Limiter inverter with RS485 load setting


Active member
Mar 7, 2018
Bought one of these, which turned out to be a little bit different to what I was expecting.


The RS485 port recieves a data stream, which is the target wattage to output from the unit and that's it....

The monitor unit is a CT (upto 30A) and also sends out an RS485 stream with the current measured wattage (no sense of direction / import / export) so you then have to wire it up before the load metering (CT) location, which is a pain.

Where it can come in very useful is because of the RS485 you can then put the inverter wherever you want and then control the output directly.

What I have done so far is setup the monitor on a separate fuse box (which contains the house sockets circuits and kitchen) and then used a couple of RS485 to ethernet adapters to then allow me to put the inverter nearly 50ft away at the bottom of the garden where the battery pack is.

The next step is to make my own data feed up to control the inverter directly.

Protocol specification :
Data rate @ 4800bps, 8 data, 1 stop
Packet size : 8 Bytes
1 : 24
2 : 56
3 : 00
4 : 21
5 : xa (2 byte watts as short integer xaxb)
6 : xb
7 : 80 (hex / spacer)
8 : checksum

Still neeed to figure out the checksum (xor / sum / ?) and what the other bytes really mean or do, if they actually do anything other than hearder, trailer, checksum. The unit can be configured for upto 5 units so one of the bytes will be for this reason (load divider).

Anyhow, hopefully this is of use.
Brief update. Unit has been running for a few weeks and seems to be ok, although the meter reading with some appliances fluctuates quite a bit so guessing the reading is an instantanious value (not average) at some point in the sine wave so any noise/choppy waveform will create fluctuations in the output that are not necessarily needed.... unless you have a separate meter feed to control the output rather than the basic meter that comes with the unit. Real shame they do not have directional metering.

There seems to be 2 versions, listed as 1000W and 1200W, however the only difference seems to be when running on battery the 1000W unit can "sustain" 800W and the 1200W can sustain 900W (go figure as to how the ratings are derived...). I have a couple of the 1200W units due to arrive and will see how they work on a separate circuit and compared to the 1000W unit. Suspecting they are identical and just typical Chinese rating confusion / nonsense, hopefully not.

Internally there does appear to be space for additional components in the 1000W unit, so the 1200W units may have these added and will find out when I get the 1200W units to have a look at and compare.
Have a URL for it?
I wish I could use these legally in Australia - they are hit and miss on their reviews but i guess if I got a 2000w unit and only run it at max 1000 ish watts it should last ok.
These are the units (check for pricing from different sellers as they are all a bit random)

They are only 1000W or 1200W rated units, so not the GTIL units that have been the choice for many and which is what I was originally planning to use. Will clarify what the difference really is soon.

The main benefit I saw with these was the simple RS485(232) control option, give it a number and it changes the output. Runs around 4Hz on output update rate.

Found out last night the output appears to start to de-rate below 47V input voltage.

In terms of grid compliance for connection they are problematic (no harmonics certification test for use) however the critical delay start timing can be changed to be compliant and this is the what I see as the main danger presented by any grid connected inverter connecting too quickly to a restored grid supply. Harmonics, add a separate external filter. Whole sepatate discussion..

My longer term intention is to use them off-grid in conjunction with a separate off-grid inverter for incremental sustained inverter capacity.
Last edited by a moderator:
I bought/used the 1000W version of this in 2018 - for about 4 months with no problem. I love the settings (battery voltage on/off, power limits) as I was too far away from main panel to use the clamp/limiter. I stopped using it in 2019 because its not ETL/UL listed - whereas all my other components (including the Inverter) are.

How do you feel about ETL/UL? - do you worry about it as a risk at all?
This is my view and not necessarily what should be followed as I have a slightly different arrangement...

Like with all laws/legal isses they are usually implemented to preven abuse of a system OR sometimes introduced as a means of creating competitive advantages in the guise of 'higher standards' (blocking imports for example).

Testing standrds seem to cover 3 elements (I'm separating outright electrical build quality here, which is a little easier to see) :
1. Anti-islanding (will your 800W inverter try to power the neighbourhood for longer than it takes for the FET's to blow magic smoke or will it reconnect and export far too quickly if linesmen are fault finding for example)
2. Harmoniocs (mainly to prevent square wave or partial wave inverters from introducing resonant issues and destroying other appliances.... typical HF units with reasonable filtering are zero issue, unless they are HF broadcasting with a really bad desing)
3. Stability (will your interter cut out if the grid becomes unstable, like if it was connected to a generator running out of fuel - this is typicallly an issue for large power stations as a phased/steepped frequency stability measure, not an issue for the home/kW size)

Voltage levels are a given basic for any equipment, so if they don't show or indicate the on/off limits then I would avoid them for example. Basic lack of control.

If your equipment is designed and operates within the required standards, just because it does not have a testing label/stamp of approval it does not mean it is therefore dangerous to the system just because it does not have a $ XX,000 stamp of :

This is also where the laws are intoduced because the few who produce real cr*p that destroys it for the many.

Summary : end of the day it's your own risk (and risk to others), just like when speeding in a car as every motorist does and it is therefore the balance of what risk is acceptable (Boeing 737).
So, after installing and a few more days of operation I now have 3 of these units active and balancing out 2 distribution boards.

The lighting board (also the shed) has one 1000W unit matching the power flow and then a couple of 1200W units, which are then matching the power from the sockets/kitchen.

The metering point in the house has two of the current clamps around the 16mm cables feeding each board (the current clamps need the plastic trimming so they will fit over the cable) and then the output from each meter is fed into an RS485 to thernet adapter, whcih is then plugged into the network. The third current clamp (each unit came with a CT and RS485 output box) is then connected on the main feeder so I can have a rough indication as to the magnitude of the residual imbalance (good way to get an idea on the CT accuracy 1+1=2.2).

The battery location in the shed then has two ethernet to RS485 adapters with the output from one going into a 1000W unit and the other going in paralle to the two 1200W units.

The "1000W" unit will only give out 800W when running off battery and recieving RS485 output level.

The two "1200W" units will only give out 900W each when running off battery and recieving RS485 output level, so 1800W in total.

I think (guess) that they derate the units when operating on battery because when they are running off solar the input power is limited by default so any fault issues will not cause significant stresses, while a battery with 100's of amps capability will quite hapily blow the PCB traces off the board. Sort of an easy addition of hopeful safety margin to faults... that's my guess.

Overall, seem to work ok, the current clamp accuracy may be a bit suspect (less accurate than I would like).

Will add some pics.

Separate note, the internal fans come on at 40C and then only relatively briefly (with 25C ambient), although I have a separate small fan (few W) blowing on the units 24x7 just to give a little more than convection flow for normal cooling. Thought is that by reducing the temperature by say just 5C the operational life of the capacitors will be extended quite a bit.... When really hot days arrive I also turn off some of the solar input to reduce the temperature on the MPPT controllers... fortunrately or unfortunately for us in the UK we don't get many 30C plus days.
completelycharged said:
Protocol specification :
Data rate @ 4800bps, 8 data, 1 stop
Packet size : 8 Bytes
1 : 24
2 : 56
3 : 00
4 : 21
5 : xa (2 byte watts as short integer xaxb)
6 : xb
7 : 80 (hex / spacer)
8 : checksum

if you understand what those bytes mean, it would be very nice (and maybe i can help) to write a python software to give data to this grid tie.

and maybe throw away that clamp and use a rs485 panel meter like eastron one.
Will try and get the full protocol out from the manufacturer (some hope...) because the units appear to be programmable for constant voltage, constant current and constant power (default)...

Suspect the 24/56 is a standard hearder for the device.
00 - single unit active (increases by 1 per unit - acts as a divisor for the power measurement)
21 - no idea

Or could be just simple 2 byte word packet.
Unit ID
Unknown/settings (21 = Constant power mode as a guess ??)
Target setting in Watts/Volts/mA

They are interesting units and possibly one of the most accessible (output setting options) cheap inverter uinits on the market at the moment, shame they lack official grid certification in virtually every country in the world....

Have not had chance to look at this yet, awaiting response from Soyo..

The brand is listed as soy-source or the primary very slow website is

The single unit has had 381kWh though to date
The two parallel units have had 142kWh each through them, so 284kWh.
The last (oldest) unit has had 701kWh though and is about a year old.

All in 1,366kWh though the small inverters.
No reply at all from the seller / company... nice.. in spite of that.......

Working !!

Checksum is 264 - byte5 - byte6 ..... which gives byte8
Why 264, still to work out.

Periodically (as random as the lottery) the units seem to send out a 15 byte packet, which contains the battery voltage and other stats like this : 23 01 01 00 00 01 DB 00 A1 00 DD 64 02 CA 01

23 header ?
01 header ?
01 header ?
00 header ?
00 header ?
01 battery voltage high byte
DB battery voltage low byte
00 amps high byte
A1 amps low byte
00 scooby no idea
DD AC voltage low byte (to test further due to issue with reading)
64 best guess is Hz x 2, i.e. 50Hz = 64 hex = 100
02 guessing this is 2 for 20C offset for temperature.... who the f writes code like this ?
CA internal temperature (magic number format of C x 10 above 20C rounded up, i.e. 103 = 10.3 rounded to 11 + 20 = 31C)
01 some sort of status bitmask

Divide battery voltage number by 10 to get actual value to 1 decimal place

Separate notes :
The fans in the 900W units turn on around 3-4C before the 800W units.
600W max output at 47.5V terminal input voltage (low voltage derating)
Multiple units on an RS485 bus, you have no idea which unit sent the 15 byte data packet
Multiple uinits on the same meter, the update frequency of the meter reading slows down (to confirm with test)

Checksum calculation (with 8 byte array, zero first value) :
invBytes(7) = (264 - invBytes(4) - invBytes(5)) And 255

4 Units operating, 3 of them under control and one pre-set at 200W output (assuming 30W loss)

Loads are added up, add in 30W loss and subtract off the 200W fixed inverter output. This result is then divided by 3 and sent to the inverters as the output setting and now shares the load evenly between all 3 units so I should now be able to net off 2600W or more (4th unit can go to 750W, so absolute max 3150W).

Few changes to do, but will see how it runs for the next few days.

The 4th unit is set close to the lowest overnight loading so overnight all but the 4th unit effectively turn off and make most use out of the adjustable output.
Bit more of an update and additional discoveries.

Discovery 1 - the inverters add up the kWh energy throughput purely based on the command value recieved even if this value is impossible to actually output. I wondered why I was seeing differences in the readings over a few days until I had a look at what the inverters thought they were doing. End result, a very poor unrelaible kWh reading.

Discovery 2 - the protocol has no real error checking and any errors are acted upon. Send the unit a packet with a command to output 80kW, fine, max output. 9MW, fine.

The units could do with basic error checking on the data packets, however they still seem to otherwise run ok, so a fix that may turn up in the future. Maybe in a year these units will be far better than the GTIL units for DIY builds rather than comparable.

Running all (meter and inverter) of the RS485 to Ethernet adapters in server mode avoids any hassles with firewalls from wherever you run your code to control them.

Setup the serial ports as 4800 baud,
input buffer for the meter of 8 bytes with a timout of 50mS (if no byte within 50mS then flush the buffer if you don't have 8 bytes)
input buffer for the inverter of 15 bytes with a timout of 50mS (if no byte within 50mS then flush the buffer if you don't have 15 bytes)

You can always do sense checking in the meter reading values (or just use meter reading value from another source) before sending a command to the inverters.

Efficiency is not as bad as I first thought, but still worse than using my 6kW unit.
Hi! I'm also working on the RS485 communication protocol, thank you for figuring out the CRC method!

Can confirm everything is the same. Will post any advances.
Manufacturer replied "Currently we do not have the RS485 protocol." So we are on our own.

I am testing a "man in the middle" setup with the inverter, in that I read what the sensor outputs, modify it, and instantly send the new value to the inverter. I noticed that the inverter is less efficient when it maxes out, so I want to cap the maximum power requested depending on the available power. The inverter is also less efficient when run at low power (<30W), and this approach should also allow to zero the power output when the demand is very low.

What do you think?
The efficency I noticed was a bit bad at maximum and very low loading so I have set mine so that I say below about 70% unless I have excess power available. Similarly if overnight is going to be less than 60W per unit I will switch one or two units off to raise the output of the othewr units that stay online.

The MIM approach is what I have done for the moment until I get my metering setup more comprehensively (I have had the parts for nearly a year now and not got around to installing them)

I'm still curious as to the other control options the units appear to offer as the know protocol so far I think is only a fraction as to what is implemented.
This thread is really interesting. I've got a few panels hooked up to a Victron MPPT with some Li-ion cells. This is also connected to a Raspberry Pi with the Victron Venus OS running on for remote monitoring. I'm getting the ET112 energy meter being delivered next week that's going on the main distribution board for the house and talk to the Pi to report the usage.

There is facility to write your own scripting on the RPi. Would be ace to get the Pi to send RS485 data to this inverter to limit the output to the actual household use for zero feed-in to the grid.
One of the sellers from Aliexpress replied "all items from Soyosource are not available now as they don't work now".

Did anyone have problems with their inverter so far?