationship to the generator chassis, but we need to assume, as is true of most generators, that the generator output can be floated from chassis, and EGC.

We can virtually never expect the two generator frequencies to be within 1/1,000,000 of each other. This would be broadcast quality crystals. This kind of difference would take 1,000,000/60 = 17,000 seconds to precess 1 cycle. With less accuracy precession is quicker.

Digitally one generator can sync with the other within a cycle by looking at the zero crossing of the master, Who is the master. This could be the first to get to a positive zero crossing. And this could be determined by the second inverter before it actually comes on line. Thus, in this case, sync occurs as soon as the second generator starts, This is all conjecture because I don't know what is done.

Once started the faster waveform could become the frequency control.

This does not give us a way to control current or power distribution between the two sources.

I would like to know the exact frequency of two different generators.

fastline I hope you can start to run some experiments. An accurate frequency measurement with no load would be real useful. Then how does frequency, terminal voltage, and current vary with a changing resistive load from 0 to maximum. I can conceive that frequency won't change.

.

 

[gar]

190822-2301 EDT

Took a look at a commodity quartz crystal. For $1 in quantity of 1 initial frequency tolerance is +/- 15/1,000,000.

.

 

[gar]

190824-1617 EDT

To continue on frequency from last night.

I have a moderate stock of crystal oscillators of various frequencies. I pulled one from stock, no sorting. This is a Crystek CCO-010 5.0688 MHz. Possibly 30 to 40 years old, never used. Five volt device, square wave output. No longer made. I would classify as a commodity product, not high accuracy, and poorest temperature coefficient.

What is unique about 5.0688 MHz. Dividing by exactly 4224 this results in 1200 Hz, 1200 baud. Divide 5,068,800 by 46,080 and the result is exactly 110 baud. And divide 5,068,800 by 44 and the result is exactly 115.2 kbaud.

I have two frequency meters. One read 5.06876 and the other 5.0688 on this particular crystal oscillator. Made no other comparisons. An error of 1 part in 50,688 is 20 parts in 1,000,000.

Is a commodity power inverter that is the frequency standard for a generator application going to be based on a quartz crystal oscillator? Could be the microprocessor in the system is quartz crystal based, and thus the generated 60 Hz could use this same time base for its frequency reference. In turn the inverter instantaneous frequency could be much better than power line. But on average over a year would not be.

To save a few pennies in manufacture of the inverter generator the time base could be a non-quartz based device.

I am curious how good the Honda frequency is?

.

 

[LarryFine]

Some linkable generators have a separate sync cable.

 

[gar]

190824-2206 EDT

My frequency meter has now warmed up for several hours and two different digital chips that wern't working started to work. Now I can resolve 10 Hz on the 9.999,999 (1 second sample time) range. Looking at the said previous crystal oscillator I get readings of 5.06880 or 5.06881 from this frequency meter. Thus, the two oscillators, time base in frequency meter, and the oscillator under test, are within 2 parts per 1,000,000 of each other. Since these are uncorrelated frequency references of different manufacture, time, and power sources it is very probale that they are very close to a true 5,068,800.000 Hz source.

To easily make accurate 60 Hz frequency measurements you do not count cycles, but rather measure period averaged over a number of cycles, and then calculate frequency from the average period.

Tonight my line frequency is very close to 60.0 Hz. I am using frequency measurement with a 10 second measurement time that provides resolution to 0.1 Hz. Over a 10 minute period all readings were 60.0 except two that were 60.1 . This is unusual, but I have not looked at frequency for several years.

.

 

[LarryFine]

Over a 10 minute period all readings were 60.0 except two that were 60.1 .

We call those leap-seconds.

 

[gar]

190828-1052 EDT

fastline:

Do you have test equipment that you can run frequency, voltage, and current measurements on your Honda? If not is there someone near you that can help?

.

 

[gar]

190830-1523 EDT

It has been bugging me on how Honda balances load when paralleling inverter generators. From the following two references I conclude that you don't just parallel them. It seems a special parallel cord is required and only certain models can be paralleled. And only two can be paralleled.

This probably means that the special cable contains some communication signals.

Otherwise balancing of load would have to be done by internal impedance, and/or frequency adjustment with load. Neither of these seem very viable. One other possibility would be to let one go to its full load and then current limit. Then what happens when both hit their full loads?

http://cdn.powerequipment.honda.com/pe/pdf/misc/EU2200i-EU2000i-Parallel-Operation.pdf
https://powerequipment.honda.com/generators/generator-parallel-capability

.

 

[gar]

190913-1740 EDT

fastline:

What have you learned about the characteristic of Honda inverter generators? And what must be done, if anything, to parallel them?

In some of the information I found, it may be implied that a special paralleling cable is required.

Honda does not seem to be very clear on what has to be done to parallel generators, or any restrictions.

.

 

[GoldDigger]

From what I understand, the parallel setup involves simply using an appropriate non-suicide cable to parallel L and N and then start one of them first. But that is not 100% certain.

 

[gar]

190913-1941 EDT

GoldDigger:

What balances the load? Or isn't it balanced, and one gets overloaded first?

Who controls frequency? First started? Or the higher frequency? Or load changes frequency?

Who of the two or what controls voltage?

Can more than two be paralleled?

.

 

[synchro]

These patents assigned to Honda describe inverter generators that can be paralleled without additional control wiring between them:

https://patentimages.storage.googleapis.com/c0/e6/2e/6de65a3fa324b7/US5258700.pdf

https://patentimages.storage.googleapis.com/78/a3/6c/0ef253a85e0f6b/US5400236.pdf


The first patent provides info on how a second generator synchronizes with a first one that's already running. The second unit synchronizes by adjusting the phase of the inverter output so that its measured output current is in-phase with the output voltage of the two paralleled generators (achieving a unity power factor). It uses a phase locked loop to control a VCXO (voltage controlled crystal oscillator) which is divided down to generate a sequence of binary words that drive a resistive ladder and create analog samples of a sine wave. This approximated sine wave and a generated sawtooth wave are both input to a comparator in order to created the PWM drive signal for the inverter.
The patent also describes the sequence of events after starting a generator, including switching to what I would call an autonomous mode if the inverter output is not already being driven by another generator.

The second patent describes additional functions for preventing overvoltage on the DC bus and providing current limiting on the inverter output.

 

[GoldDigger]

Sounds like there is no special provision for load sharing, but as the load increases eventually, whatever the split ratio, one generator will reach maximum current and the other will take up any remaining load up to its maximum current. At that point, I would assume either the voltage will sag or both generators will shut down.

 

[synchro]

Sounds like there is no special provision for load sharing, but as the load increases eventually, whatever the split ratio, one generator will reach maximum current and the other will take up any remaining load up to its maximum current. At that point, I would assume either the voltage will sag or both generators will shut down.

Agree, and there 's not much description of the load sharing. Figure 12 of the first patent shows the V-I curves of two generators and how they each supply a certain current at a given output voltage. The picture shows the two generator outputs havIng the same impedance (slope V/I) but displaced so that one provides more current than the other. How fast each alternator is running would determine how much DC bus voltage and inverter output current each one could support. The generator's have controls on the speed of the engine and alternator but I didn't see that described anywhere (although I haven't gone through every part of the patents in great detail).

 

[gar]

190914-1015 EDT

GoldDigger:

I tend to agree. I read parts of both referenced patents. In detail parts of the first, and less of the second.

These patents seem to cover only the operation as a slave unit. In other words what the generator does when it is the slave. Thus, something else sets frequency, and this slave tracks that frequency. There is no discussion of it being the master. When the generator of this patent starts up it assumes it is the slave, and it servos off of an existing voltage. From a patent perspective there is no need for a discussion of how a generator is to decide if it is a master, and set itself as the master. The patent only needs to cover the slave aspect.

We need some experiments on a number of Honda generators, all of the same paralleling model number.

Run as a single generator what is the voltage current curve of each unit plotted vs current, and on the same plot frequency vs current for resistive, highly inductive, and highly capacitive loads. Another test needs to be a very nonlinear load, a capacitor input DC supply.

How close do these curves match for different units of the exact same model. Clearly in the above tests these would start in a master mode, and there is no slave.

These tests would be enough with which to start. From this information one could then judge what paralleling experiments to run.

.