Gaia/Masspower Grid Tie Inverter Teardown
I've been using this inverter for a while now and the 3 month warranty has passed so I can now tear it down.
This specific inverter is branded as "Gaia" but looks similar to other chinese made inverters and this one is rated at 500W with a 22V to 60V PV input with MPPT.
The inverter exterior. The graphics is one big sticker on top of the heatsink.
The DC input side.
Other models usually have one red LED on the left and three green LEDs to the right. This one only has one Bi-color LED which I only see blink red when there is no PV input. I have not seen it light up green.
AC input/output side.
Inverter nameplate underneath.
Four screws hold the bottom plate and once those are out, we get an overview of the main board.
Lots of silicone and a big foam rubber block on top of two large electrolytic caps.
Interestingly, this inverter is made by Masspower which makes a lot of small grid tie inverters.
The PCB is marked as SUN500G version 3.
Typed the link in my browser and this is where it went.
A small sub board is seen here which is basically a small TNY266 based offline switchmode power supply.
This powers the control circuitry from the mains 240V. This is what is drawing about 1.5W from the mains when there is no PV input.
The white wire carries the full 240VAC mains as input to the SMPS. If you noticed, in the board overview pic above, this thin wire snakes around all over the entire board without any support not to mention it is rated at 300V max. I can see a fire hazard in the future. This needs to be replaced with adequately rated wire!
Some large output filtering inductors for EMI suppression.
The small red round thing to the side is a 6.5A fuse.
And there is the other end of the white wire!
There is a relay that is jumped over. Not really sure what it is for but it appears to switch the AC mains connection as an auto select for 115/230VAC.
These four devices are the main switches for the DC-DC converters. Two for each transformer.
The lug bolted to the heatsink is a thermistor for temperature sensing (fan control and over temp protection).
These are IXYS IFP110N15T2 devices which are rated at 110A, 150V 13mOhm.
These two devices are STmicro STP75NF75 in parallel which is probably used as an input polarity protection by the looks on how they are wired.
These are 75V 80A 9.5mOhm devices.
On the other side, there are a bunch more parts!
The four on the left are FDP18N50 devices which appear to be wired as an H-bridge and is used to convert the HV DC to AC.
The rest are two pin devices which are diodes. These are marked as "NFC47G U860" which appears to be a variation of an MUR860 diode which are 8A 600V parts.
Hmmm... A crystal oscillator. There must be a microcontroller underneath.
And then there is this tab. Which basically contacts a pad on the PCB which is a ground connection for the chassis.
Removing all the brackets that hold the parts to the heatsink, we get to see the underside of the main PCB.
I see a current shunt resistor. This measures the PV input current going into the DC-DC converters.
Main uC. Atmel Mega88PA chip.
This appears to be doing the housekeeping duties.
Big SOIC chip is a UCC3806 low power dual output PWM current mode controller.
Small soic chip is an LT1013 dual precision op amp.
This SOIC chip is a FAN3224C dual gate driver IC.
It appears that the driver circuitry for the relays are populated but the relays themselves aren't installed.
Here is another one for the other relay seen earlier.
Here is a DC side relay that is not installed.
It is wired to switch the PV input positive.
There is also an internal 30A fuse for the PV input. These are rated at 32V so it may or may not cause a fire when it blows as it is not adequately rated but it is better than nothing.
And the AC side relay that is also not installed.
The transformers are labeled for 20V to 60V DC input.
Hmmm, where did this lug come off? I don't remember taking it off.
Ah, there it is. It was wedged between the end plates and the heatsink. This SHOULD have its own screw with a star washer to the heatsink not just wedged in like that.
I have not taken efficiency tests yet but online reviews mostly put it with about 20% loss so it is about 80% efficient. Not the best but not the worst also. I am using it with two 160W PV modules in series so that should be about 320W peak. Best I have seen is about 212W going into the mains but most of the time it pushes around 180W. It does get pretty warm and the fan is temperature controlled. I will do more efficiency tests when the sun is predictable (there are scattered clouds this past week).
09 Jan 2016:
A few months ago, the inverter would sometimes work, sometimes, be off for no reason. There is no specific pattern so I just washed the entire board with IPA and it seemed to have fixed it. This morning, it was dead again so I unplugged it, plugged it in again and still dead. A few minutes later, it was running so I left it alone. Tonight, I looked at it and it was dead again. It should have a red blinking LED when there is no PV input but it was still consuming 1.2W so I previously assumed that the internal aux supply is working. A dead red LED says otherwise.
I looked at the auxiliary powersupply which takes in mains 220V and puts out around 12V or so for the logic circuitry.
Plugged it to 220V and it is dead.
The soldering is crappy and looks to be hand soldered and the flux has not been cleaned.
It is a simple offline switcher based on the TNY266PN controller chip.
Output voltage regulation is done via a zener diode. Not even a TL431!
The white connector is unused. It does not have a feedback loop assumed to be used for a fan (later testing showed it to put out 11.7V unloaded)
Here is a closer shot of the soldering work.
I resoldered all the joints and washed it thoroughly with IPA.
Powered it up and it showed 10.2V on the output!
The main board is labeled with "12V" but it seems to be not critical.
Red blinking LED is back! Let's see if it works when the sun comes out.
This was bought around 21 Jan 2015 (date on warranty sticker) so it is almost 1 year old, and I already had to fix it twice.
Page created and copyright R.Quan © 29 May 2015.