A quick update.

 

Our first full month (billing period is 18th-18th) with the monitoring on made 121 units including a couple of miserable days where we made less than 1 unit. We never actually went into export (although it came close), going to need more panels.

 

So around 480 Baht against our regular usage of 5,700 Baht, about a 10% reduction.

 

Payback would be about 4.3 years at the current rate, we shall see what happens when the dry starts (no clouds but smog).

Nice!


Sent from my iPhone using Thaivisa Connect

I also note that the 1,200W (4 panel) inverters are less than the price of two 600W (2 panel) inverters. 

 

https://www.aliexpress.com/item/32231406246.html 1,200W $164.70

https://www.aliexpress.com/item/33036019251.html 600W $98.08 ($196.16 for two)

Both prices are for the newer WiFi versions.

 

I think I'll try the 1,200W unit next.

 

I'm also working on an automatically regulated dummy load to prevent us going into export when the meter reader is in town whilst still getting the maximum from the panels when we put the kettle on (or fire up the irrigation pumps).

 

 

1 hour ago, Crossy said:

I'm also working on an automatically regulated dummy load

I think that that is what SWMBO calls me ????

This is the production graph for our best day (4.6 units), clouds were in evidence at times so there's still room for a little extra. Note the time is in China Standard Time (+1 hour from Thailand).

 

 

"Reverter" is Chinese for inverter.

Did you install the extra heat sinks with the fan blowing, and what temp difference did it make?
Sorry if I missed this update.

Sent from my SM-J700F using Tapatalk

9 minutes ago, carlyai said:

Did you install the extra heat sinks with the fan blowing, and what temp difference did it make?
Sorry if I missed this update.

 

Both the inverters currently in operation have fans which have dropped the peak temperature (measured by the inverter software) to about 60C. I did try additional passive cooling but it didn't really make a massive difference, for now I'll stick with the fans.

 

Do you know how your meter will react to a positive energy export? wi it run backwards, stop or does it run forwards irrespective of the flow of energy?... During a peak sunny day you could shut down your ACs and other high energy item to see the meter in an export phase...

4 hours ago, sfokevin said:

Do you know how your meter will react to a positive energy export? wi it run backwards, stop or does it run forwards irrespective of the flow of energy?... During a peak sunny day you could shut down your ACs and other high energy item to see the meter in an export phase...

 

You think our standing load involves A/C for me? It's all keeping her ruddy koi and the like alive ????

 

Anyway, the meter disc definitely goes backwards, we've not exported enough at any one time to check that the numbers also reverse, that's a job for when we install the next batch of panels. Maybe next month depending upon funds.

 

Well, on Thursday afternoon we had a power failure whilst the system was at almost full output. When I got home both the 20A MCB and the RCD feeding the inverters had tripped.

 

Close both breakers, no bang but it was immediately obvious one unit wasn't generating. Steady red light ????

 

 

Spoke (by email) to the manufacturer's support chap (he's been very helpful in the past).

 

Anyway, with the lid off it's fairly obvious what the failure is - the "nothing there" used to be a fuse.

 

 

 

Fuse annihilated, MOSFETs fried! Spares on their way tomorrow, free (just pay shipping).

 

It seems to be a common failure mode of these inverters, remove the supply whilst they are feeding power and the driver devices die.

 

But if it's just the FETs and the fuse it's a relatively easy and cheap fix. I will make sure I have spares in stock.

 

 

so, was it a Mains bump? (brown out)

 

or you think it was a failure initiated from your device?

Just now, tifino said:

so, was it a Mains bump? (brown out)

 

or you think it was a failure initiated from your device?

 

It was an external failure, the juice was off for 30 mins or so. The FETs seem to die at the point the mains goes off, then the fuse pops when it returns.

 

It only seems to be an issue if the inverters are working hard at the time, usually it's dark (storm clouds) when the power goes out. I'll be investigating bigger (V and I) MOSFETs.

 

The support chap is very helpful so I have no real issues continuing with these units, fix it when it dies.

 

 

maybe put in some 240VAC delay relay timers (normally used for separating aircon restarts)  

1 minute ago, tifino said:

maybe put in some 240VAC delay relay timers (normally used for separating aircon restarts)  

 

My thought is that the FETs dies when the supply goes off (power has nowhere to go, voltage rises to the point the FETs give up). Failure mode seems to be a G-S short which sends the fuse into orbit when the supply returns.

 

I'll be talking to the support chap and engineers, maybe there's a (more expensive) device that's more robust that will work as a replacement.

 

The fact that the manufacturer is talking to me is, to say the least for a Chinese outfit, unusual.

Need to rig a snubber to the circuit.  Common problem, especially if there is any capacitance in the system.

1 hour ago, tjo o tjim said:

Need to rig a snubber to the circuit.  Common problem, especially if there is any capacitance in the system.

 

Any suggestion for values? There's an awful lot of energy that's suddenly got nowhere to go.

Still waiting for the manufacturer to come back to me with shipping for replacement FETs.

 

These are the beasties MMFT65R195P, 650V 20A N-Channel MOSFET MMFT65R195P.pdf  2 bucks a pop from AliExpress https://www.aliexpress.com/item/32956754206.html I shall get a few spares in.

 

I think I'll also replace the surface mount fuse with a regular 5x20mm fuse and holder. There's already some damage to the copper, if the SMT fuse pops again it may become difficult/impossible to replace. 

 

I also note the presence of a 470V MOV at the mains input. I would have thought that would be fast enough to kill any damaging spikes, evidently I was wrong ????

 

51 minutes ago, Crossy said:

Any suggestion for values? There's an awful lot of energy that's suddenly got nowhere to go.

Sorry, I R a PE, but nearly flunked basic circuits...  Past experience is with VFDs on large pumps on substations that still had power factor correction capacitors installed (and large 6-pulse Liebert UPSs).  In those cases you just needed to limit the reverse-bias voltage with a fairly small RC network, generally on the DC bus. The other approach that might work is a surge suppression capacitor, but I was told the snubber was more effective. 

21 minutes ago, Crossy said:

I also note the presence of a 470V MOV at the mains input. I would have thought that would be fast enough to kill any damaging spikes, evidently I was wrong ????

 

Yeah... I made that mistake too.  That is where a surge suppression capacitor would help limit the rate of rise so the MOV has time to do its job. 

1 hour ago, tjo o tjim said:

Yeah... I made that mistake too.  That is where a surge suppression capacitor would help limit the rate of rise so the MOV has time to do its job. 

 

Interesting, there's a space on the board for a cap across the MOV, I shall ask the engineers.

 

EDIT I also wonder if those gaskets are heat conductive, since I don't need IP65 I may replace them with heatsink tape and see if that improves the internal temperatures.

 

Sorry if this is unhelpful, but there's a YouTube video by Sam Ben-Yaakov on building a MOSFET snubber. Bit more involved than a single capacitor.

Sent from my SM-J700F using Tapatalk

Didn't you say a few posts ago that you have a surge suppressor fitted?

Here is a less than helpful bit. I have had experience with both direct drive output (high frequency) and transformer output (low frequency) inverters. The only ones which lived are the transformer isolated ones. I would guess the transformer impedance is useful. I should mention that the inverters I am talking about are not the micro types.

1 hour ago, Muhendis said:

Didn't you say a few posts ago that you have a surge suppressor fitted?

Here is a less than helpful bit. I have had experience with both direct drive output (high frequency) and transformer output (low frequency) inverters. The only ones which lived are the transformer isolated ones. I would guess the transformer impedance is useful. I should mention that the inverters I am talking about are not the micro types.

 

Yeah, the idea behind using these baby inverters was/is two fold:-

• On a budget, add panels or inverters one at a time, any time without major (any) re-configuration. Just add the inverter to the end of the daisy chain and plug in its panel(s).
• The inverters do MPPT on a per-panel basis, so, at least theoretically, you should be getting best power transfer at all times.

I also wanted to be able to control the inverters indvidually to tailor the output for those days when we don't want to export (when the meter reader is in town). Unfortunately the ability to talk to the inverters locally has gone away with the latest versions so an alternative "do not export" solution is in development (Arduino controlled dummy load which monitors for export and dumps excess power).

 

44 minutes ago, Crossy said:

 

Yeah, the idea behind using these baby inverters was/is two fold:-

• On a budget, add panels or inverters one at a time, any time without major (any) re-configuration. Just add the inverter to the end of the daisy chain and plug in its panel(s).
• The inverters do MPPT on a per-panel basis, so, at least theoretically, you should be getting best power transfer at all times.

I also wanted to be able to control the inverters indvidually to tailor the output for those days when we don't want to export (when the meter reader is in town). Unfortunately the ability to talk to the inverters locally has gone away with the latest versions so an alternative "do not export" solution is in development (Arduino controlled dummy load which monitors for export and dumps excess power).

 

All good stuff...........until. 

When you loose one micro, do you loose all output or is the damage limited to the one?

3 minutes ago, Muhendis said:

When you loose one micro, do you loose all output or is the damage limited to the one?

 

Just the one, unless (like ours) it opens the protection in the distribution board, but all bar the dead unit work when the breaker is reset.

 

The inverters are daisy-chained, just a straight through mains connection from one side to the other.

34 minutes ago, Crossy said:

 

Just the one, unless (like ours) it opens the protection in the distribution board, but all bar the dead unit work when the breaker is reset.

 

The inverters are daisy-chained, just a straight through mains connection from one side to the other.

Ah well. That's another theory gone flying out the window. The thought occurred to me that the micros failed to sync when the city power returned and so bang went the units but even so the question of why one unit failed and the other(s) didn't needs to be addressed. What with that and the high running temperature I think I would avoid the opportunity to go down the micro inverter route.

42 minutes ago, Muhendis said:

The thought occurred to me that the micros failed to sync when the city power returned and so bang went the units

 

There's a number of theories abounding upon the web, the main one being that these units don't like the supply being removed when they are supplying power near maximum. This is borne out by another poster having a similar issue when he removed the mains to add an extra panel, the inverter didn't return ????

 

It's certainly annoying, but hopefully will be a sufficiently rare occurrence that it won't be a major issue (our power failure are usually when it's dark from rain clouds). I have a source for parts (not the manufacturer). Now I've had the lid off one I have some ideas for improving the cooling.

 

EDIT I also wonder if there could be transmission-line effects going on here, fast edges get reflected from the open end of the daisy chain.

 

Maybe looping the end inverter back to the source would help. What it would do is allow the removal of a failed inverter without having to shut down the entire array and bridge out the gap. 

 

Good for keeping the mind active.

40 minutes ago, Crossy said:

Now I've had the lid off one I have some ideas for improving the cooling.

I'd be interested to see the cooling arrangement plus your improvements.

The business side of the heatsinks.

 

 

The power devices are on the back of those 3mm thick extrusions. There are some pretty minimal fins on the long sides (bottom and top of the photo).

 

This is the cover, I don't know if that black gasket is heat conducting but it certainly does not cover the entire back of the heatsink so it's limiting the heat flow out to that nice big potential radiating surface.

 

Task A is to remove that gasket (I don't need the IP65 rating) and put that cover on with some heatsink compound or tape. The compare with an unmodified unit. 

 

Some matt-black heat-conducting paint or (better) black anodising of that cover too is worth looking at. 

 

I could replace that entire plate with a finned beastie, but that may be more than somewhat costly.

 

 

2 hours ago, Crossy said:

the main one being that these units don't like the supply being removed when they are supplying power near maximum

That is undoubtedly a design shortfall. I would suspect feedback loop stability of an individual unit.

What to do to fix it? Well, not a lot from the customers perspective except to automatically disconnect when city power is lost, but that's rather like shading the solar panels to keep them cool. In other words the solution defeats the object.