EEVblog 1620 - Deye Solar Hybrid Inverter EXTREME TEARDOWN

[@EEVblog]

UPDATE: Deye have responded and asked me to make this post: "As an employee of Deye in Australia, I would like to address this issue with complete transparency. Deye, as a globally leading manufacturer of energy storage inverters, has always been committed to providing our customers with cost-effective and high-quality products. Our product failure rates are relatively low compared to our peers in the industry. During the past couple of years, the pandemic caused significant supply chain challenges, making it difficult to procure certain components while the demand for energy storage solutions surged. To ensure continuous product supply, we had to use alternative components. Please rest assured that all these alternative components are A-grade and have undergone rigorous testing to meet industry standards and certification requirements. The inconsistency between the components in our products and those mentioned in our catalog is due to our failure to update the catalog （2022-2023 Version）in a timely manner. We sincerely apologize for this oversight. We have taken immediate steps to rectify this and will update our catalog as soon as possible._ "

[@Petertronic]

Hmm, this unit was made in 2023, I would have thought that was well after the supply shortages recovered. There is NO shortage of Panasonic, Omron etc decent quality relays that I am aware of.

[@kurohyuki5912]

They can switch suppliers but they can't update a brochuer? Hmmmkay.

[@ThePonderi]

Interesting. But the inverter is widely used. We will see.with time

[@dr_jaymz]

When substituting components for chinese equivalents sometimes they can be superior especially recently. The challenge is to maintain that consistency in the supply chain because the first 10,000 off may even be rebadged as chinese when they are not until you replenish further down the line and find out they change it. If the manufacturer monitors failure rates and keeps track of what fails the results can be surprising, why spend 5$ on a relay when $1 has no additional failures. We perhaps need to change our perceptions that its automatically rubbish if note made here or a brand we don't recognise, in a lot of cases they come down the same production line. Its maintaining that quality and keeping fakes out of the supply chain thats a pain - and that applies to branded stuff as well.

[@hilo90mhz]

@@ThePonderi my brand new new Deye 50K uses Nichicon capacitors, didn't go further inside to check the semiconductors. But some are still using the old brands.

[@BrianHG.Ocean.Fitness]

@22:05 - The HERIC patent expired last year. It's now public domain.

[@stevenbliss989]

GREAT!!!!!!!!!!!!!!!!!!! :):):)

[@simontillson482]

I don’t know how they ever managed to obtain or enforce that patent. Multilevel inverter topologies have been published in literature for decades, and the single-phase tri-level topology is both commonly known and bloody obvious. Where is the innovation in it?

[@ignispurgatorius5297]

@@simontillson482 The whole patent system is a swamp of morons frankly. It starts with the patent offices and ends with the courts. My company holds a patent for a simple mathematic operation on a measured value in a niche industry. Everytime someone mentions it I can't help but shake my head... Likewise some of our competitors hold patents that block us because they managed to patent basic mechanical principles. One of them even managed to patent a simple translatoric servo system. It's not about actually protecting ingenuity, it's just a way for big companies to troll one another.

[@WizardTim]

Although don't forget while the original EP2086102A3 patent has expired *in some regions* , the name "HERIC" is a legally separate and distinct registered trademark which will never expire (if renewed correctly).

[@BrianHG.Ocean.Fitness]

@@WizardTim I don't believe any inverters sold have ever boasted in print 'Contains HERIC(TM) technology.' This would infringe on the use of the trademark name. Today, the inverters just boast a global conversion % efficiency.

[@TheDefpom]

Company should be done for false advertising, making false claims about their product, they are obviously using the name brands as a sales point otherwise they wouldn't even mention the brands used.

[@zentechnician]

Made of the finest Chinesium you can get though. We promise.

[@tschuuuls486]

They almost got away with ommiting the AC main relays in their microinverters (or not using them). They left them out because it's a failure prone item. They used them for the regulatory test units but not in retail ones. German regulators didn't like that move and pulled them from the market and mandated them to install relays.

[@p_mouse8676]

I totally agree, but in that case we have to shut down 99% of all companies?

[@ursodermatt8809]

what you expect? they are chinese after all.

[@brentdennard6722]

@@p_mouse8676yes we should! But not permanently, just long enough for them to wake up to the fact that regulatory agencies don’t want them skirting the system.

[@andrewmclauchlan9205]

Just had a visit to a Deye dealer in Melbourne and checked out the 12kw 3 phase hybrid. Amongst other things I was told that Deye make every component in the unit including the resistors. I didn't argue and said thanks and left. I think the product is quite good but have my suspicions about the dealers!

[@robertheritage-i8h]

Where is the dealer?

[@andrewmclauchlan9205]

@@robertheritage-i8h Truganina

[@FrozenHaxor]

Who the f would open an entire production line just to make resistors. Unbelievable BS.

[@Zardox-The-Heretic-Slayer]

you'll probably find a single 10u 16v nichicon capacitor on the board somewhere so that they can still make the claim that they use their parts.

[@animarkzero]

My Sun-8k SG04LP3-EU which is three phase has the real deal inside---> Nichicon all the way😃 It was produced in August 2022! So maybe the had to cheap out on the way....?

[@adon8672]

​@@animarkzeromaybe this is part of the Chinese companies' response to America's trade wars. Preemptively swap out western and Japanese components before they get banned.

[@CornishMiner]

Excellent video. Fascinating. Thanks to everyone who contributed.

[@IanScottJohnston]

The planets must have been in alignment for Dave to receive a parts unit like this, and the same as he’s installing himself. Once in a lifetime clearly welcomed!…..nice!

[@Glensen]

I have a Deye hybrid 48v 3-phase 10kW. Runs like a dream. Did the initial setup, and never touched it since, never notice when grid goes down. I have 32kWh of battery capacity. Choose a 48v system for safer and cheaper batteries. And to get 3-phase and 48v battery Deye was one of very few manufacturers. And apparently they are the "goto brand" in africa because they just work.

[@paulkirn5223]

An explanation of the DRM interface. A key requirement of AS4777. 2 is Demand Response Mode (DRM) defined in the standard AS4755 Demand Response Standard which enables the inverter to respond to signals sent to it remotely. These signals trigger the inverter to change the mode of operation, potentially turning it off, on, or ramping the output up or down. To allow the electricity network operator to better manage the grid. The RJ45 would be connected to a ripple receiver (decabit/K22 signalling receiver) this device is usually provided by the local distribution network provider for the local area, the RJ45 interface board is simply monitoring contact closures from an attached receiver. DRM modes are: DRM 0 - Operate the disconnection device, DRM 1: Do Not Consume Power, DRM 2 Do not consume at more than 50% rated power, DRM 3 Do not consume at more than 75% rated power / Source reactive power if capable. DRM4: Increase power consumption. DRM 5: Do not generate power. DRM 6 Do not generate at more than 50% rated power, DRM 7: Do not generate more than 75% of rated power AND Sink reactive power if capable, DRM 8: Increase power generation. DRM modes DRM 1, DRM 2 and DRM 3 are also commonly used for controlling domestic loads such as air conditioning. EEVblog 1283- What is Mains Ripple Injection? kzbin.info/www/bejne/hqCXk2qAndWrgbM

[@ajlitt001]

There's no secondary side DC blocker because they figure that blowing the 150A fuse is the least of your problems if two of those FETs turn on at the same time.

[@EEVblog]

Yeah, fair enough.

[@RLPE]

But they must have some sort of DC elimination strategy for normal operation - it's easy to end up applying a small amount of DC to the transformer over a long period from changes in modulation (phase shift, duty cycles etc.) and then you end up with core saturation. Either there is a blocking cap hidden along with the transformer in that potted enclosure, or they're doing something fancy like measuring the core flux or one of the other uncommon ways of mitigating this issue (maybe something to do with the extra leads), in which case, why put one on the high voltage side?

[@bgdwiepp]

Probably PSFB and not running it all the way to 100%/50% dutycycle, bleeding off any extra flux in that "short" time

[@otherphoenix1]

I was thinking it could be susceptible to flux walking and maybe it was an llc as there is a large series capacitor labelled Cr5 on the hv side but there are other parts missing in that case. It's likely operated differently.

[@nshattock]

I would have thought that they would be using an LC resonant tank circuit to reduce losses and therefore only need one capacitor. I don't think it would have anything to do with safety. Not sure if they are doing a full DAB topology without looking at transformer construction and switching waveforms

[@SansNeural]

Really nice to see the guts and a bit of how it works. Last year I implemented a UPS for my whole home office with a 3 kW hybrid inverter from China. 24 volts of battery and just 400W of PV panels to keep the DC topped up. Works pretty well and has let me ride through numerous 1-3 second power drop-outs and about an hour once when a storm knocked out the power. Since the batteries spend most of their time at full charge and doing nothing, I'm testing a grid-tie inverter added on to the DC side to kick in only when the air conditioner is running. Should help defray the cost of cool this summer.

[@zentechnician]

Look how nice and soft that potting is. Super lucky there. The hard stuff I just drop whatever it is into a pot of boiling water for a couple of minutes and whatever the potted secrets it was hiding become so much easier to to uncover. Some of these ebike controllers I do work on are encased in stone like resins that require medieval methods.

[@EEVblog]

Yeah, was still annoying and sticky enough to not easily be able to get the coils out though.

[@zentechnician]

@@EEVblog That was a really good teardown man. Between you showing the parts numbers and their ability to produce a decent looking silk screen a person could practically duplicate that thing, if it were worth duplicating that is. You were showing those links etc and talking about them needing more is definitely reminding me if these potted controllers. These are very high wattage dealies from electric dirt bikes so 10k+ wattage. They are linking these phases etc so unevenly it drives me nuts. You spend 2 hours depotting the thing and you still have to get to the power stage board underneath the brains of it and then it's just all kinds of ugly when you get to it. Just biggkes the mind. No wonder they encased it like Han Solo, they were embarrassed for anyone to see it. I try to comfort myself by thinking they use a thermal camera and just add more links at hot spots and someone actually cares but that probably not a thing. What I see happening is this, one company haxs a bunch of different names they operate under, they saturated the market so it looks like there are choices. They own all the most well known brands for this specific market so they are only competing with themselves but no one realizes this. They purposely provide some of the worst documentation . China has quite a few people I've heard, but apparently one that speaks English and understands these controllers is not possible. People get fed up and just switch brands. To be fair they really go out of their way to obfuscate the fact that they are all connected. They even choose different protocols etc between brands, maybe to make you buy another programmer or maybe it's just the difference in usage in Asian markets. People are using what are motorcycle controllers on bicycles. They get you coming and going, I guess it's good business maybe? I don't run my business that way but I don't live in China either so there's that. I hope those inverters are cheap. The Chinese metric system I always just assume the numbers they give you can probably run at 33% comfortably and %50 if you want to push it so you buy 3 if the things, use 2 to equal 1 good one, have a 3rd as a spare which you are definitely going to need and you still have money left over. How can you compete with that? Sometimes you put out the money for the good one and it dies too . Step right up, spin the wheel. Everyone's a winner. Uh huh.

[@patrickpafarnis5798]

@@zentechnician Very nice explanation of how the Chinese market works.

[@zentechnician]

@@patrickpafarnis5798 To be fair our companies do the same stuff. I think it was a National Geographic article maybe that I read about a Consumer Reports type company that realized that 3 of the washing machines they had fur review were the same machine. Whirlpool owns lots of brands they dosone Kenmore stuff and Amana, I think Maytag too. Many more. So the machines all performed exactly the same, the difference was different art of the control panels and slightly fancier plastic knobs on the Amana. So there was a cheap one then the next one up was like 60 bucks more, and then the Amana with the fanciest knobs was $130 more. Everything underneath that outer shell was exactly the same aside from some branding, but it was all made in the same factory. This us like one of those brain on drugs commercials. Who taught you this? I learned it from you dad I learned it from you. lol

[@monad_tcp]

@@zentechnician They must like gambling if that's how they do business, I kind of get why they like gacha games so much.

[@Roobotics]

12:39 I don't think it was merely happenstance or 'running out of room', that they moved all the electrolytic caps away from the switching PCB, where all the toasty FET devices reside.

[@michaelvw11]

I am using this exact type of hybride inverter for over a year now in combination with a solar assistant controller. It's been flawless so far. The solar assistant controller is used as a epex price switching device, buy cheap electricity and unload from the battery when the price is high, ideaal combination.

[@deepwinter77]

I'd go for a Fronius inverter made in Austria very high quality. Excellent company they also make some of the best welders in industry. They've been around for ages too.

[@markusweissenbock6337]

Thats right, power-semis and electrolytics are exclusively Japanese, European and/or US sourced.

[@fredfred2363]

Appreciate the inverter power topology diagram. Interesting. Thanks. Like you said, the drivers must be super critical- if two opposing switches accidentally overlapped, bang- game over.

[@vincei4252]

I believe Deye makes the Solark 5K, 12K & 15K solar inverters. I've had a 12K for 4 years that has been a happy little camper. I saw a partial non destructive teardown of a Solark 5K and it indeed did have quality components and Nichicon caps. Maybe they take more care when they manufacture for partners? Solark does say they have a lot of control over what Deye manufactures for them.

[@ursodermatt8809]

and that is one of the reasons the solarks are so much more expensive. i am sure.

[@vincei4252]

@@ursodermatt8809 I know. I wasted money and time buying cheaper inverters. All my problems went away when I bought the Solark.

[@tibbydudeza]

They an OEM manufacturer - they make inverters for SunSynk as well besides for themselves - same case and board but the panel software is different, and they got their own dongle and website.

[@boots7859]

Thats the key point, external well known brands check to insure BOM parts match, because they have a reputation to maintain. SOP for Chinese manf's of Western goods will simply run extra shifts with similar Chinese parts to minimize unit cost and utilize the R&D thats aleady been done and the end profit it simply amazing. Amazing what you can make when you have cheap labor, cheap parts, and high-end existing IP already paid for.

[@monad_tcp]

The beam counters had a blast with that thing !

[@threeMetreJim]

You may have noticed that the Chinese manufacturers don't seem to believe in multi-phase converters to spread the load, just everything in parallel with the fattest inductor possible. The long thing is the mains output filter for the sinewave inverter and the round one is the boost inductor. I have a similar one, part disassembled, a growatt 2.5kW model that was dead before use. You may get a 'bang' for your buck sooner than expected.

[@EEVblog]

We'll see, fingers crossed.

[@volodymyrzakolodyazhny]

For modern cheap Chinese hybrid inverters ' you have to dissassemble them first (DON'T TURN ON), look carefully, add all points of soldering needed, etc. Only then, after carefull inspection - you can turn it on, under a power.

[@vincei4252]

I tried a Growatt Inverter for about 2 weeks and cut my losses pretty quickly. Over volting the batteries and forcing them to shutdown into protection mode.

[@piclife1178]

Looks like an EMC nightmare. Thanks for the video, very interesting.

[@MrJohnBos]

It looks like the patent on the HERIC inverter has expired and no longer applies.

[@NiHaoMike64]

How do the two extra IGBTs improve the design? I can see the point if it were a split phase design with a split bus cap, but for single phase, doesn't that achieve the same result as simply just turning on both low side or both high side IGBTs?

[@RLPE]

@@NiHaoMike64 It's the same result except for the common mode voltage applied between the DC and grid side, which if done at high frequency may create common mode noise issues due to the environmental capacitance between the solar arrays and ground. So maybe HERIC reduces the common mode noise filtering requirement.

[@NiHaoMike64]

@@RLPE Wouldn't it be more straightforward to have the AC output inductance all on the hot side and just have the neutral side be switching at a low frequency?

[@otherphoenix1]

A normal h bridge switching 3 level creates a pwm frequency common mode square voltage on the dc side. This is because during the 0 state with both high or both low switches are on and thus a common mode voltage is presented to the grid. Using the extra two switches for this 0 state means the grid side is effectively disconnected from the high voltage dc side and no common mode voltage is presented. Also need to realise that the grid is earth reference so any common mode voltage present actually appears on the hv bus side wrt earth. Now all those heat sunk devices (on an earth heatsink) and even the solar panel array are all pinging away at pwm frequency to earth. That's an emi nightmare, it'll probably just blow up from self interference. SMA had their own H5 that achieved the same thing. Also various H6 and other esoteric versions of h bridge have been proposed to overcome this.

[@otherphoenix1]

​@@NiHaoMike64 To be equivalent the pwm operated half bridge would need to operate at twice the frequency... Good luck with 1200V igbt. Or the inductor be twice as big. This is however a more feasible idea with GaN and is called a totem pole PFC. You can also look at a single phase neutral point clamped topology that gives 3 levels from a half bridge tied to the active while the neutral ties to the dc bus mid point.

[@appliedengineering4001]

@27:58 in the video. The reason they are not using a blocking cap on the battery side is because the voltage is not high enough to overcome the losses. You would need a very large cap bank on that side to break even. In the end, it's just better to omit the cap all together. Remember that the current has to go through 2 sets of MOSFETs. So if one set shorts out, the current sensor can shut down the other set of MOSFETs and break the circuit.

[@matthewchadwick7690]

At 5:48, there is a date of 2022/1/8. Maybe they switched to Nippon Chemi-Con later since the brochure is 2023?

[@ursodermatt8809]

i think the specs where for the american solark which is something like 3 times more expensive.

[@EEVblog]

@@ursodermatt8809 Deye branded brochure, and was given to the Australian owner

[@rasz]

Other way around.

[@Nikom_biz]

Thank you very much, your circuit design review style is interesting. You are a great engineer, and the video is excellent. Wishing you energy and inspiration! From Ukraine...

[@k7iq]

Deye also manufacturers larger hybrids than the 5K of course. I took one of their 12Ks apart as well which was similar to this one. There is a US company that private labels and has engineering input to these. Sol-Ark. They specify the inverter output power at battery + solar. Higher watts than with battery and DC to DC converter alone. Their web access used to be hosted in China and I think that is changing to servers being able to be hosted in the US. They are obviously a top notch manufacturer. Those caps might be just fine. Would have to measure. We use a Chinese cap company (jianghai ) that use to (still does ?) manufacture Panasonic snap-ins that are awesome and low ESR. The USA versions have PV Arc-Fault which is now required here.

[@HermanWillems]

lol USA really has a love hate relationship with China. Here in Europe we have far more real European options, for example from Europe. Where are your TRUE American inverter companies ?

[@turboslag]

I wonder what all those high end component manufacturers think about their brand names being used to market a product but not actually using their devices?! Maybe they should get the heads up so they can respond!

[@teardowndan5364]

The "HERIC" patent is effectively just pair of synchronous rectifiers head-to-head to do for AC exactly the same job that synchronous rectifiers do for DC. Haven't read the patent but I'm guessing those transistors are also used for PFC boosting as well when charging batteries from AC. If someone made a fully synchronous bridge-less PFC circuit for a power supply, it would probably infringe this patent too,

[@cinobro6393]

This teardown was so extreme it blew the socks clean off my feet.

[@randacnam7321]

Aishi (pronounced 'aye shur' cuz Pinyin) is a common cap brand to find in compact fluorescent and LED lamp ballasts, in which applications they seem to survive well enough given the roasting they get. The series connected Hongfa relays may be a cost cutting measure. I have 2 Sol-Ark 12K inverters made by Ningbo Deye (pronounced 'duh yuh') and all of the main AC relays are Japanese made Panasonic/Matsushita rated for 63A.

[@tejeshwvardhan1165]

wow let me get my popcorn for the show! do more teardown on grid connected inverter!

[@rowlandcrew]

I disassembled my dead solar edge inverter to find the failure. It appeared to be durably designed, e.g. no burned fuses, igbt's, relays, diodes. What failed were the AC HV film caps and the MOV's. Getting the unit apart was annoying because they heat sink glue the solder side so everything needed to come apart. Quite explosive failure on the MOV, big black skid mark. The film safety cap apparently incrementally burned off layers internally, did not short, and ended up with pico farads and a bulge. Getting the safety caps were a problem with lead time on digikey probably because of high demand in making inverters. Once replaced the unit worked as before. I have a power monitor that can plot inverter output with one second resolution where I have noticed occasional sub minute power drop outs. It appears the film caps continue to have internal shorting events, shutting down the inverter, however the inverter software apparently expects this so it waits ~30 seconds and restarts. Therefore it appears the lifetime of these inverters is proportional to the average rate of internal shorting events times the decrement in capacitance/event to the point where the capacitance is so low that the internal inverter generated switching noise is no longer meeting specified limits. Unless the MOV pops. Because solaredge built in a capacitance glitch algorithm, I expect they have a secret specification for the lifetime of their inverters.

[@johnsonlam]

Thanks Dave, fun tear down shows sometimes marketing tricks just greatly exaggerate lies.

[@Timuragashka]

I wanted to mention a potential issue I’ve encountered with this inverter model. If one of the solar array wires accidentally shorts to ground, there doesn’t seem to be any built-in ground fault protection. This could lead to damage to the inverter, especially during events like lightning strikes when SPDs (surge protection devices) short to ground. It would be great if future models included protection for these kinds of faults to prevent such failures.

[@KeritechElectronics]

Discombobulated litz wire, eh? It's a hyae-naaa... Despite the parts (who said it was ALL Nippon Chemicon or OnSemi?), it's a nice unit. As for service links, I don't really get it that they used those teeny tiny pieces rather than a length of wire soldered on top like at 37:05. That should be doable with pick&place machines too. The bi-directional DC/DC converter is beautiful in its simplicity and it's worth nothing that in a H bridge, the body diodes on non-conducting MOOSEFETs will also work as a FULL BRIDGE RECTIFIER! Quite an effective one, too - with very low drop... and with good cooling too.

[@EEVblog]

Those links are handy because you can fit thousands of them on one slim reel in the PnP machine, low weight, fast to place etc. With those body diodes, I've been meaning to do a video on how regen braking on cars work which uses those nicely.

[@matthewmarks6951]

@@EEVblog You surprise me - I would have thought that body diodes would have far too much voltage drop compared with turning the FETs on, but I bow to your expertise!

[@maxtorque2277]

@@EEVblog Regen braking absolutely does not use the body diodes! Voltage synthesis is active at all times ie gates are being fired. Body diodes are really only conducting during dead time period and these days that's very short

[@MMMM2MMMM2MMMM]

When are you going to install the Deye on your Enphase system? I would like to know how you would install the Deye AC coupled.

[@gkasprow]

I don't see any external inductors at the battery converter. The blocking cap could be part of resonant circuit. Isn't it resonant converter when charging battery and simple H bridge when discharging?

[@MartinKL]

Those Vishay diodes would be hard to substitute on the gray market. I bet they tried though. I'm sure they own a good thermal camera.

[@florremdiw]

where could I find the original inverter topology presented here? (from a eevblog member called phoenix?)

[@ElektronikLabor]

15:43 Does anyone know the article number of that or similar smd current links❔ I need some for an 120A application

[@flatfoot]

Great thanks. A little more detail on the CT inputs would be lovely - I've got issues with one & don't fancy stripping it down

[@uni-byte]

Did you ensure there was absolutely no Nippon-Chemicon caps in there? After all, they did not say "no Aishi capacitors" did they? 🤣

[@Drillmechanic]

There is a small inductance on the secondary winding of the transformer, so it is necessary to install a capacitor of large volume, but for low voltage, to synchronize the resonant frequency of both sides. But I think that it is more expensive to arrange this bidirectional resonant frequency synchronization than to install a large capacitor. And since the secondary winding has very little inductance, there is no need to compensate for these MosFET on/off switching losses!

[@thehouse-ff7mk]

The battery side of the HF transformer has no blocking capacitor as it is a phase-shift full bridge topology. A blocking cap would create an LC resonant circuit with the leakage inductance of the transformer. It needs a proper square-wave waveform to moderate its power flow. Infineon make an engineering sample PSFB DAB design.

[@chrisw1462]

@35:35 The other advantage of series relays is, IF they both open at the same time (or VERY close to it) it reduces or eliminates the arc across the contacts. The path between the two is effectively isolated when they open, so no place for the arc to go.

[@typxxilps]

3:03 - who would have thought that they have a coin cell battery on the back of the display? That means the battery cell will fail sooner or later. Would not have believed that.

[@mihaiachim5299]

don't worry - the inverter will fail much faster than the battery😅

[@matthewmarks6951]

@@mihaiachim5299 Yes, and also there should be no load on it unless there is no grid, PV or storage battery! I've got a heating programmer that's well over 20 years old and its CR2032 still powers it fine if the mains fails.

[@hilo90mhz]

It's just for the realtime clock.. could last 10-20 years in that duty. The display PCB controls all the communication and user interface and scheduling for different time based modes.

[@stusue9733]

@@hilo90mhz You may well be right but given where it is, wont it be a throw-away when it does? Given we are "trying to save the planet"(TM) that doesn't seem like the bet idea ever.

[@hilo90mhz]

​@@stusue9733it's actually quite easy to get to the back of the screen, I have opened mine. Just the 8x hex screws and you're there. So really not hard to replace.

[@dr_jaymz]

There's a important point here regarding the PV switching that he doesn't go in to in much depth but the parasitic capacitance of the panels to ground is a significant one. After installing a hybrid system I noticed a huge 50Hz square wave electric field in my workshop. What happens is that the panels virtual ground is switch to neutral on the positive and negative part of the sine wave. This means that the panels jump up and down by several hundred volts with respect to earth so if you have them close by in the roof then can get capacitive coupling. It's the nature of the design. if you don't ground your panels then the frames will have a significant AC voltage with this weak coupling and if they are grounded or you have them on a flat tin roof then presumably thats a parasitic loss?

[@CurrentConnected]

This is crazy different from the Sol-Ark 5k that everyone claims is manufactured by Deye (Pronounced Doy-YAH) that we tore down in our video: kzbin.info/www/bejne/nmXQioCCoNiKrNU

[@daveluxton8317]

agreed. however, the main DSP chip seemed to be the same. Your teardown vid was awesome btw. We used to be a Deye dealer, but now we sell SolArks. reliable and best support in the biz.

[@DanBurgaud]

29:10 if there is a DC blocking caps here, it would be very very very VERY large capacitance. ie, there is 100A flowing here. Deye opted to use a small 150A fuse or allow the IGBTs to explode instead

[@brunonikodemski2420]

The round transformer is wound with Litz Wire, and used for higher frequency systems. We used this for frequencies above about 250KHz. That's why youre seeing the fast rectifiers. Vishay is actually a good brand.

[@LBCAndrew]

I just noticed you've got a JBL Studio Series S26 speaker behind you. I've got a pair of S310, pair of S38, and S-Center for my TV setup. Great speakers.

[@jordanhindes769]

Would love to see a talk through dc blocking in transformer topologies such as H bridge, push pull etc and maybe loop in snubbers, resetting the core techniques. Maybe flyback converters Very interesting topic !

[@tomg8069]

Do you know if they used tlp or hcpl as a drivers for igbt and mosfets? I would expect yes to have separation....but can see a few dc dc converters for ppwering optical drivers...If they used drivers like ir without separation then rather bad idea.... Personally i would use push pull on battery side. Also missing current sensor on primary side but yes... LEM sensors like lzsr200 are expensive ;) do they use phase shifting for battery transformer or just pwm one side and use built in diodes on second side? This is less effective but easy to controll. Quite poor mcu for such a machine...

[@mihaiachim5299]

@13:05 at 150A/36-48 v should arc like a welder in the right conditions on that small distance …. :p 😅

[@volodymyrzakolodyazhny]

Yes, definitely.

[@JanicekTrnecka]

If that fise is going to blow, I think that arcing in this particular area is one of the "minor" issues.

[@guytech7310]

I presume the 150A fuse substitutes for the DC side H-Bridge blocking cap. Issue with blocking cap is heat, reliability & board space (AC caps are bulky) and cost.

[@manuelcilia391]

you make my head spin, but very informative

[@springer9406]

At 15:30, perhaps those "pads" (none connected to each other) are for heat dissipation?

[@mrnmrn1]

Not likely, because shiny metal surfaces have low IR emissivity, so their capability to dissipate heat radiatively is very low. Copper with solder mask on it radiates heat an order of magnitude better. Emissivity of shiny metal is like 0.05-0.1, with any non condutive coating it goes up to 0.8-0.98.

[@springer9406]

​@@mrnmrn1 LOL. OK sure, pal.

[@smartups1]

Can you show his wave form of inverter . One of these I saw zero crossing distortion .

[@bigjd2k]

So the solar panels are live! As well as being high voltage themselves. Spicy!

[@hybridsolarwale]

We are happy with LUXPOWER HYBRID INVERTERS, some are running since 6 years and warranty period was 5 years . Luxpower more stable

[@brentdennard6722]

Very revealing. I’ll never buy this company’s products now thanks to you Dave. It’s looks like a fantastic quality product but the lies… the lies are enough to scare me away.

[@EEVblog]

Fair call.

[@hilo90mhz]

Pretty sure it's a matter of the sales people not updating that brochure... I have their HV model and it does have real TI DSP and I believe Nippon capacitors.. didn't check the fets.. but they have used all those brands in the past, they're just not keeping up with the marketing materials as they lower costs.

[@mihaiachim5299]

@ 30:47 Loctite SF 7200 to remove the conformat coating😊 ( but keep It away from the screen of your fluke multimeter - It will put a hole in It 😅)

[@blockbertus]

Sounds like someone is speaking from experience. ;-)

[@SurvivalSquirrel]

@@blockbertus Any stuff would work. I learned this through an accident ;-)

[@bonnotdidier5572]

Hi , if the two PV inputs are in paralell, can one ONLY PV string be wired on the two inputs in paralell? what about the mppts work ? any conflict ?

[@RollsFPV]

Just wondering if this unit can be used in an off grid application with no grid connection?

[@Juergen_Miessmer]

Yes of course.

[@mfx1]

It looks like that board/those caps are dated before the brochure so maybe they switched to better parts after this board was made?

[@markmuir7338]

I’d like to see the waveforms during different modes of operation. Especially if the frequency shifts out of grid spec on the generator input to tell the microinverters to shut down. This is how the Tesla Powerwall/Gateway system handles off-grid operation with microinverters when the battery is full.

[@mohamada1]

Can you get me datasheet or the manufacture of SMD high current links for PnP benefits Thanks

[@dinissilva5688]

So the transformer is not isolating the mains voltage from the solar panels? Can't it be dangerous to have the mains voltage not isolated from the PV inputs? When I saw the transformer, I though it was going to isolate the the mains voltage from the PV inputs and not the battery inputs from the rest.

[@ArtemMelanich]

I'm amateur in this stuff.. but can we connect a battery through a beefy capacitor to add this short circuit protection? Although BMSes usually have some kind of protection for that I think.

[@jimviau327]

Nowhere did we see a claim pertaining to the Brand page. Did they specifically mention that those are the brands utilized? That's exactly their games. They lure you into make-believe, but will never mention it as a legally binding fact. Showing an image doesn't mean much. China sell you dreams. We just have to get used to it and decide appropriately. Nice teardown, by the way. I love it.

[@Beta_Trader]

Very impressive tear down! Thanks. What about a tear down of Lux power inverter?

[@kozlovskyi]

Why there is no SiC + GaN inverters on a market?

[@stefanvanzyl9090]

nobody is making large quantities of cheap high powered GaN mosfets or IGBTs yet. Certainly not cheap enough to offset their cost in the device's expected 5-15year lifespan with efficiency improvements only

[@conkerconk3]

How long do you reckon it would take to design the circuity for one of these from scratch, maybe including test circuits too? I'm just getting into the EE field and don't know how to properly gauge the time it takes to design a large circuit, let alone these monster circuits.

[@EEVblog2]

Oh wow, that's a how long is a piece of string question.

[@TheDefpom]

Not sure I like the inductors mounting inside there, just dropped in and touching each other on wonky angles, might mean there is only a couple of layers of enamel to stop a short between different transformers, that isn't good.

[@mihaiachim5299]

There is some kind of plastic insulation between them @ 7:54

[@brentdennard6722]

Yeah, you overlooked the plastic sheets dividing each toroid. Pretty great build quality, but why the lies? I hate false advertisement sooooo much

[@TheDefpom]

@@cannesahs yeah, thanks

[@mihaiachim5299]

@@brentdennard6722 the cure is not to buy :) But we love good lies…. That’s the problem … We love good specs for good money … and sometimes you can achieve this through an honest and successful design and other times through lying…

[@EEVblog]

@@brentdennard6722 Yes, there is an insulation sheet between each one.

[@dragonfireproductions790]

I install setups using these deye inverters it's been running for 3 years straight with no hiccups

[@feicodeboer]

When the grid goes down, how does the system prevent that the generated current goes into the grid instead of your home? Did I miss something or is there external circuitry required?

[@feicodeboer]

From the manual I understand that there is a separate 'backup load' that gets the fine stuff on grid outage.

[@Juergen_Miessmer]

If the grid goes down, some relays will separate the grid from the inverters AC rail.

[@sergepetrov8598]

Here people use hybrid invertors as household-size UPS, even without any solar.

[@jetfu400]

Do more videos like these, review the insides of inverters. Make comparison vs high frequency inverters vs low frenquency. Which is more durable, reliable from the inside out... Appreciate your videos thanks very much!!!

[@TimonEckholdt-cx9ns]

1620? I can't beleve it! Have seen around 1k videos on so many diferent devices. ❤❤

[@EEVblog]

I'm old.

[@TimonEckholdt-cx9ns]

No way

[@dama054]

Great video quick question would you class this as a low frequency inverter or a high frequency inverter

[@philipbotha6718]

I highly doubt they're using a three level inverter. In almost all multi-level inverters the IGBTs are still either on or off. The losses otherwise are too high. The MOSFET may suitable for use in a three level inverter, probably neutral point clamp (NPC) but this does not look like a multilevel topology.

[@AntonyBriggs]

What's the image handling software Dave is using in this video?

[@triplex7144]

How the unit performing? Im looking at getting one

[@davidwillmore]

Do you really need a dc block on the battery side? The hbridge already offers two steps of switching. If a mosfet fails, then the controller just turns off the corresponding mosfet on the other leg.

[@EEVblog]

Electrical part failure and permanent MOSFET shoothrough. BUt as someone pointed out, it's 150A fused.

[@mihaiachim5299]

@11:45 As far as I can find on the internet they are just 105C …

[@waxore1142]

Regardless of the manufacturer's ability to obtain the exact high-quality components in this current economy it seems like a very well built device...

[@boots7859]

The tooling, molds and production lines were probably all part of the initial ramp-up expense made by the initial Western company. Why shut the plant down at 1700 hours and let it sit not making money?

[@waxore1142]

@@boots7859 was referring to the PC board components that are sourced out.

[@pauliexcluded1]

When was this made? I am curious is this was a covid supplier issue.

[@piertestini8963]

super like for this great analysis, very interesting. although a question arises: technically how does the inverter control in which direction the current flow on the AC side, from converter to grid or reverse, it's AC and therefore it doesn't seem simple to me.

[@maxtorque2277]

The grid has an impedance, as does the inverter. The output filters also act to form a bit of a buffer. The inverter simply controls it's applied voltage to either take current from the gird (output V lower than grid) or to push current onto the grid (Vout > Grid) The votlage is of course in the AC domian and hence the inverters waveform must comensurately match that of the grid, ie it must be "sync'd" to the grid frequency

[@matthewmarks6951]

Lots of complications: being "grid tied" it has to synchronise its AC output to the phase/frequency/voltage of the grid, and also instantly disconnect if the grid fails, for safety reasons (to avoid electrocuting anyone working on the supply). In terms of power direction, at its simplest I guess if the voltage on the DC bus is higher than the grid, it exports; otherwise, it imports.

[@otherphoenix1]

The simple answer is that they use closed loop sinusoidal current control performed digitally inside the dsp. If you can control current into a stiff voltage (the grid) you can control power. Another way to think about it is that it is a like buck converter when generating and becomes like a boost converter when charging. But with sinusoidal current.

[@matthewmarks6951]

@@otherphoenix1 Explains why it's difficult to find an explanation of how this works! What "sample rate" does the system need to operate at to control the output?

[@torginus]

I remember Deye having some controversy in Germany after they removed some parts required for local certification a few months after they started selling inverters there.

[@Juergen_Miessmer]

These micro-inverrters were allegedly grey imported by german vendors Not shure if it was deyes fault or if they had been built for other countries.

[@MatthewSuffidy]

So how do you know you have one of two bad relays so far to avoid having two? Amazing how all the power goes along a PCB though.

[@p_mouse8676]

Partial coating? How does that makes any sense?

[@EEVblog]

In theory the rest of the board has solder mask whcih does the same thing. You only need to cover the exposed solder connections.

[@mihaiachim5299]

The coating is expensive… all costs add up

[@mrnmrn1]

The most likely places for moisture induced leakage current and corrosion is where the lead spacing is fine, like SMD ICs and other SMD parts. When your board is full of connectors you can't submerge it into a pool of lacquer, so they hand paint it, and to save time and material, they paint it only where it is really neccessary.

[@otherphoenix1]

A lot of the coating is over parts that are either safety isolation or high voltage functional isolation. You don't need it everywhere if you want lowest cost.

[@jaro6985]

You can't conformal coat connectors or they will fail, so you selectively coat. And as mentioned coating blank spots on the board is a waste of money.

[@Creative-hi8my]

What role do Q14 and Q18 play in the circuit?

[@mihaiachim5299]

@ 31:53 everything is “ liver flavored” From the Garfield movie: when Jon comes home and tells Garfield that he brought him his favorite food with liver.. and Garfield replies: actually liver flavor😂 - this Inverter is just like that 😂 😅

[@dombaratadomba5029]

super informational tear down bro.... i like this video.....

[@fredfred2363]

One of the assembly houses I know in Shenzhen said there's an embargo on importing 'western ally' components. So Japanese, Taiwan, US and European made components are difficult to get without a big bribe.

[@pspkutube]

Awesome!!! Loved it. Thank you! I think the shift over to Chinese components might be a part of the decoupling that China has been doing in its industry to handle potential sanctions in future. Given that these are now quite well understood technologies, can't we open source these designs and have reference boards built using the world's leading components? This might help spread out manufacturing to various countries, and more importantly, accelerate the move to Green Energy. One more thing I would like to see is modularity in the design that enables easy troubleshooting and replacement. Today, debugging is a big problem in this space. This is critical for durable solar deployments that last many decades.

[@tonyh6309]

Where are the MPPT boost input filter capacitors? All I can see are a couple of grey film caps, maybe 5 to 10uFs tops? The Solar array current spec is, I believe, up to 13A or so; typically you'd design the inductor ripple current of a boost convertor to be around 30% of the average so perhaps 3A peak to peak. Unless the switching frequency is insanely high or the boost inductor is exceptionally large I wouldn't have expected 10uF to be anywhere near enough to avoid an amp or more of ripple through the entire PV array. That wouldn't be good for efficiency and terrible for EMI. Several hundred nH of wiring inductance is also going to make things interesting. Also where are the bus caps? Obviously I'm missing something here...

[@reversetransistor4129]

The power in power out ratio would be interesting to know, measured, of course.

[@bentheguru4986]

Not a bad looking setup compared to the flood of other cheap inverters with high failure rates.

[@mik13ST]

Don't know much about inverters but why does it use a flyback boost converter? Those are inefficient, right?

[@iamdarkyoshi]

Protip: Conformal coat dissolves with flux remover or acetone Also, soft potting compound can often be de-laminated with compressed air