EEVblog 1610 - Deye Hybrid Solar Inverter

[Brumby]

Well, an effort was made.

Maybe consider acquiring a borescope camera...?

[Kleinstein]

With the panels oriented east / west there is a good chance that one would not get the full nominal power anyway. So the panels may well be OK from the size.

[Brumby]

If things come down to a warranty claim, I can imagine the real, on-site physical realities may not play as important a part as the simple numbers on paper would.

Whether or not you would do this, let's look at the following hypothetical... 

• Say you were wanting up to 8kW of solar power on the best, sunniest day possible but, due to circumstance, you had to install a string of solar panels rated for a maximum of 12kW to achieve that. 
  
• Then say you had a piece of attached equipment rated for 10kW which failed.

The numbers on the specification sheets would depict a clear mismatch - so what would you expect the result of a warranty claim to be?

As an extension to this question, let's say you have run this system for 4 years covering some extreme examples of high insolation and your history shows the maximum instantaneous power generated at ANY point was 8.2kW.  What then would the answer to the warranty claim?


Edit: Maybe I'm just getting cynical, but when it comes to warranty matters, my life experience has taught me to be very mindful and tread carefully.

[EEVblog]

With the panels oriented east / west there is a good chance that one would not get the full nominal power anyway. So the panels may well be OK from the size.

For a good part of the day, yes. But my roof is fairly shallow pitch, so in summer there is going to be pretty close to full power on both arrays for a good part of the day.
Better to be safe than sorry.
And the two spare 440W panels I can mount elsewhere and have the DC cables come down direct so I can play with various microinverters and other products. e.g. I could hook them up to my Hoymiles microinverter mounted on the wall and connect into the Generator input of the Deye for extra potential battery charging.

[EEVblog]

If things come down to a warranty claim, I can imagine the real, on-site physical realities may not play as important a part as the simple numbers on paper would.

Whether or not you would do this, let's look at the following hypothetical... 

• Say you were wanting up to 8kW of solar power on the best, sunniest day possible but, due to circumstance, you had to install a string of solar panels rated for a maximum of 12kW to achieve that. 
  
• Then say you had a piece of attached equipment rated for 10kW which failed.

The numbers on the specification sheets would depict a clear mismatch - so what would you expect the result of a warranty claim to be?

As an extension to this question, let's say you have run this system for 4 years covering some extreme examples of high insolation and your history shows the maximum instantaneous power generated at ANY point was 8.2kW.  What then would the answer to the warranty claim?

Edit: Maybe I'm just getting cynical, but when it comes to warranty matters, my life experience has taught me to be very mindful and tread carefully.

I'd say there would be some buffer built into that MAX string rating for this very reason. e.g. panels can actually give out more than their rated label power on a high solar insolation day.
But if you installed a system with a panel label capacity greater than the max, they'd have every excuse to not honor any warranty IMO.

[Brumby]

But if you installed a system with a panel label capacity greater than the max, they'd have every excuse to not honor any warranty IMO.

That was the essence of my point.

[madires]

Yep, surprisingly good build quality. But I'm still quite wary because of the RelayGate/DeyeGate last year (https://www.eevblog.com/forum/renewable-energy/deye-pv-microinverters-lack-a-safety-relay-(eu)/).

[nctnico]

If things come down to a warranty claim, I can imagine the real, on-site physical realities may not play as important a part as the simple numbers on paper would.

Whether or not you would do this, let's look at the following hypothetical... 

• Say you were wanting up to 8kW of solar power on the best, sunniest day possible but, due to circumstance, you had to install a string of solar panels rated for a maximum of 12kW to achieve that. 
  
• Then say you had a piece of attached equipment rated for 10kW which failed.

The numbers on the specification sheets would depict a clear mismatch - so what would you expect the result of a warranty claim to be?

As an extension to this question, let's say you have run this system for 4 years covering some extreme examples of high insolation and your history shows the maximum instantaneous power generated at ANY point was 8.2kW.  What then would the answer to the warranty claim?

Edit: Maybe I'm just getting cynical, but when it comes to warranty matters, my life experience has taught me to be very mindful and tread carefully.

I'd say there would be some buffer built into that MAX string rating for this very reason. e.g. panels can actually give out more than their rated label power on a high solar insolation day.
But if you installed a system with a panel label capacity greater than the max, they'd have every excuse to not honor any warranty IMO.

In such cases the MPTT trackers just throttle down to the capacity the inverter supports. When I installed my 4.2kW Growatt inverter it was set to 2500W max. by default and it throttled down to stay within that limit. After I configured it correctly, It peaks to 4.5kW when a cloud moves away and then throttles down to 4.2kW. But I've noticed it can do this balancing asymmetric so one string delivers 2500W and the other 1700W. But this is on relatively cold days during the spring.

You could say that my inverter is undersized but it is kind of what is recommended for a 4500Wp setup. And realistically, on a hot day it peaks at around 3900W due to the panels being quite hot.

[EEVblog]

Yep, surprisingly good build quality. But I'm still quite wary because of the RelayGate/DeyeGate last year (https://www.eevblog.com/forum/renewable-energy/deye-pv-microinverters-lack-a-safety-relay-(eu)/).

I'd still err on the side of giving them the benefit on the doubt on that one. It's unlikely they deliberately made the decision to remove a relay in just one country to save a few cents. Of course if they screwed up then they should fix it.

[EEVblog]

UPDATE:
Someone has the exact same inverter (in pieces) and is sending it to me for a complete teardown.
Deye actually market the use of premium component suppliers in the manual, but I've seen photos of this unit and it uses Aishi caps   

[Xena E]

UPDATE:
Someone has the exact same inverter (in pieces) and is sending it to me for a complete teardown.
Deye actually market the use of premium component suppliers in the manual, but I've seen photos of this unit and it uses Aishi caps   

Does it make anyone else cringe when they see HV lytic caps (particularly those ones¹), apparently sitting directly on top of the solder mask?

¹ more bang for your buck.

[thm_w]

Does it make anyone else cringe when they see HV lytic caps (particularly those ones¹), apparently sitting directly on top of the solder mask?

¹ more bang for your buck.

What is the issue with that?

In such cases the MPTT trackers just throttle down to the capacity the inverter supports. When I installed my 4.2kW Growatt inverter it was set to 2500W max. by default and it throttled down to stay within that limit. After I configured it correctly, It peaks to 4.5kW when a cloud moves away and then throttles down to 4.2kW. But I've noticed it can do this balancing asymmetric so one string delivers 2500W and the other 1700W. But this is on relatively cold days during the spring.

You could say that my inverter is undersized but it is kind of what is recommended for a 4500Wp setup. And realistically, on a hot day it peaks at around 3900W due to the panels being quite hot.

Yeah I would hope so if its properly designed, assuming the panel string is within the VOC max, I would think the only downside is the >3250W power going into heating up the panels a bit. I assume the rating is there more so people don't put 5kW all on one string and only get 3kW out.

[nctnico]

UPDATE:
Someone has the exact same inverter (in pieces) and is sending it to me for a complete teardown.
Deye actually market the use of premium component suppliers in the manual, but I've seen photos of this unit and it uses Aishi caps   

Does it make anyone else cringe when they see HV lytic caps (particularly those ones¹), apparently sitting directly on top of the solder mask?

Nope. I don't see a reason why this is bad at all. Actually, having the capacitors mounted tight to the PCB makes sure they don't wiggle under vibration which could break the pins at some point. Adding some glue to stick the capacitors to the board is even better for durability.

[Phoenix]

Your comment about the MOSFET packages around 15min - they look like a custom mounting bracket that allows a single screw to apply pressure to two TO-247 packages. It's not a special package.

[Phoenix]

I've done a quick markup of the mainboard layout too if anyone is interested. I can also help with a more detailed reverse engineering when you get the disassembled boards - especially around some of the hairy details if you want to deep dive things like conversion topologies, purpose of specific components etc.

[EEVblog]

I've done a quick markup of the mainboard layout too if anyone is interested. I can also help with a more detailed reverse engineering when you get the disassembled boards - especially around some of the hairy details if you want to deep dive things like conversion topologies, purpose of specific components etc.

Thanks. I'll post high res photos when I get to this.

[EEVblog]

Only just got around to this, but teardown photos here:
https://www.flickr.com/photos/eevblog/albums/72177720317489696

[SiliconWizard]

Damn, there's black resin over all ICs.

[Phoenix]

Nah it's just conformal coating. Look at it with light from the right angle you'll read them. The trouble with photos is that you can't tilt them for more detail.

[EEVblog]

Damn, there's black resin over all ICs.

Main processor is an Advchip DSP
http://advancechip.com/mobile/32fd/61.html

[EEVblog]

Nah it's just conformal coating. Look at it with light from the right angle you'll read them. The trouble with photos is that you can't tilt them for more detail.

Yes, in this case you need near horizontal light to just be able to see the numbers.

[Phoenix]

Main processor is an Advchip DSP
http://advancechip.com/mobile/32fd/61.html

They aren't subtle about what they are trying to copy...
https://www.ti.com/product/TMS320F28335#features

[nctnico]

So much thermal grease. Yuk!

[EEVblog]

So much thermal grease. Yuk!

Impossible to avoid in the teardown, it just gets everywhere.

[nctnico]

So much thermal grease. Yuk!

Impossible to avoid in the teardown, it just gets everywhere.

That is what I mean. Using heat conductive pads would have been so much nicer from a service standpoint (and likely manufacturing as well).