Jump to content

Lower cost hybrid Inverters(Y&h) , how are they different from Off-Grid versions


luk88

Recommended Posts

Hi from a new member ūüôā

 

I've been using Solar for quite a while and I have some electronics repair experience. I initially wanted to build a pure off-grid system in addition to my small existing on-grid setup. 

 

However, one of the off-grid inverters I bought (Y&h 4.8kW 24v) just to power lights, security system and other small loads has grid tie capability. I tested it and it works pretty well for me. 

So now I wonder, what is the electrical (or design) difference between off-grid all-in-one inverters that cannot do grid export and those that can. Is it just a software feature? 

 

Both types have the one AC inverter for the output (during grid tie it's in bypass-line mode). Both can sync to the grid (as both have the 10ms switchover UPS option). 

 

So what functional blocks a typical off-grid inverter lacks to do grid export? By a typical off-grid I mean something like EASUN smh-ii 7k (Axpert MKS ii?). 

Also, Y&h claim to do their own manufacturing. But main boards look very similar to Voltronic products. Does anyone know if they add the grid export by using custom firmware only? 

 

As far as I can tell theirs (Y&h) are the only cheap battery inverters that can do grid export and limit the exported current at the same time. They have new options 37 (enable hybrid mode) and 38 (limit grid tie current) so with some clever programming it is possible to achieve zero export without a CT. 

 

I looked at EASUN igrid range and although they can do grid export there is no way to limit. 

 

Another notable thing is none of the Y&h inverters can be parallel 'Ed. I'm also interested if this can be changed, but perhaps it's for another thread. 

 

Link to comment
Share on other sites

7 hours ago, luk88 said:

Hi from a new member ūüôā

 

I've been using Solar for quite a while and I have some electronics repair experience. I initially wanted to build a pure off-grid system in addition to my small existing on-grid setup. 

 

However, one of the off-grid inverters I bought (Y&h 4.8kW 24v) just to power lights, security system and other small loads has grid tie capability. I tested it and it works pretty well for me. 

So now I wonder, what is the electrical (or design) difference between off-grid all-in-one inverters that cannot do grid export and those that can. Is it just a software feature? 

 

Both types have the one AC inverter for the output (during grid tie it's in bypass-line mode). Both can sync to the grid (as both have the 10ms switchover UPS option). 

 

So what functional blocks a typical off-grid inverter lacks to do grid export? By a typical off-grid I mean something like EASUN smh-ii 7k (Axpert MKS ii?). 

Also, Y&h claim to do their own manufacturing. But main boards look very similar to Voltronic products. Does anyone know if they add the grid export by using custom firmware only? 

 

As far as I can tell theirs (Y&h) are the only cheap battery inverters that can do grid export and limit the exported current at the same time. They have new options 37 (enable hybrid mode) and 38 (limit grid tie current) so with some clever programming it is possible to achieve zero export without a CT. 

 

I looked at EASUN igrid range and although they can do grid export there is no way to limit. 

 

Another notable thing is none of the Y&h inverters can be parallel 'Ed. I'm also interested if this can be changed, but perhaps it's for another thread. 

 

An utility approved grid tie will for now use a CT or relay. It has the synch function to the grid. Also it MUST switch off export if the grid fails. 

What you mention as new function does normally not work as well as a CT is these inverters normally rely on exporting a % to the grid but could/would trip a prepaid power meter. That's where the CT or relay comes in to measure and with comms tell the inverter to throttle. 

The typical hybrid that can export has its shortcomings. Those with CT also have the non essential loads whereby excess PV can be exported to these high loads but without exporting to grid due to the power meter tampering in prepaid meters. 

Link to comment
Share on other sites

13 hours ago, Scorp007 said:

An utility approved grid tie

But who said anything about "utility approved" ūüėĀ

As far as I know none of the Y&h, iGrid, SMX etc are utility approved. But it's not stopping people using them. 

 

It is not ideal, I agree completely, but in this thread I wanted to find about about the technical differences in the design between a non-grid export enabled all in one inverter and one with such function. 

 

Of course you're right you should not use a non-utility approved device to push to the grid, but people have different needs. For example one may have long running loads and the particular Y&h inverter has an option to limit current to certain amount of amps. So as long as the load is slightly higher there is no actual pushing to grid. But the feature is still called grid export, very confusing for many. 

13 hours ago, Scorp007 said:

will for now use a CT or relay.

This is ideal, but as mentioned being able to set max amps is useful too. Not approved for export, but useful. 

13 hours ago, Scorp007 said:

 

It has the synch function to the grid.

Both export enabled and ones without export if they advertise 10ms switchover and UPS mode are able to sync to the grid. 10ms is half the grid waveform so to achieve under 10ms switchover they have to sync. 

 

Perhaps the ones with grid export "sync better"? More precisely? 

13 hours ago, Scorp007 said:

Also it MUST switch off export if the grid fails. 

Yes, this is very important. The models from Y&h I speak of do have "is landing protection" built in. 

13 hours ago, Scorp007 said:

What you mention as new function does normally not work as well as a CT is these inverters normally rely on exporting a % to the grid but could/would trip a prepaid power meter. That's where the CT or relay comes in to measure and with comms tell the inverter to throttle. 

Now we talk about it's use rather than the design of the electronics. It is a separate matter, but yes I agree it doesn't work as good as a grid export inverter with a CT or one that communicates with a smart meter (what you called a relay I guess), for example Deye inverters etc. However a Deye inverter costs 4x the price for the same power so for some people this cheaper grid export is useful too. They don't really have to export to the grid if there is a configurable amps limit. They may have a load exceeding it. 

13 hours ago, Scorp007 said:

The typical hybrid that can export has its shortcomings. Those with CT also have the non essential loads whereby excess PV can be exported to these high loads but without exporting to grid due to the power meter tampering in prepaid meters. 

Indeed. The kinf of "cheap Hybrid" I talk of has one big shortcoming. As they use the same inverter parts that would normally be supplying loads to do grid export it cannot do grid export and draw from the battery at tye same time. What do I mean? Surely we don't want to export excess pv and use the battery at the same time, right? Imagine the following situation. 

 

Your battery is full, the sun is shining and you have a 30A compressor running continously to run some industrial process on the AC in side of the inverter and it's infeasible to change the wiring or it's so powerfull the inverter couldn't start it, but grid export makes it 90% cheaper to run. So you use the grid export . Your "cheap hybrid" also has some computers on the load side. Then a cloud comes and blocks the sun. It stops exporting, this is normal, your compressor runs 100% from the grid, but the inverter also powers the AC out loads (the computers in my example) from the grid despite the full battery. Why? Because you have to switch to SUB to enable grid export. 

This is how they work. 

So my main question is, is this just a software feature? Can any all in one inverter with 10ms switchover/ups do such type of grid export in principle? (if it had correct software) 

Link to comment
Share on other sites

32 minutes ago, luk88 said:

But who said anything about "utility approved" ūüėĀ

As far as I know none of the Y&h, iGrid, SMX etc are utility approved. But it's not stopping people using them. 

 

It is not ideal, I agree completely, but in this thread I wanted to find about about the technical differences in the design between a non-grid export enabled all in one inverter and one with such function. 

 

Of course you're right you should not use a non-utility approved device to push to the grid, but people have different needs. For example one may have long running loads and the particular Y&h inverter has an option to limit current to certain amount of amps. So as long as the load is slightly higher there is no actual pushing to grid. But the feature is still called grid export, very confusing for many. 

This is ideal, but as mentioned being able to set max amps is useful too. Not approved for export, but useful. 

Both export enabled and ones without export if they advertise 10ms switchover and UPS mode are able to sync to the grid. 10ms is half the grid waveform so to achieve under 10ms switchover they have to sync. 

 

Perhaps the ones with grid export "sync better"? More precisely? 

Yes, this is very important. The models from Y&h I speak of do have "is landing protection" built in. 

Now we talk about it's use rather than the design of the electronics. It is a separate matter, but yes I agree it doesn't work as good as a grid export inverter with a CT or one that communicates with a smart meter (what you called a relay I guess), for example Deye inverters etc. However a Deye inverter costs 4x the price for the same power so for some people this cheaper grid export is useful too. They don't really have to export to the grid if there is a configurable amps limit. They may have a load exceeding it. 

Indeed. The kinf of "cheap Hybrid" I talk of has one big shortcoming. As they use the same inverter parts that would normally be supplying loads to do grid export it cannot do grid export and draw from the battery at tye same time. What do I mean? Surely we don't want to export excess pv and use the battery at the same time, right? Imagine the following situation. 

 

Your battery is full, the sun is shining and you have a 30A compressor running continously to run some industrial process on the AC in side of the inverter and it's infeasible to change the wiring or it's so powerfull the inverter couldn't start it, but grid export makes it 90% cheaper to run. So you use the grid export . Your "cheap hybrid" also has some computers on the load side. Then a cloud comes and blocks the sun. It stops exporting, this is normal, your compressor runs 100% from the grid, but the inverter also powers the AC out loads (the computers in my example) from the grid despite the full battery. Why? Because you have to switch to SUB to enable grid export. 

This is how they work. 

So my main question is, is this just a software feature? Can any all in one inverter with 10ms switchover/ups do such type of grid export in principle? (if it had correct software) 

Points taken. Sorry I could not add to the actual question asked. ūüėÄ

Link to comment
Share on other sites

On 2024/06/23 at 10:33 AM, luk88 said:

So now I wonder, what is the electrical (or design) difference between off-grid all-in-one inverters that cannot do grid export and those that can. Is it just a software feature? 

That's a good question . Answer is YES , software .  Nowadays just about all commercial / residential  inverters  employ grid-tie topology , even if they are configured as off-grid only (like the Axperts etc). That excludes the smaller type inverters you get that is about 400W- 3kW  and those that are PWM AC type .  All Axperts from the MKS3 upwards are  grid-tied topology , it is ONLY the software that  prevents grid export.  All these inverters support ac/dc blending of power and that can only be entertained by grid-tied topology. The exception is the Axpert King , which converts grid ac  to dc and then blend that with the other dc sources , but these inverters are not as efficient as the bidirectional grid-tie ones , they are predominantly marketed where grid is dirty or where you require 0mS response where grid is lost.

image.png.686ce2aa3eeff15ffd990b6f376aa6cf.png

 

above is block schematic of the grid-tie technology . So this is the same for SunSynk on-grid or Axpert off-grid. The only differences as far as hardware  is where an inverter is certified/approved for on-grid export , that inverter has earth leakage built-in in order to detect PV leaks to earth (I believe the spec states no more than 300mA leak current). And of course the inverter support external means of power measuring in order to control export.

 

EDIT : The rules of engagement  for off-grid inverters is relatively simple . The software will not allow the DC power (point 3 in the diagram)  to exceed the AC load power  (point 2 in the diagram) at any time.  In practice though a small amount of export power slips thru the grid   when the load  power is suddenly reduced from say 2.5 kW to  300W (example kettle or hair-dryer switching off) . This is due to the response time of the control loop .

 

 

Edited by BritishRacingGreen
Link to comment
Share on other sites

Very interesting @BritishRacingGreen, thank you for answering. I wish one day someone makes an open source firmware for some of these inverters... 

I wonder if the source code is floating somewhere (not open source, but known to the people in the know). How else could a bunch of "manufacturers" use the same menus, same features etc? 

Link to comment
Share on other sites

2 hours ago, luk88 said:

I wish one day someone makes an open source firmware for some of these inverters..

Sadly I think that might never happen , well , given my age , at least not in my lifetime. If the inverter function is merely inverting DC to AC  ,  I think that's not impossible task. But the grid-tie algorithms required is real heavy lifting and its failure modes are unforgiving. Apart from all the high order digital filtering required by the DSP  to sync to grid in terms of magnitude , frequency and phase , you will have to generate high resolution PWM sine wave and able to make fine adjustments to phase angle in order to control both direction and  magnitude of power .  These days the main DSP also performs the MPPT tracking from first principles. To name just a few challenges .  But life is full of gurus and experts , who knows? Perhaps you are one of them.

2 hours ago, luk88 said:

I wonder if the source code is floating somewhere (not open source, but known to the people in the know). How else could a bunch of "manufacturers" use the same menus, same features etc? 

Not that I am aware of .  The closest you will currently get is what @Coulomb has done  supporting the Voltronic firmware.  Him and his colleague have reverse engineered the DSP firmware to native binary assembly language. And they have mastered the art and they have methodologies and developed tools  to reverse engineer the binaries , to the tune that they can bake in / patch in software fixes for certain manufacturer bugs , as well as add functionality. I am not aware of many people that can do this , it is that difficult , and it is THAT kind of magic!

I have toyed with an idea to fool  an Axpert to export without it being aware that it is exporting. Remember the current sensors readings are the fundamental  inputs to the DSP power control algorithms. For example there is a sensor in leg 2 (load)  and leg 3 (dc) in the diagram above. The Axpert has no sensor for path 1 (grid) . So we somehow add to the instrumentation amplifier  reading an offset that represents ratio metrically a power of 1000W. This I want to initially do in the analogue hardware domain , not software. Proof of Concept . Lets assume the actual load output is 2.5kW . But the new 'cooked' reading the DSP gets is 3.5kW . So it changes the ac phase angle  so much so as to deliver the required 3.5kW from DC , given we have excess PV power to do so. The DSP reaches 3.5kW on the dc side , but the actual load is only 2.5kW , so the 1000W is now pushed to the grid , and we are exporting!.   Or so my theory goes.

If that proof of concept works , we can approach someone like @Coulomb to support an unallocated can bus message identifier in the firmware , of which its payload contains the required offset that must be added to the load sensor ADC reading , this time in software . But that's also probably a huge task. Then you entertain an external embedded controller that can send can bus messages , and also read Modbus from  an energy meter , so we can prevent exporting to Eskom.

Alas , I started with this idea 18 months ago , and its still only living in my head , too much other priorities.

 

Edited by BritishRacingGreen
Link to comment
Share on other sites

11 hours ago, BritishRacingGreen said:

These days the main DSP also performs the MPPT tracking from first principles. To name just a few challenges .  But life is full of gurus and experts , who knows? Perhaps you are one of them.

I wish ūüėÖ

Very true(about the difficulty). And this is even without features like parallel operation. I read an article about control mechanisms in parallel inverters to figure out if one could parallel ones without such feature and there are things like circulating currents between two inverters etc to avoid which uses similar control to power factor correction. It's quite a bigger problem than just make sure you supply half the current.

11 hours ago, BritishRacingGreen said:

Not that I am aware of .  The closest you will currently get is what @Coulomb has done  supporting the Voltronic firmware.  Him and his colleague have reverse engineered the DSP firmware to native binary assembly language.

Did they publish some info about what cpu/dsp is it? What tool chain they use? What the fw image layout is? Are there some debugging tools onboard? (serial, JTAG etc) Any "entry" points like accessible spi bus for decoded program flash? One would need a spare inverter for sure, just to do some hacking... I actually have all the tools I might need (programmers, interfaces,a fast logic analyser etc). If the 10kW unit from Y&h lives to my expectations I'll be ordering another one to replace my current 4.2kW. So then I'll have the 4.2k as spare to play with.

Or are they tight lipped about the actual methods? I'd love to have a go, but reversing it from scratch is such a big task. No surprise someone who invested so much time would want to get dome return on investment. Still I hope for more open information even if it's just the minimum.


Now that you say about it I think that's exactly how "tweaking" by manufacturers is done. Unless they do get the source and a nice SDK, but if it existed it would've leaked by now, they are probably doing exact same type of hacking.

 

 

11 hours ago, BritishRacingGreen said:

And they have mastered the art and they have methodologies and developed tools  to reverse engineer the binaries , to the tune that they can bake in / patch in software fixes for certain manufacturer bugs , as well as add functionality. I am not aware of many people that can do this , it is that difficult , and it is THAT kind of magic!

I've dabbled at such work at various brief times. Mostly as a hobby to crack unknown Cctv cameras to do what I want, but these are simple devices running mostly known type of software (Linux with uboot). Inverters probably running some sort of real time OS might be orders of magnitude more difficult. There is also a possibility they don't have an RTOS and it's written in bare metal code. Depending on the cpu and tools(debugger, emulator) available it may be even better. Id love to find out more.

 

11 hours ago, BritishRacingGreen said:


I have toyed with an idea to fool  an Axpert to export without it being aware that it is exporting. Remember the current sensors readings are the fundamental  inputs to the DSP power control algorithms. For example there is a sensor in leg 2 (load)  and leg 3 (dc) in the diagram above. The Axpert has no sensor for path 1 (grid) . So we somehow add to the instrumentation amplifier  reading an offset that represents ratio metrically a power of 1000W. This I want to initially do in the analogue hardware domain , not software. Proof of Concept . Lets assume the actual load output is 2.5kW . But the new 'cooked' reading the DSP gets is 3.5kW . So it changes the ac phase angle  so much so as to deliver the required 3.5kW from DC , given we have excess PV power to do so. The DSP reaches 3.5kW on the dc side , but the actual load is only 2.5kW , so the 1000W is now pushed to the grid , and we are exporting!.   Or so my theory goes.

A cool idea. If you ever try please describe the results even if they are negative.

 

11 hours ago, BritishRacingGreen said:

Alas , I started with this idea 18 months ago , and its still only living in my head , too much other priorities.

 

I have the same problem. So much stuff to do, so little time!

Link to comment
Share on other sites

@Coulomb I have traced the PWM drive of the LV SCC circuit as best to my knowledge . This is from the large flat PCB version. The King PCB I have does not have device symbols but  it looks very much the same.

So I don't think my theory of a failsafe watchdog is applicable . The PWM signals is merely gated on or off by a separate GPIO of the CPU . It may be a convenient / helper function to turn PWM  off when CPU want to disable PWM, instead of reconfiguration of internal PWM subsystem?

image.png.7f4142afdb7aae87214db026e6d020fa.png

 

Edited by BritishRacingGreen
Link to comment
Share on other sites

8 minutes ago, BritishRacingGreen said:

@Coulomb I have traced the PWM drive of the LV SCC circuit as best to my knowledge . This is from the large flat PCB version. The King PCB I have does not have device symbols but  it looks very much the same.

So I don't think my theory of a failsafe watchdog is applicable . The PWM signals is merely gated on or off by a separate GPIO of the CPU . It may be a convenient / helper function to turn PWM  off when CPU want to disable PWM, instead of reconfiguration of internal PWM subsystem?

image.png.7f4142afdb7aae87214db026e6d020fa.png

image.png

i am sorry , this is exactly the wrong thread for post , should have gone in my 'journey' thread. 

Link to comment
Share on other sites

16 minutes ago, luk88 said:

Did they publish some info about what cpu/dsp is it? What tool chain they use? What the fw image layout is? Are there some debugging tools onboard? (serial, JTAG etc) Any "entry" points like accessible spi bus for decoded program flash?

No they have not as far as I know , and I think it has very much to do with a third party's need to understand their methodologies. BUT , @Coulomb is with us to support community , he has a very professional and selfless  outlook on  community driven  product support . Yes they have software tools and debugging support . I will recommend that you send Coulomb a Private Message and declare your expertise and interest. 

Link to comment
Share on other sites

28 minutes ago, luk88 said:

Or are they tight lipped about the actual methods? I'd love to have a go, but reversing it from scratch is such a big task. No surprise someone who invested so much time would want to get dome return on investment. Still I hope for more open information even if it's just the minimum.

 

No they they are not tight lipped . I can tell you now that I am only 3 years in this business of solar energy and if it was not for the extreme amount of time he has spent with me , I would have given up already . Just make contact with him , I infer from your posts you know your embedded stuff .

Edited by BritishRacingGreen
Link to comment
Share on other sites

20 hours ago, luk88 said:

Did they publish some info about what cpu/dsp is it?

Some info on my JTAG post.

Quote

What tool chain they use?

A very brief summary in this recent post. I've written more elsewhere.

Quote

What the fw image layout is?

It's just one big block of code, except for one flash sector (sector A, with the highest addresses in C2000 DSP) reserved for the bootloader.

C2000 firmware starts at 0x3D8000 for firmwares that support paralleling, and 0x3E8000 for the VM firmwares that don't.

Quote

Are there some debugging tools onboard? (serial, JTAG etc)

There is a JTAG port for all firmwares; for the C2000 DSPs it's the 14 pin header with pin 6 cut off. But you can't use it unless you wipe out the entire firmware, including the bootstrap loader, or you know the 128-bit password. ARM display firmware is not password protected at this stage.

There is one large test command built in, but they usually don't give much of a clue as to what the code is doing.

Some control boards have a USB connector that doesn't seem to have any purpose in normal operation. I have not investigated.

Quote

Any "entry" points like accessible spi bus for decoded program flash?

The flash is all internal, only EEPROM and sometimes an "I/O expander" are on SPI buses. I have a very limited set of tricks to make life slightly easier when analysing firmware. I don't publish many of these, not to keep the glory to myself, but because when they become known, the developers can usually do something simple to work around them, and then the advantage is lost forever.

20 hours ago, luk88 said:

There is also a possibility they don't have an RTOS and it's written in bare metal code.

That's pretty much it. There is a set of about 6-8 tasks that timeshare; it's a standard TI "operating system". They use something very similar in the ARM display/comms code. There is a set of events that can be sent from task to task. Event 0 is the timer tick; the period of the tick can vary between tasks. Higher priority tasks get faster ticks. There are two huge interrupt routines that do the analog to digital conversion reading and other time critical tasks. 

On 2024/06/25 at 2:17 AM, luk88 said:

I wonder if the source code is floating somewhere (not open source, but known to the people in the know).

I do believe that there is leaked source code about. The fact that there are so many Axpert VM II clones about, years after Voltronic stopped making them, suggests that the source code for that firmware is circulating amongst the clone manufacturers.

 

Edited by Coulomb
Link to comment
Share on other sites

20 hours ago, BritishRacingGreen said:

The PWM signals is merely gated on or off by a separate GPIO of the CPU . It may be a convenient / helper function to turn PWM  off when CPU want to disable PWM, instead of reconfiguration of internal PWM subsystem?

That seems plausible. In DSP code, when they want to turn off the PWM, there is a block of boiler plate code (possibly a macro or maybe a function that is frequently inlined), and it's a dozen or so instructions long. I assume that flash space and processor speed is limited in those 8-bit MPUs, so this hardware solution may be justified.

Thanks for that tracing. I should figure out where that IO pin is used in the firmware, and document it.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

√ó
√ó
  • Create New...