y Philippines Retirement Solar Off Grid System

[Pete]

Hi Solar, the inverters that have blown up on me, Powerjacks 8kw. Have taken out the mosfets, the driver transistors and also some of the resistors on the Mosfet boards.
I repaired some of the mosfet boards as spares, I ended up replacing the mosfets and some LED indicators on them plus most of the resistors. I did repair one driver board, but they are not easy to work on, being so small.
I just replaced all the driver transistors on it.
My powerjack has blown up twice using my power saw. I think it is the starting current that pulled the battery voltage down too low and caused the mosfets to blow.
I have since added a 500 Farad capacitor bank in parallel with my VRLA batteries and have not had a blow up since. My tools and air compressor start much easier too.
From what I have read Lithium batteries have a lower internal resistance and should be able to supply starting currents easily.
I did read some old stuff on this site about Oztools blowing up powerjacks when the grid failed. I think it was when he was using them as chargers and the grid failed that they blew up.
Have fun
pete

[solarnewbee]

Only wish I could recreate the setup the 10 minutes prior to the blow up to see if there some bleed over. Looks like I’m going back to mechanical ATS. The switch over was so smooth and quiet. I took220v from the inverter to a relay, relay contacts close on 5v to pin4 to pull-up to bring pin 5 high activating the inverter SSRs. Then pin 6 goes low deactivating grid SSRs. Vice versa for switch to grid.

Checked for voltage leak across the SSRs and get just .34v. Of course thing maybe be different when you have voltage coming from both directions. Could be all coincidence and something else has occurred. The top is always off for cooling the transformer so I or someone else could have dropped something in there or the transformer has shorted. In that case I have a spare, nope, it’s 24v Damn!
What’s normal readings for a 48v transformer?

Replaced a popped transistor on the LF board. Could just make out it was a 2A and found one on a 8kw board. It’s now a donor board. I have a drawer full of irf1405 fets. By the data sheet they beefier that the ones in there and gate voltage range is the same.

I have a dandy little smd multimeter I picked up on aliexpress and it’s really nice. All the 47r were blasted leading up to the fet  gates. Everything else is within range.

Put LF board back in and it fired up and acted normally as in tried to run but got no feedback so shutdown driver and whatever other circuit on the board lights up. Flashed a few times like it was hunting driver was solid on. Of course with no fet boards in those first few rows nothing happens.

Powerjack got out of the grid tie and battery charging business. All the tracings are on the boards but the components are missing.

I could order new smd resistors (16) and wait another week or solder in some 1/4 watt. Any thoughts?

Next?

[lighthunter]

Wow youve got a lot in that last post.

1/4 watt R will work fine.
Regarding ATS, ive had plenty of woes with contacts, sacrificed plenty FETs. Now its a large long throw 80Amp dual pole with a phase sync detector and havent lost a FET since. Not sure why youd want SSR type....
The switch sequence you describe sounds like a problem but maybe im misunderstanding. 

"
I took220v from the inverter to a relay, relay contacts close on 5v to pin4 to pull-up to bring pin 5 high activating the inverter SSRs. Then pin 6 goes low deactivating grid SSRs. Vice versa for switch to grid.   "

This to me reads like you are doing make before break with ssrs, unless your two sources are synced and at same voltage, fuses and every other weak point will clear.
I must be misunderstanding.

Your mechanical ATS would have the load on the pole and inverter on one throw while grid on other throw. This allows a brief moment where neither source is connected to load making it impossible for the two sources to fight.  The ssr needs similar scheme where the one in use deenergizes first then short pause before the other energizes.

Sounds like you got the repair going the right direction!

All the best!

[solarnewbee]

Maybe I should post my code. There’s absolutely no delay between relays switching. Of course the goal was to get rid of the long delay that was shutting down TVs, fridge, modem on a daily basis. Son in law works as a call center operator for T-Mobile here  and when the modem drops out he loses calls. Fiber modem takes 3-4 minutes to acquire which is weird. Something else I failed to incorporate was a battery backup for the Arduino that’s powered from the grid. If the grid does something flaky and the Arduino becomes unstable then rapid switching may occur. At the very least the inverter will get hit with a sudden rapid on/off load. Somewhere I have a MeanWell psu din rail mount that has a battery input and supplies 5&12v.




    if  digitalRead(inverteron_pin4 == HIGH);
         digitalWrite inverterssr_pin5 HIGH;   Looking now maybe this line should be second
         digitalWrite meralcossr_pin6 LOW; this line first maybe a delay 50?
     else
         digitalWrite meralcossr_pin6 HIGH

If a triac has a voltage/current on both legs nothing on the gate, will they interact? Is there a device to put in between the 2 sets of SSRs that will divide the 2? Diodes won’t do it. 🤔

There’s a circuit called “zero cross” ties to the gate and one leg of the triac. Can’t find an explanation for what this does but assumption says it prevents what happened to me with the house circuit paralleled to both sets of SSRs. My other thought was that inverter was out of phase with the grid. Never have check to see if we have a mains neutral. I will now tho.

[rossw]

There’s absolutely no delay between relays switching.

Triacs and SCRs will typically stay turned on until their A1/A2 (or A/K) current falls to zero.
So if a triac is triggered at say, 10 degrees, it will stay on until 180 degrees.
If you turn on the OTHER SSR at 12 degrees (after "removing the trigger source) of the first one, you'll have almost a complete half cycle with both devices on.

If a triac has a voltage/current on both legs nothing on the gate, will they interact?

Most triacs can also be triggered with a high dV/dt on A1/A2.

There’s a circuit called “zero cross” ties to the gate and one leg of the triac.

Zero-crossing is a technique used to try to reduce RFI/EMI and overall stress. By triggering the device at "close to" the point the waveform crosses zero,
there is minimum current, so minimum switching noise.

Again, the warning about "trigger at some point, and the device remains "on" until current reaches zero (typically the next zero-crossing)
Triggering on zero, and off at zero means your devices are always on for multiples of half-cycles.
You REALLY need to watch your triggering to keep the magic smoke inside the packages.

[lighthunter]

Hi S/N,   

I agree with Ross post.
I hear your concern about the transfer hiccup. Though i never had a fiber modem to worry about, A loaded refrigerant compressor will make a unfriendly noise with a huge surge if the transfer isnt done correctly.

Doing the math with 50 hz and one full cycle of time is 20 milliseconds. This is the minimum delay that should be used. Theoretically 1/2 cycle would be enough but to be on safe side... 20ms or greater.  This is similar to a double pole relay. Most ATS specs are around 50 ms.

20-50 ms shouldnt bother most equipment.

If you want to make it even smoother add a permission input to your arduino. Then connect it to the contacts of a phase align detector. (Below) Its really crude but works beautifully. The relay drops to deenergized state when there is no danger of misalignment.

The final step i would do is place a low wattage filament light bulb in series with each source (inverter and grid) in this case.
And connect your new modified ssr ATS up to each bulb source and connect a third bulb as the load. Initiate many transfers and watch the bulbs each transition for flickers or brightness changes. If all goes well you wont even loose a light bulb, if you do, at least its not your inverter By doing this you could tune your settings for best performance. Whatever you find as a lowest time delay, i would double it at least to give margin for unknowns like temp change behavior.

I never went that far, i got excellent performance with A JQX-62F 80A relay and the sync circuit so never went further. One more thing to consider when comparing an ssr to a relay...  A transfer switch has a terrible life. It has to bear the abuse of every possible scennario such as a 2HP water pump on startup if the transfer happens during such an event then surge current could easily exceed 100 amps. Physical contacts are better suited for these unknown extremes than an SSR. Even if subjected to this abuse they wont usually fail immediately and if sized appropriately they can be inspected with magnifier and changed before an inverter gets hurt. Just 2 cents. The SSR scheme can work i think ive just never done it and am thinking sizing is critical. Ive friends whove physically dismantled dragon brand ssrs to find components less than half the rated values so be careful of that as well. The failure mode of a mechanical relay is usually welded contacts which prevents switching but doesnt have to ruin inverter. If ANY breakdown of an SSR happens it will almost surely leave debris for the partner device to trip over kind of like an H bridge failure in your inverter, all it takes is one to fail and usually wipes out every one.
Im sure it can be done succesfully just have to design for the worst case.

[lighthunter]

The circuit above is simply a capacitor charge pump driving a rectifier bridge supplying power to a 24v relay coil. (The 250 ohm represents the coil resistance) when the sources are in sync, the relay has no power and drops out giving your transfer circuit the safe to switch signal.

[solarnewbee]

Thanks Ross and LH

There not any high inductive load in the house to note.

The mechanical ATS I was using uses 2 60a rated disconnects motor switched of course. Somewhere I have a contactor rated for 80a (our service is only 60a) but I lack the circuitry to switch it like an ATS . I had always wired the ATS as the inverter being grid, lose the inverter, switch to the grid.

I suppose a simple logic circuit would do the trick for a mechanical ATS. Maybe simply energizing a 2 pole contactor from the 220v out as the inverter comes online

So Ross, your saying  it’s possible that since there was no delay during switching that the triacs were held open without time for a close cycle then grid and inverter were connected until poof?

Just found the contactor I must been using it for transfer before it’s got labels. I’ll install that and the try the light bulb test. Maybe in a few days tho. Spent all days Saturday in bed with body aches. My daughter came back from her Moderno second shot and got fever and body aches and mysteriously I do too the next day. I had the Johnson & Johnson last April with same effect, miserable all night.

Keep you apprised

[rossw]

There not any high inductive load in the house to note.

That probably helps, but even with purely resistive loads the issue I mentioned still exists. (more later).

The mechanical ATS I was using uses 2 60a rated disconnects motor switched of course. Somewhere I have a contactor rated for 80a (our service is only 60a) but I lack the circuitry to switch it like an ATS . I had always wired the ATS as the inverter being grid, lose the inverter, switch to the grid.

I suppose a simple logic circuit would do the trick for a mechanical ATS. Maybe simply energizing a 2 pole contactor from the 220v out as the inverter comes online

When I was on-grid (I had 3-phase power too), I had two, mechanically interlocked 3-phase contactors. It was physically impossible for both to have their contacts engaged at the same time.

The grid 3-phase fed 3-phase to the big loads (mainly a big HVAC). When running on genset (which was single-phase), it would drop the grid contactor, and as soon as practical afterwards (mechanical interlock) would pull in the genset contactor. It was wired so the single phase fed all 3 phases together. The advantage with this was the aircon used phase-to-phase for its control relays, so it simply couldn't operate in this mode, but everything else did

So Ross, your saying  it’s possible that since there was no delay during switching that the triacs were held open without time for a close cycle then grid and inverter were connected until poof?

Hypothetically, lets say you have grid (A) and inverter (B), both waveforms exactly 50Hz, pure sine, identical voltage, free-running inverter with no phase lock.
Using src/triac as switches with simple gate triggering, lets assume for the point of argument that you're running on inverter (input B) and the triac is being (re)triggered constantly. Then, the grid is restored. You try to change from inverter to grid after waiting a few seconds to be sure it's back and stable.
Lets also assume that at the time you go to switch, the grid is 90 (electrical) degrees ahead of the inverter. You turn off the trigger to input-B and immediately trigger input-A with no delay. The B input triac will continue conducting (because it is still passing current), but now triac A will turn on (at the next zero-crossing, which is only 90 degrees away). Triac A will also trigger and you have inverter and mains connected together, but 90 degrees out of phase. (There are worse scenarios - 270 degrees is a much bigger bang)

Spent all days Saturday in bed with body aches. My daughter came back from her Moderno second shot and got fever and body aches and mysteriously I do too the next day. I had the Johnson & Johnson last April with same effect, miserable all night.

Yuck. It's taken me months, and eventually friends in high places to pull some strings, but my wife and I both got our second pfizer shots (wife on tuesday, I got mine yesterday). Been too busy to notice any side effects whatsoever beyond a slightly sore arm at the injection site.

[solarnewbee]

Replaced all the mosfets and resistors and assumed LF and driver board were working properly since it operated stable solo. Switched on the first battery breaker and bang goes one mosfet. Replaced that and tried again then something went pop and mosfets on the positive side blew. Maybe I just need to order an LF board. The LF boards are reduced down to pretty small stuff and I only have so much patience with my bad neck. It was refreshing getting back to a little hot rework and using my favorite electric desoldering gun.

I have an email into Helen at powerjackinverterparts.com so I can send pictures of my boards and buy new ones. Get past this that windmill might make it up!

[Pete]

My experience with Powerjack inverter repairs is that the driver board is often faulty.
It is best to use an oscilliscope to look at the signals that the driver board is outputting.
It does not take too much of a glitch for the mosfet to lose drive and turn into smoke.
I bought a dead 8kw powerjack that I was going to repair and have as a spare, I got it going but it made strange noises when running, like it was trying to blow up then it would recover.
Fortunately it had two transformers in it , so I bought two of the 8010 aliexpress type inverter boards and made two smaller inverters with the transformers.
They draw a lot less idling currents and run much quieter than the powerjack did.
Of course they are not grid connected, as I got rid of our grid and cut the poles down when we moved here
good luck
Pete

[solarnewbee]

Just sent Helen at Powerjack a chuck o money for 8 fet boards, newer version LF board with driver. Since the ac board is labeled ac charger board I may have talked her into sending an LF Baird with charging components.

Missed something on the backside of the LF brd. Not sure when that trace blew but it’s right under the driver.

Pete what components did you replace on the driver when you repaired yours?

I want to test my transformer just for good measure. I know shorts but any ideas what reading I would get on a 48v split coil ?

[Pete]

Hi Solar, when I repaired m driver board, I only replaced the NY, and MY transistors. Oh and the optocouplers.
They were hard enough to get at,
Me and Surface mount stuff don't get along that well that I want to change too many.
On the power board I changed all the mosfets and some of the resistors, I think the ones I changed were 220 ohm but it was a while ago.
Pete

[solarnewbee]

Well, replaced everything but the transformer and now it buzzes 3 times shows 333v on the screen momentarily and goes into alarm. I took reading on the transformer when I had it out and Al but one averaged .7 ohms and 1 was 1/3 ohms. Not knowing what it should be I can’t say if that’s bad readings.

[solarnewbee]

The more I think about it, it seems the xfmr may be shorted. When it was tied to grid thru the SSRs maybe out of phase the windings heated up enough to short out. now it just won’t start against that short. I can’t get answers from Helen about ohms readings for that transformer but what I got does seem really low. Hvac compressors have some heavy copper in them and they read around 2.5-3 ohms and not quite 0 when shorted since my meters usually take a beating in that business and the leads are cheap.  Thank God I sold it and have time to sit around and fret over my broken inverter haha.