? about parallel operation of inverters

[brutus1955]

what i want to try to do, is to take two ups
inverters, tap the first at the logic drive to the FET's
driver, and jumper this logic over to a second inverter
which has had its logic disconnected.

the goal is to be able to then parallel connect the outputs of both inverters for more power

is this possible?

bob g

 

[itsmoked]

is this possible?

Not even remotely...

Keith Cress
kcress -

http://www.flaminsystems.com

 

[brutus1955]

ok, why?

is the logic not able to provide enough or sink enough current to provide for additional fet drivers?

if this is the case, what is keeping me from inserting a
intermediate driver, that takes the logic input and has enough capacity to provide for the fet driver darlington pairs?

or alternatively, maybe just change out the darlington pairs with higher power units and let them drive the additional fet's on the other inverter board?

i guess i had hoped for a bit more of an explanation
than a simple "no"

so if this method will not work, what would it take to
get both inverter boards in sync with one another, so that their output transformer secondaries could be connected in parallel?

thanks
bob g

bob g

 

[itsmoked]

These are not just open loop signals. They are closed loop. That means the driver needs to see what's happening as it creates the control signal. Often they control the actual cycle profiles to specific levels. This means the control signals are looking for specific signal shapes and amplitudes. These expected values are going to be badly distorted, confused, and wrong if you attempt to kludge two output stages together.

You also can't just hook two independent output stages together without investigating how they will share the load, and if they will share it equitably.

Further these things are engineered to great detail to just barely run correctly without the drive circuitry burning up from the loading it's subjected to. Now you hook two full drive loads to one driver and you're going to be overloading the driver -badly-. This will cause it to drive the output FETs badly -read slowly- which will cause their losses to skyrocket. High losses lead rapidly to letting all the smoke out.

Just buy a bigger unit. It will save you a fortune if you value your time at more than 10 cents and hour.



Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

There are inverters that can be operated in parallel. Siemens 6SExx types can be parallelled if an optical link is connected between master and slave(s). The link synchronizes gating pulses.

It works in a mode similar to what brutus proposes. But doing it "just like that" would not work. The reason is that the EMI level in an inverter is quite high ("very" is a better description) and that parallel operation needs exact switching to be succesful. The need for exact switching precludes the filtering that is necessary to make the system operable.

The closed loop that Smoked mentions is there to ensure that every IGBT reacts properly to the gating pulses and that they do not have too high a residual Vsd (which could happen if there is a short on the load side). If the master gating is applied before that logic, the protection might work. The problem is that, even if it does work, the signals exist in the slave inverter's control processor and thus doesn't have any influence on the master pulses that actually control the inverter.

Parallelling of inverters surely can be done. See

http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19640007846_1964007846.pdf

for an early test. But it is not, as Smoked says, anything you should try at home.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ScottyUK]

What are you trying to achieve: redundancy or just to wring more power out of the drive?

If the former then it's realistically a non-starter, and for the latter wouldn't it be easier to install a single larger power module? The chances are you can drive a slightly larger power switch using the same driver circuitry within reasonable limits, although don't expect to take the control stage from a 1kW drive and apply it to a 100kW power stage because gate capacitance will overwhelm the driver and the maximum frequency big switches can achieve is much lower than those small switches are capable of.

With the cost of drives through regular suppliers and even on Ebay is there any likely cost benefit of what you're doing?


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[raithrovers1]

As mentioned before, most modern UPS systems can be paralleled up but only with systems of the same design and manufacturer. You can also get systems that can have additional inverter modules inserted inside the unit to add power rather than redundancy.
I agree with all the other posts as to why what you are suggesting is not feesible.

UPS engineer

http://www.powerups.co.uk

 

[OperaHouse]

I could do it. It would be more trouble than it is worth. If you have to ask this question and don't understand the potential problems, it means you can't. A UPS is a nice source of parts for a project. Other than that they are useless. You can by a late model dead 2KW inverter on ebay for $20 and fix it in an hour.

 

[brutus1955]

maybe i should have put a bit more info in to start with

the ups unit i am working with are common as dirt, scrapped when the batteries die usually, and can be had for less than 10 bucks each.

now then, with cheap inverters available, that are sine wave
it just seems a crime not to give it a go to try to stack them or use them in ways not originally planned by the oem.

what am i out if i fail? some time and a few bucks, and even that is not a big thing because i can use the transformers for another project.

as for taking the logic off board to control two inverter boards, i figured it possible (now i realize there are all sorts of reasons why i probably can't work) but,

i have seen it done in some inverters such as the 3600 watt vanner inverters,

if you open one of their cases, you will find one logic board that houses the microprocessor and all its related crap to generate the logic needed to form a sine wave

from that logic board the pulse are moved via a pair of equal lenth ribbon cables (same stuff as what is used to connect harddrives, floppy drives, etc in computers) to
remotely mounted FET driver/FET switcher boards. each of which is connected to its own transformer on the primary side, and then parallel connect on their secondary side.

it would seem that it is possible to do this only if the logic side was used, and not jumpering from the FET driver
to two seperate FET swithes where the power levels are much higher and emi would therefore be an issue.

at least the emi would be much more of an issue anywhere around those switches and drivers than it would be at the relatively low power levels of the logic drive.

also it is fair to note this is a project to see if it can be done, and what can be learned in the process.

so maybe there is less than a 1 in 10 chance of success?

or even 1 in a 100? if successful there is practical use for such a system, by integrateing some microcontrol to bring in and out stages depending on loading the overall system efficiency can be much higher than if one were to simply buy a larger unit and run it at low power levels most of the time.

anyway thanks for the input guys, even if you are not optimistic about the project

bob g

 

[brutus1955]

i might also add

i won't be running in open loop
the idea being maintain the closed loop of the first inverter board to the logic as oem designed

then as the second board is brought in parallel on its output the feed back is maintained

otherwise there would be no feedback and certainly there would be problems.

the plan is to learn if this is possible, then take those
lessons learned to stack two or maybe three, 3kva inverters,

the 3kva ups inverters are much more expensive, so i would rather torch a pile of much cheaper and available 1.4kva units working the kinks out.

 

[Skogsgurra]

Good luck! It may even work.

Sometimes, ignorance helps. Guys that don't know about the perils of the journey can complete what others wouldn't even think of (remotely).

But don't come here for help if it doesn't work - we already told you.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[brutus1955]

so the odds are something greater than 10 to 1 against?

not to worry, if it don't work, i won't be coming back to find out why.

however if it does work, does anyone want to hear about it?

if not, thanks for the input anyway

its appreciated



bob g

 

[ScottyUK]

Bob,

Let us know either way.

When you first energise it have a video camera rolling: that way if it's a really big bang you can put it on YouTube!


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[raithrovers1]

bob

I admire your adventurous spirit and would be very happy to hear if your project is a success. I am mainly worried about the current sensing side of the feedback loop. This will only be monitored on the output of the inverter that you will be taking the pulses from. This means that in theory, the other inverter could supply more curent that it can handle without any limitation. therefore if there is a parallel problem and this unlimited inverter feeds current into the other inverter (if out of phase for example)then all I can see is big bang and lots of smoke. Once again, have video camera rolling for you tube!!!!
Please let us know how things go.

UPS engineer

http://www.powerups.co.uk

 

[OperaHouse]

On that note, each individual fet in the H bridge feeds back to each individual power supply through an internal diode. If one pair of fets shorts, the other inverter power supplies will supply current through a common ground without it being sensed.

 

[Skogsgurra]

Bob!

Don't let them stop you! I am getting seriously interested in what you will arrive at. Carry on and KUTGW!

Also: Be careful! Exploding power transistors and electrolytic capacitors are not nice to get into one's face. Wear protective goggles! At least.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[brutus1955]

not knowing what KUTGW means i guess it is full steam ahead

i picked the inverter with the failed fan circuit to do the chopjob on, and lifted the main board out of the case

the solder side is the exposed side, and the component side is down in the case, so in case of explosion it is well contained.

the topology is as follows as near as i can tell

working from the output back

there is an output transformer that puts out 120ac on the secondaries and input 13.5 vac no load to ~15.8 vac at 3/4 load, so i am not dealing with terribly lethal voltages here at least on the drive side.

driving the primary side are two banks of FETs each having a parallel set

each set is driven by a pn2222a and a 2907 transistor
which as near as i can make out so far make up what in effect is a darlington pair

this pair is driven directly from the logic off the processor ic.

what i have in mind on the slave board is disconnecting the logic to the darlington pairs and jumpering the logic from the master boards logic

that way one logic controls two seperate sets of darlingtons which in turn drive two seperate sets of FET's
which in turn drive two seperate transformers

now there is no feedback from the second transformer output so i will have to parallel connect the two transformer secondaries in order for the master to be able to monitor the combined output and make correction if the voltage goes high or low.

at this point it looks fairly straight forward, but
i am thinking i should make up a tree type jumper set so that all the leads to both sets of darlingtons are equal length and therefore resistance.

it would appear to me that what i am contemplating doing is just doubling the same archetecture that is already on one board, they use a darlington pair to drive each bank of fet's, i am just adding more banks of both darlingtons and FET's.

i did find the current transformer and its output goes directly to the microprocessor via a couple of resistors, it might be that i can connect the slave CT back to the micro in tandem with the master CT? gotta think about that a bit more. i really don't think i will need to do that as long as the outputs are connected in parallel and not run seperately where one would have feedback and the other would be running open loop.

just got to take a bit more time to sketch up the connections from the board, double check and as the say in carpentry, "measure twice and cut once"

i was somewhat relieved to see that there is no DC/DC converter in the front section that boosted the 24vdc buss
to something over 150volts or somesuch.

working with a driver that is under 24vdc albeit at fairly high amperage is much safer poking around in

at least for me

trying to take my time, be patient and think it through well is the challenge at this point.

i just wanna do it and see what happens

thanks guys

bob g

 

[Skogsgurra]

Keep Up The Good Work (KUTGW)

The 2N2222 and 2N2907 are not a darlington pair. They are a complementary pair with 2N2907 being a PNP transistor and 2N2222 an NPN. Needed to drive the gate capacitance on/off.

Don't forget the camera! And remember the goggles! Insulated gloves is a good idea, too.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[brutus1955]

Gunnar:

boy you sure are something less than optimistic huh?



thats ok, i frequently work around 440/440 and am aware
of getting bit, and have had caps explode on me as well,
thats always good for getting the heart rate up as well.

one question though
the two transistors 2222 and 2907
are they not situated to make up the equivalent of a darlington pair. in that one takes the lower level
and raises it while the second handles a bit more power.

its been about 30 years since i was in electronics tech school, so i am working to relearn as i go.

anyway looks like it might be exciting huh?

placing any odds on success?



bob g

 

[Skogsgurra]

Darlingtons always come in NPN or PNP. Never both in one pair. What you see is probably a level shift plus a driver. Difficult to say when no diagram available.

I AM very optimistic. But also somewhat careful. 10:1 seems to be a fair bet. Or 100:1.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...