v1.46 is now ready to download

Actually, it’s not 89.5 but rather 85. I subtract 50 from 900. I better change it to not subtract any amount to give you the full range. This will be the only change so it should be safe. It will be v1.46 and be in the server in 5 min. I’ll let you know if you want to force update.

Hi Lael, I was looking through the code and nothing was changed wrt the pseudo ion. Recall, that the implementation of the pseudo ion = 0 is by changing the cutoff voltage to the max dc input voltage so that the inverter will detect the cutoff voltage has been reached and turn off. However, for some reason, I have the cutoff voltage changed to 0.5 V lower than the max dc input. For example, on a 45-90V inverter the max dc input is supposed to be 90V, but it's actually set it to 89.5V. Is it possible your DC input is greater than 89.5V and that is why ion = 0 isn't working? If so, I can change it 90V.

Hi Lael, Just pushed out v1.45 which includes LWT. The topic is 2ms/SB/x/st where x is the device number. The payload is ’connect:offline’. The keep alive time is 60s. Retain flag is set and QOS of 1.

Hi Lael, That’s great! I want to apologize again for the inconvenience. This should not happen again as I’m instituting a check on my test setup to check auto updates before releasing. Previously I used over the air update which is different than the automatic http update because it uses a hw rest instead of the sw reboot so it got missed. About LWT messages, it is on my list to do so I’ll bump it up on my priorities since you asked. I’ll try to get it in the next version.

V1.44 is now on the update server. If you want after you are connected to your network, you can use the debug page and check for updates. It should find v1.44 and download. Once downloaded it will reboot using the old version which will erase your log in so you will just need to redo your log in in stand alone mode. No need to cycle AC power because it’ll be using v1.44’s reboot which is now good. Sorry for the inconvenience.

Ok. I finally figured it out and you are exactly correct. It does connect to your network but then when it reboots, it looses the stored log in because of a bug that I introduced to make the reboot feature more robust, but apparently it makes things worse. So basically the software reboot does not work, but a hard power reset will work. So the only way to connect to your network is do the following. 1. Wait until it’s in standalone mode. Fast blinking red LED. 2. Then connect to sunbeam. 3. Go to 192.168.4.1 4. Go to network configuration 5. Enter your SSID and Password. Don’t hit save yet. 6. Watch the yellow LED on the SunBeam board and get ready to cycle the AC to the inverter. 7. Hit save. The yellow LED will start blinking about 1 per seconds. When it stops blinking probably within 3-5 blinks. LED will be off for more than a second. Then cycle the power immediately before it tries to reboot. You have about 2 seconds. SunBeam should connect to your network. 8. Do not use the reboot button on the home page or you will have to do this again. 9. I will need to upload a fix V1.44. When you do get it, you’ll need to do this one more time minus the AC power cycle and then it should be fine afterwards.

Hmmm that is strange. Are all the other settings the same as what you set them?

Hi Lael, I made an update in v1.43 to support user selectable MQTT message rates from a message every 1.1 to 11 seconds. The selection is under the network configuration page under the MQTT section. I just uploaded the update 1.43 so you should get it within 5 hours.

Hi Ben, Thanks for the information. I'm in contact with someone who got a new Sun2000 with the same relay modification on the motherboard. To me, this is really good news because it means the manufacturer is trying a hardware solution to brownout and it validates the SunScreen approach. Based on your description, it looks like the relay engages whenever there is 12V and not when it senses a good AC signal. Since the 12V DC is from the AC and the inverter micro is 3.3V, it looks like the relay should disengage as the 12v drops and probably before it reaches 3.3V to ensure there is no AC when brownout happens. SunScreen keeps the micro from going into brownout whether there is AC or not. As long as during brownout, there is no AC, there should be no damage. If the relays are draining the batteries sooner, I think that is still okay as long as the battery can last during the momentary power loss which is usually a few seconds. If the power is out long enough for the battery to run out even with the relays, it will switch off very fast and the micro probably doesn't experience brownout very long and the AC would be off by then. Your rerouting of the 12V does allow the battery to last longer, but I think it's unecessary because brownout is still prevented during the loss of AC power. However, without your rerouting, it does make their relay modification ineffective because the relay doesn't turn off when the AC is dropping out. Your modification does allow basically "double" brownout protection, one using their relay and the other using SunScreen. On another note, it appears there are multiple causes to the AC short problem and that the manufacturer is trying to address. 1. is what we talked about already with brownout. and 2. is an AC short when the 2 IGBTs prematurely fail during normal operating conditions most likely from heat when there is too much voltage or current somehow. If you look at the new board, they've changed the IGBT to higher voltage and current rating and they used a full body package (it doesn't have the metal tab). I think they are the components mounted to the end of the heatsink. For the Sun2000, the IGBT is now 1200V, 24 A where previously it was 600V 20A. The full body is supposed to be able to dissipate the heat better. So this appears at first to be a good solution because it makes the IGBT more robust. But I noticed they have turn off delay 3 times worse than before which means significant more heat which could negate the higher ratings. I've tried slower IGBTs on my unit and they do blow sooner so turn off delay is a factor. I'm now trying IGBTs that are signifcantly faster and it seems more stable. I'm thinking they might not have realized the issue could be a heat issue from slower IGBTs. Anyways, if you don't mind taking a picture of the IGBT on the new Sun1000 so I can see which ones they are using now. The one for the Sun2000 are CRMicro G24T120AK3SD.

Hi Lael, Pretty much the same average range for both import and export with a maximum of 50W.

I didn’t see the second part your earlier post until now. Yeah, there’s no way to sync the Sun GTIL2’s time since it works similar to the LCD back light command. It’s only used by the LCD so the command is within the LCD and it’s not exposed. On the other hand, the date is sent to the control board for whatever reason. SunBeam could send its date instead but it won’t change what’s displayed on the LCD screen. Since I don’t know what the date is used for, it’s probably best not to mess with it.

Hi Lael, good to know. Thanks!!

Or you can work timer to make sure it stays off while charging.

Great. When you do get 6.1, there’s a chance ion and the LCD screen work timers won’t work if I’m right so you might need to go back to your older version so make sure you save a copy or wait for me to implement a manual switch. If it does work, then check SunBeam’s debug page to see if the start/stop received counter increments every minute. If it does that means ion commands are working and the LCD screen work timers will work and I was wrong. About en and enl, Sunbeam actually moves en into enl and saves it as it crosses midnight. Then it zeros out en. If you can delay turning it off until 12:01 then that would work better. I think adding FW version MQTT will be added in the next version.

Hi Lael, 1.39 did some housekeeping that I needed to do. The serial number packet that is received from the control board is 11 digits, but I noticed the last digit is always 3 for everyone AND only the first 10 digits are displayed on the LCD screen and on the label on the side of the unit. So to avoid any confusion, SunBeam now shows only the first 10 digits. I also changed things for a wider release like changing the debug page (adding a clear reboot count button and made check for updates more reliable and friendlier, also the packet details were removed to save memory since I'm planning to leave the hidden debug page). I also added a couple more MQTT status keys - "enl" is the lifetime energy generated x 10 and "bvoltp" is the cutoff voltage when the power limit is set whereas "bvolt" is the cutoff voltage when the current limit is set. These are separate for older LCD screen firmware like yours. But for SunBeam, there is only 1 "bvolt" command that will set bvolt = bvoltp just like the latest LCD screen firmware. All this is now reflected in a new version of the manual that is online now.

Yeah, the initial drop to 0 happens when any command gets sent. Then for my Sun2000, it stays at about 35-40 W and remains there until I send or the LCD board sends a true ion command . I’m guessing every minute either the LCD Screen or Control Board checks to see if it is really off and if it isn’t it’ll turn it off. So wherever the original command is issued within that minute, will determine the response time. For whatever reason, I’m guessing this 2nd turn off method is dependent on if you have the work time feature. On another note, if you do try to add the work timer feature w/o updating the control board, can you let me know if it works? If it doesn’t work, then both the real and pseudo ion commands might not work if the new work timer firmware sends the real ion commands and the control board doesn’t execute it. The reason is because I’m using the detection of real ion commands to automatically switch to real ion mode. I may have to change it so that it’s a manual selection rather than automatic in case someone tries to update w/o changing the control board and it doesn’t work.

That’s great news!! Finally! Everything should still work including the remote work timers from SunBeam. You just don’t have the one provided by the Sun1000. I can’t think of any limitations using especially the cut off voltage as an ion command. The one you set in remote control mode is saved. You basically will not see the new value of the cutoff voltage when ion is 0. It’s set to the max DC in voltage for the inverter type. Give it a good shake down and let know if you have any questions or notice anything weird.

1.38 is ready for you to try. Click check for updates from the debug page.

Yes, it’s cheap. I’m actually testing the pseudo code now. I’m putting the finishing touches. I should be done within 10 minutes. I’ll let you know. I did notice on my Sun2000, it relies on the ion commands to completely turn off the inverter if the cut off voltage is reached. If I block the ion commands then the generated amount won’t go lower than 40W. Let’s see what happens on your end.