I forgot one more important hint: The two peaks (Inverter 3, no patches) in yesterday's charts are exactly at 9:00. On this day and time, the inverter switched from grid to battery mode. The switchover was no longer active today.

Edited December 5, 20223 yr by Volker

15 hours ago, Volker said:

The two peaks (Inverter 3, no patches) in yesterday's charts are exactly at 9:00. On this day and time, the inverter switched from grid to battery mode.

I can't think why switching modes makes any difference, except that it might somehow "reset" the power tracking code somehow.

My patches are obviously inferior, but clearly even Georg's aren't doing all that well in these conditions. His tends to spend a lot of time at 120 V, mine at 90 V. Clearly, once it gets into that state, it finds it hard to get out, unless or until there is a burst of PV power.

I'm tempted to try a patch with Georg's patches, plus every minute or so forcing the target voltage to a high value, perhaps the maximum MPPT value. The tracker seems to have no trouble going down in voltage, so it should not take long to get to a sensible voltage setting. This would be a very crude form of "scanning" for the MPPT.

I might go hunting for the code that might be resetting the MPPTs when it changes to battery mode. That might be appropriate code to call or to copy and run it every minute.

[ Edit: Added "in these conditions". ]

Edited December 6, 20223 yr by Coulomb

On 2022/12/06 at 11:11 AM, Coulomb said:

I can't think why switching modes makes any difference, except that it might somehow "reset" the power tracking code somehow.

My patches are obviously inferior, but clearly even Georg's aren't doing all that well in these conditions. His tends to spend a lot of time at 120 V, mine at 90 V. Clearly, once it gets into that state, it finds it hard to get out, unless or until there is a burst of PV power.

I'm tempted to try a patch with Georg's patches, plus every minute or so forcing the target voltage to a high value, perhaps the maximum MPPT value. The tracker seems to have no trouble going down in voltage, so it should not take long to get to a sensible voltage setting. This would be a very crude form of "scanning" for the MPPT.

I might go hunting for the code that might be resetting the MPPTs when it changes to battery mode. That might be appropriate code to call or to copy and run it every minute.

[ Edit: Added "in these conditions". ]

The power tracking can be restarted by the following methods:
- Inverter is switched from grid to battery mode.
- Inverter changes the power balance strategy.
- The PV field is disconnected by hand for a moment and then reconnected.
I had already considered an hourly switchover as an emergency solution...
The power then actually always increases by a factor of 2 to 4. If the PV current is too low, however, the reading quickly falls back to the previous value.

On 6 November it was overcast again, but there was a little more light.

On 7 November at 12:41 I disconnected and reconnected the PV fields of Inverter 3. In the diagram "2022-12-07 Inverter 1 3 4 5.PNG" all PV fields are shown and the time is marked (orange is the field with south orientation). The day started with snow on all PV fields, but it disappeared quickly. It was often sunny until 12:15, then it was grey like the previous days.

Georg's firmware starts early (this is its biggest advantage), but it does not use the power well with small brightness fluctuations (see "2022-12-07 Inverter 1 and 3.PNG" time 12:41+.

OK, I have an idea. I think that setting the direction to zero when the current is low is a bad idea. When you're already at minimum PV voltage, then if the direction is forced to zero, the algorithm heads to lower voltage, but because it's already at minimum voltage, the voltage stays the same, and the current is probably constant, so it keeps heading down in voltage. If the direction wasn't forced, then the first action would be to reverse direction, so the voltage can go up even if the power stays the same.

The problem with this idea is that with either of the patches, the current threshold is quite low, so I think it might almost never end up on the "low current" area of the code where I propose to make the change.

Still, it's an idea. Maybe I'll run up that plus the "reset every minute" idea into another patch. Busy for a few hours though.

This is patched firmware version 356.05, based on Axpert Max II 10 kW firmware version 56.05. It has the patches of version 156.05 (current threshold for MPPT1 and MPPT2 reduced from 0.50A to 0.05A), plus:

* Every 60 seconds, the MPPTs are "reset" by forcing the direction to 0 (so that reducing voltage should increase power), and the solar reference voltage is set to 450.0 V (the highest MPPT voltage). This will hopefully effectively perform an MPPT scan, although the scan will stop at the first "hill" found.

* For MPPT1 only, when in low current mode, the direction flag is no longer forced to zero. I suspect that this is contributing to the "stuck at 90V" problem.

* The firmware version variable is set to 35 605 (representing 356.05), and because this is more than 32 767, it will be interpreted as negative by any code that treats it as a signed integer. Hopefully there is no such code, but this patch will tell us for sure. Even if the version comes out crazy, it should not effect actual inverter operation.

NONE of Georg's other manOG-10 DSP Reflash_356.05 patched.zipy patches are implemented. The premature float bug is patched as before.

Use at you own risk.

 

I really like what's going on here!
Coulomb fixed my axpert II 8k inverter firmware. But it can definitely be improved.
I would like to help you at work! But unfortunately I don't know assembly language. And the inverter has two stings of the same size but with different orientations.
I have a video of the factory firmware 90v getting stuck. After SBU-SUB switching, the PV voltage jumps, but within a short time it gets stuck at 90v again.
I hope I can help.

 

On 2022/12/09 at 12:58 PM, Coulomb said:

This is patched firmware version 356.05, based on Axpert Max II 10 kW firmware version 56.05. It has the patches of version 156.05 (current threshold for MPPT1 and MPPT2 reduced from 0.50A to 0.05A), plus:

* Every 60 seconds, the MPPTs are "reset" by forcing the direction to 0 (so that reducing voltage should increase power), and the solar reference voltage is set to 450.0 V (the highest MPPT voltage). This will hopefully effectively perform an MPPT scan, although the scan will stop at the first "hill" found.

* For MPPT1 only, when in low current mode, the direction flag is no longer forced to zero. I suspect that this is contributing to the "stuck at 90V" problem.

* The firmware version variable is set to 35 605 (representing 356.05), and because this is more than 32 767, it will be interpreted as negative by any code that treats it as a signed integer. Hopefully there is no such code, but this patch will tell us for sure. Even if the version comes out crazy, it should not effect actual inverter operation.

NONE of Georg's other manOG-10 DSP Reflash_356.05 patched.zipy patches are implemented. The premature float bug is patched as before.

Use at you own risk.

 

Since last week Wednesday we have been in real winter here. The temperatures are always below freezing and yesterday it snowed for the second time.
I cleaned the snow two times from the PV1 fields of inverters 1 and 3.
And I moved the south field (still PV1 of inverter 5) and cleaned this field today as well - Georg's firmware is now active here.
In the meantime I had no time for the project... Tomorrow I will install the new firmware. I hope it doesn't snow again....

Table: Your two firmware variants are compared here (12...06-Dec-2020). Georg's version has a higher energy yield per day - it sticks less.

On 2022/12/09 at 12:58 PM, Coulomb said:

representing 356.05

I have just installed the firmware: only Inverter 3 (156.05 to 356.05).

QVFW
(VERFW:00356.05

On 2022/12/09 at 12:58 PM, Coulomb said:

This is patched firmware version 356.05, based on Axpert Max II 10 kW firmware version 56.05. It has the patches of version 156.05 (current threshold for MPPT1 and MPPT2 reduced from 0.50A to 0.05A), plus:

* Every 60 seconds, the MPPTs are "reset" by forcing the direction to 0 (so that reducing voltage should increase power), and the solar reference voltage is set to 450.0 V (the highest MPPT voltage). This will hopefully effectively perform an MPPT scan, although the scan will stop at the first "hill" found.

* For MPPT1 only, when in low current mode, the direction flag is no longer forced to zero. I suspect that this is contributing to the "stuck at 90V" problem.

* The firmware version variable is set to 35 605 (representing 356.05), and because this is more than 32 767, it will be interpreted as negative by any code that treats it as a signed integer. Hopefully there is no such code, but this patch will tell us for sure. Even if the version comes out crazy, it should not effect actual inverter operation.

NONE of Georg's other manOG-10 DSP Reflash_356.05 patched.zipy patches are implemented. The premature float bug is patched as before.

Use at you own risk.

 

The picture from 12 December shows Georg's version in Inverter 1 compared to 156.05 in Inverter 3 and the pictures from 13 December show Georg's version in I1 compared to the new version 356.05 in I3.

The changes cause a complete restart of the MPPT algorithm at 450V. Therefore the found optimal PP is always left. However, I think the MPPT controller works fine at higher voltage values.

Therefore, it would be better to restart only when the voltage is below a certain value (e.g. 175 V) and then start the restart in a medium range (e.g. 200 V). Then the power drops would no longer occur (see pictures A and B from 13 December).

Of course, it would be even better to examine and adjust the algorithm directly. But the proposed solution would be a good compromise.

On 2022/12/14 at 2:49 AM, Volker said:

Therefore, it would be better to restart only when the voltage is below a certain value (e.g. 175 V) and then start the restart in a medium range (e.g. 200 V). Then the power drops would no longer occur (see pictures A and B from 13 December).

I agree with the above. I've implemented this in version 456.05, with your suggested figures of 175.0 V and 200.0 V.

This is Axpert Max II 10kW main firmware version 456.05, based on factory firmware version 56.05, for use with old reflash tool and using the BDR96 reflash command,
  with just one of Georg's changes to prevent getting "stuck" near 90V PV voltage.
  That change is for the current threshold, now 0.05A was 0.50A
  "Resets" the MPPTs every 60 seconds, but now only to 200V and only if presently below 175V
  Prevents resetting direction in MPPT1 when at low current
  Also, for MPPT1 only, don't reset the direction flag to zero when in "low current".
  Also fixes the premature float bug.
Use at your own risk.

 

OG-10 DSP Reflash_456.05 patched.zip

Hi, thanks to all for this hard work!!!

I also have this Axpert Max II 10K and 90V bug. My firmware is:
Main CPU Firmware Version: 82.02
Secondary CPU Firmware Version: 122.12

So, no firmware patch for me?

4 hours ago, eliafino said:

So, no firmware patch for me?

Sadly, no. My notes say that you are blessed with the latest DSP chip ('28066), and the only firmware updates we have are for the older '2809 chip.

The secondary (display) firmware isn't relevant to the "stuck at 90 V" problem, but again you are blessed with the latest display chip (GigaDevice).

Maybe one day we'll learn the secret of patching firmware for one device to become valid firmware for the other device, but at this stage, that's not feasible.

7 hours ago, eliafino said:

Hi, thanks to all for this hard work!!!

I also have this Axpert Max II 10K and 90V bug. My firmware is:
Main CPU Firmware Version: 82.02
Secondary CPU Firmware Version: 122.12

So, no firmware patch for me?

Try to ask for firmware your supplier, is it possible that you can succeed. I have done that several times but with no results unfortunatelly

17 hours ago, utasek said:

Try to ask for firmware your supplier, is it possible that you can succeed. I have done that several times but with no results unfortunatelly

The suppler tells me that the upgrade is only possible by returning the inverter. It cannot be done with the USB cable, not even with the supplied serial cable. The USB OTG upgrade doesn't work either. He also tells me that some had their control boards blown.

5 hours ago, eliafino said:

The suppler tells me that the upgrade is only possible by returning the inverter.

That will come as a surprise to many who have already updated their firmware several times.

It's hard when your supplier tells you things like this.

On 2022/12/15 at 6:22 AM, Coulomb said:

I agree with the above. I've implemented this in version 456.05, with your suggested figures of 175.0 V and 200.0 V.

This is Axpert Max II 10kW main firmware version 456.05, based on factory firmware version 56.05, for use with old reflash tool and using the BDR96 reflash command,
  with just one of Georg's changes to prevent getting "stuck" near 90V PV voltage.
  That change is for the current threshold, now 0.05A was 0.50A
  "Resets" the MPPTs every 60 seconds, but now only to 200V and only if presently below 175V
  Prevents resetting direction in MPPT1 when at low current
  Also, for MPPT1 only, don't reset the direction flag to zero when in "low current".
  Also fixes the premature float bug.
Use at your own risk.

 

OG-10 DSP Reflash_456.05 patched.zip 1.6 MB · 0 downloads

At the moment it is very cold here (up to 14° below freezing). I was able to clear the snow from the two comparison fields twice, but then a thick layer of ice formed there. There is supposed to be a thaw soon, then I will test further...

Hi Coulomb I have an Aexpert max II 10 Kw from SolarPower24, can you please tell me what processor I have and if I can flash your firmware, thanks

Edited December 22, 20223 yr by ClauBebe

19 minutes ago, ClauBebe said:

I have an Aexpert max II 10 Kw from SolarPower24

Hi, please report CPU firmware version

21 minutes ago, eliafino said:

Hi, please report CPU firmware version

Hi, main CPU Firmware version 56.02 (U1)

Secondary CPU Firmware version 22.10 (U2)

                                                          0000 (U3)

the unit is bought from november 2021 and YES I HAVE STUCK ON 90V ON PV1 AND PV2 ! I have 2 strings of 375w * 10 pics of Canadian Solar , Halfperk Cut

 

Here is unboxing

TX

 

Edited December 22, 20223 yr by ClauBebe

here is voltaj on MPPT1 and MPPT 2 , on both of them is identic  10 pics of 375W Canadian Solar PV, both strings on south

Edited December 22, 20223 yr by ClauBebe

14 minutes ago, ClauBebe said:

main CPU Firmware version 56.02 (U1)

I think you are lucky, right @Coulomb?

 

On 2022/12/15 at 6:22 AM, Coulomb said:

I agree with the above. I've implemented this in version 456.05, with your suggested figures of 175.0 V and 200.0 V.

This is Axpert Max II 10kW main firmware version 456.05, based on factory firmware version 56.05, for use with old reflash tool and using the BDR96 reflash command,
  with just one of Georg's changes to prevent getting "stuck" near 90V PV voltage.
  That change is for the current threshold, now 0.05A was 0.50A
  "Resets" the MPPTs every 60 seconds, but now only to 200V and only if presently below 175V
  Prevents resetting direction in MPPT1 when at low current
  Also, for MPPT1 only, don't reset the direction flag to zero when in "low current".
  Also fixes the premature float bug.
Use at your own risk.

 

OG-10 DSP Reflash_456.05 patched.zip 1.6 MB · 8 downloads

3 days ago, snow and ice thawed away due to a warm rain. As I could not fully interpret the results yesterday, I waited for another day and yesterday the forum did not work…

 

The inverters have been running with the following firmware since then:

QVFW

(VERFW:00056.06 [Inverter 1, by Georg, green in the diagrams]

QVFW

(VERFW:00465.05 [Inverter 3, brown in the diagrams]

QVFW

(VERFW:00056.05 [Inverter 4, original, in the diagrams red]

QVFW

(VERFW:00465.05 [Inverter 5, in the diagrams orange]

 

Notes:

a) I1 PV1 is comparable to I3 PV1, but I3 PV1 has more shading from the south by trees (10 modules each, to the west, but almost horizontal).

b) I3 PV2 is comparable to I4 PV2, but due to the orientation the sun cannot shine directly on it (10 modules each to the east).

c) I5 PV1 almost south and receives direct sunlight from noon onwards (8 modules)

d) I1 PV2 same direction as I3 PV1 (but 4 and 7 modules and different shading)

e) At the winter solstice the sun is at 21.5° at noon! In the last two days there were only very short periods when the sun was visible through the cloud cover - almost only indirect/weak light.

 

I'll keep the interpretation short, please do it yourself:

The comparison between I3 PV2 and I4 PV2 is simple: the original software is poor, but it regulates much more "smoothly".

With I1 PV1 to I3 PV1 you can see that Georg's version can still "hang on". But his curve is very smooth. Whether the "shadow adjustment" remover is good or not, I don't know (yet).

I think the latest firmware flutters too much and this causes losses. Surely the last change should have prevented it?

With the latest firmware I would have expected no flutter behaviour (I3, brown curves, PV1 and PV2 each 10 modules)!?

But I5 (orange, PV1 8 modules) also has this firmware and here the control looks actually better - I can't really explain it.