Author Topic: No brand/5500W inverter disaster porn  (Read 7843 times)

0 Members and 1 Guest are viewing this topic.

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #30 on: May 05, 2021, 09:04:12 pm »
Thanks LH, I will definitely ask if I can't understand the Chinglish.
Yep I agree on social media, all it seems to have done is speed people up more, so that their interactions are very short and not so sweet.
We don't do any of it at all.
Thanks for your help
Pete

Offline dochubert

  • Sr. Member
  • ****
  • Posts: 265
  • Karma: +4/-0
  • 11 powerjacks, 9 that run, 3 actually used daily
Re: No brand/5500W inverter disaster porn
« Reply #31 on: May 05, 2021, 09:27:52 pm »
Hi Guys,
Here's some info you might not have;

8988-0
8990-1
8992-2
We're all going to DIE!  (eventually)

USA = Communist former republic
---  dochubert --

Offline dochubert

  • Sr. Member
  • ****
  • Posts: 265
  • Karma: +4/-0
  • 11 powerjacks, 9 that run, 3 actually used daily
Re: No brand/5500W inverter disaster porn
« Reply #32 on: May 06, 2021, 10:14:38 am »

Should have mentioned that the info is for the 10kw unit that I have.  The 5500w may be different.
We're all going to DIE!  (eventually)

USA = Communist former republic
---  dochubert --

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #33 on: May 06, 2021, 05:31:48 pm »
Thanks Doc, I did notice that the one you have is a little different to the one that is in the article but still good info to know.
I have saved the pics so that I can check when my inverter arrives.
Cheers
Hope the new system at your place is behaving well
Pete

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #34 on: May 07, 2021, 01:36:54 am »
The good news is that after I replaced the MY and NY transistors and Optocouplers on the driver board that the inverter appears to be working fine.
Tomorrow I will put the capacitors back on the main Mosfet board and see how it goes then.
I have only run it with a 0.22 ohm resistor in series with the positive supply so far and it puts out 240 volts and doesn't scream at me.
So the signs are all aligned so far.
Thankyou all for your help, the ideas shared here saved me blowing up mosfets and a lot of work.

Pete

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #35 on: May 07, 2021, 02:30:09 am »
Good work Pete! sounds like your just about complete with repair. One thing that helped me out in past was to shut off the lights during testing to see if any tiny lights are flashing as carbon arcs around fets sometimes its hard to see all the carbon from the burned fets. Thanks for the photos on connections Doc! looks like i didnt need the middle one so only the switch and feedback are absolutely necessary. I really like this thing so far. Maybe its less prone to failure than pj!

I did a bit of reading up on bipolar vs unipolar sine wave synthesis. Its more complicated than it first seemed.  Turns out bipolar switching scheme uses rail to rail power supply voltage in both polarities to drive the transformer primary. (Max vout=.78 x vbat) which is 34.32vac when batt is 44vdc. And uses high frequency switching on all 4 fet gate drives.

This method is simpler and produces maximum voltage possible at the transformer primary but has drawbacks.

Switching losses are higher.
Significantly more distortion.
Harder on fets.
Needs larger inductor to smooth to good sine wave.

Two types of unipolar switching schemes exist, one uses high frequency switching on all 4 fet drives, the other uses low frequency for two and high frequency for the other two (8010 unipolar does this)

This type is significantly more efficient.
More accurate sine wave output.
Smaller inductor filter is needed.
Maximum output voltage is less than power supply.

   I couldnt find proof of lower output in writing but it seems obvious, as the wave simulation uses a different volt averaging scheme to minimize stress on the switches  and load. (the reason the transformer is quieter with 8010 unipolar)

The 8010 has capability to do bipolar so i may switch it as an experiment when i have a backup ready to use :)  switching 8010 to bipolar will involve altering the feedback loop to include freqsel pin on 8010 and making sure the other two drivers are capable of the high frequency gate switching. I'm pretty sure the egs002 is not capable of the 8010 bipolar mode simply by shorting the jumper.

In conclusion unipolar has some serious advantages. I do think  the highest reliability inverter is to add (or subtract) the necessary turns to transformer and go with unipolar.  *note, if reducing primary turns, you are increasing flux density and you dont want saturation, safer to add secondary turns)  Factory wound units generally have primary low volt side on inside layer so theres no risk of me going there HA! ::)


 
Health Warning: May contain traces of nut!
LH

Offline noneyabussiness

  • Full Member
  • ***
  • Posts: 155
  • Karma: +7/-0
  • I think it works....
Re: No brand/5500W inverter disaster porn
« Reply #36 on: May 07, 2021, 05:46:23 pm »
https://www.anotherpower.com/board/index.php?topic=1116.15

Oh and make sure you make the modification above that OZ referred to on first post, VERY important for no blown  mosfets on low frequency inverters...


Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #37 on: May 07, 2021, 09:00:25 pm »
Yes, good point!! i did do that just in case it mattered, the people who designed these boards may have their own clever way of fixing that issue but i didnt wanna find out the hard way they didnt. Its kinda neat, the 8010 has a bit of protection built in like under volt and overvolt of the AC feedback overtemp as well, then the egs002 board has protections built in and the makers of this power board added overcurrent and short circuit over and under battery volts. I wonder if that isnt how the problem Oz found came about, the old saying, too many cooks... well you get it!!

The thing just works way too good not to like it or want to change it. Best part if something did happen the plug in egs002 board is only about $7.00 and beyond that. repair would be just the normal resistors and fets we are used to.
Health Warning: May contain traces of nut!
LH

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #38 on: May 08, 2021, 08:53:49 pm »
Hi guys, In my spare time, Ive been trying to understand the switching sequence to synthesize a pure sine wave from a digital pulse train. The whole thing used to seem simple to me until i started looking at it closer.

The two most popular techniques of generating a sine wave are called bipolar and unipolar. The meaning of bipolar implies that you are supplying energy in both polarities throughout the positive and negative half cycles.  Unipolar on the other hand implies that you are supplying energy of only one polarity for each half cycle of the AC wave.  Just knowing these facts raises an eyebrow as to whats going on with the fets, inductor and transformer.

Theres always lots of inductance in these circuits and one principle of an inductor is it always opposes a change in current. Like a flywheel so to speak it has electrical inertia. So for whatever reason, maybe simplicity, bipolar switching technique pulses energy through the transformer primary in both directions as it makes a half cycle.There are only two possible switch conditions yielding v+ or v- at the primary. It would be a bit like driving down the road in forward gear for a short distance then reversing directions for an even shorter distance then forward again further than the first time. With very short pulses of timed energy the wave is shaped properly. One concept i never realized is except for dead time between switch events, the load (xfmr pri) is always connected to the power source either positively or negatively. Theres no "off" time with bipolar. The reverse polarity has a braking effect on the current flow through the inductor. By controlling the duty cycle any desired resultant current flow can be achieved. Just the concept of it doesnt seem nice to the switches or anything else in the way.

Unipolar switching also pulses energy to shape the half cycle but it doesnt use reverse gear until the alternate half cycle where it is needed. Between the energy pulses (wait time) both fets on top or bottom are turned on to allow inductor current to continue to flow.

Im doing some interpreting and reading between the lines here but im pretty sure the reason bipolar switching always has a pair of fets ON so the load has a conductive path is because of what happens if you try to stop a flywheel in motion or inductor with current. (Spark plug ignition uses that principle)
So unipolar is described as a 3 voltage level +Vd -Vd and 0, basically the 0 volt state is when both upper fets are on (or both lower) when the inductive energy can bypass the power source but still continue moving.  If you ask me,  unipolar just makes a lot more sense and seems much easier on switches.  I still do not understand fully why the voltage output of unipolar should be less but i expect theres a good reason for it. Some articles i read were very clear the unipolar switch technique is very superior to bipolar  with much less distortion and very low harmonics and much smaller inductor needed and more efficient. I guess its easy to see why.
Health Warning: May contain traces of nut!
LH

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #39 on: May 08, 2021, 09:14:27 pm »
Nice explanation LH. Like you say an inductor resists current direction changes. It is a challenge when trying to switch the inductor current in reverse as the inductor stores energy in the core and the back emf can be pretty horrific.
The designers must have some way of dumping that back emf safely or the mosfets would just go bang.
I am guessing the mosfets are turned off just in time to soak up the inductors stored energy before reversing the output.
Beyond me how to design that, I just fix things when they break. As long as they once worked then fixing them is possible with patience.
Nice thread some great info here
thanks
Pete

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #40 on: May 09, 2021, 05:08:43 am »
" It is a challenge when trying to switch the inductor current in reverse as the inductor stores energy in the core and the back emf can be pretty horrific."

Yes indeed, once the FETs are switched on, the inductor is now connected opposing the source voltage and the stress is  between the inductor and the source, so the current is braked and excess inductor power charges battery or capacitor bank rather.  Thus one of the reasons for the much larger low esr capacitor bank in the bipolar inverter.  Its not horribly inefficient because energy is not wasted but theres a lot of whiplash creating distortion that the filters need to remove. One of the reasons Oztules inductor modification helped efficiency. Id like to assume the added inductor increased the life of the FETs as well. The working inverter in some ways is similar to a combustion engine where timing and counterbalance are extremely important to the smoothness of the cycle and longevity of the machine.
Health Warning: May contain traces of nut!
LH

Offline rossw

  • Senior Moderator
  • Hero Member
  • *******
  • Posts: 879
  • Karma: +35/-0
  • Grumpy-old-Unix-Admin
Re: No brand/5500W inverter disaster porn
« Reply #41 on: May 09, 2021, 05:29:45 am »
Like you say an inductor resists current direction changes.

Clarification. It's not changes in DIRECTION.  It's simply changes.
If you try to go from 3A to 5A, (no change in polarity) an inductor will resist it.
If you try to go from 20A to 15A, an inductor will still try to resist it!

It's the change (variation) in magnetic field that is the effect at play, not a complete reversal.

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #42 on: May 10, 2021, 10:58:54 pm »
Hi rossw! Thanks for pointing that out. I'm still puzzled as to why the 8010 unipolar has less voltage building ability, almost seems like it must be a software limitation since its the same hardware circuit. Yet unipolar is described as switching between +V and 0 wheras the other is +V and -V perhaps thats the difference.

Ive also noticed one other major benefit to this unipolar inverter. We have a 3.5kw grid tie connected to two arrays, most days that thing is pegged out at 3650W, its programmed to skip one AC cycle every second and monitor the grid its connected to. This (skipped cycle) would always cause the lights to flicker every second. Now with the unipolar there is no flicker. Also when the wife would use a space heater on low, ( uses 1 diode for half wave) the grid tie would go offline. Now with this 8010 unipolar, the heater can be on low using 750 w of lopsided load and the GTI is fine with it. So i guess the wave accuracy and low distortion claim is true.
Health Warning: May contain traces of nut!
LH

Offline rossw

  • Senior Moderator
  • Hero Member
  • *******
  • Posts: 879
  • Karma: +35/-0
  • Grumpy-old-Unix-Admin
Re: No brand/5500W inverter disaster porn
« Reply #43 on: May 11, 2021, 10:55:36 pm »

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #44 on: May 23, 2021, 08:29:07 am »
Yes, good article! not new information compared to others i had read but very informative nonetheless. I was hoping to find one that states unipolar has max output voltage .64VDC vs bipolar maxvout of .73vd.  Instead of stating the relationship between vd and vout vs switching type, they do infer the concept by pointing out the unipolar driven Hbridge drives the load between VD and 0 while the bipolar Hbridge drives the load between +Vd and -Vd. I suppose most who read such articles dont need it spelled out any further :) Ha! Eventually i will test and prove the theory but havent as yet.

I must clarify something i said earlier though. I mentioned something to the effect of unipolar inverter is more reliable and less prone to failure than bipolar counterpart. While i still think this is true, I havent proved it, its only a guess based on the switching emf. It turns out, the reason my bipolar pwrjck bit the dust a while back was because of a failing transfer relay. Not the inverters fault. If we get two consecutive cloudy days i have to switch over and though i synchronize the two sources before a switch. I have on occasion pushed stop button on controller which transfers all without regard to phase alignment.  Thats a fatal error, it wounded a relay and later caused the failure. I am thinking instead of transfer scheme. I can get an even better seamless solution by grid powered charger to 48v battery at about 1kw set to come on at some min battery V. I have an old transformer and rectifier with caps from a plasma supply that will work perfect. Not quite as efficient but zero risk of inverter failure from a relay. How do you guys do this? I know those with gensets just charge the battery but wondering if anyone else uses transfer relays and have inverter failure because of it. Seems like a cool idea but i cant recommend it. If someone insists on it i would say make sure your relay has 1/8" between poles and never switch without synchronizing.



Health Warning: May contain traces of nut!
LH