Topic: Lucky Miner LV07 - Firmware wanted (Read 687 times)

I once said that I didn't intend to make other firmware adjustments, and today I'm doing it anyway. 

Okay I understand what you mean now! 

see https://osmu.wiki/bitaxe/api/

e.g. http://bitaxe-ip/api/system/info

Could you please explain these 2 things a little better from your last post?



What do you mean by endpoint?

Miner	Firmware		Comments
LV06	latest standard firmware from skot		boardversion = 0.11, otherwise take the CVS-file of the 204 Ultra
LV07	the standard firmware from skot needs to be modified  in order to handle the 12V input and the 2 asic chips		the lv07-cvs form TCH-LV07 works, after firmware changes

I am testing skot's firmware with customizations for the LV07. The latest build has been running stable on my LV07 for 1.5 days. See here: https://github.com/un-painted-org/ESP-Miner/releases (EDITED)


Have you tried monitoring the heap memory via the /api/system/info endpoint?

And I wouldn't bother checking the GUI for memory leaks. You can just try the latest www-bin from skot. This will fit the TCH-LV07 firmware without any problems. Only the volt display is still scaled to 5V there. And if importing the www-bin causes problems, you can always import a different www-bin via the /recovery endpoint.

Here's an update on the ESP-Miner-TCH firmware running on the LV07 miner and the memory leak issue:

It does appear that the memory leak is occurring only when the logged into the LV07 through the browser. The heap memory leak definitely occurs when the dashboard is the active menu item in the browser. I also tested the Realtime Logs page (with the logs hidden) and this to causes heap memory to go down over time.

Usually I keep the dashboard page open all the time and then open the Realtime Logs page in another tab to check the heap memory status. And as you can see in a previous post I logged a 9+ hours session showing the reducing size of the memory heap. The dashboard screen can show up to about 90 minutes of hashing data. That's quite a lot of info.

So to do a test, I kept 1 of my LV07s dashboard open and checked from time to time on the logs page, and sure enough, the heap memory was reducing in size. Then I logged into my 2 other LV07s only momentarily just to check the memory heap, and they were both staying at around 148K! So it appears as long as you're not logged into the miner through the browser, the memory leak does not occur!

I will say this though, my heavily modified v2.1.1 running on all my LV06s don't have this issue. So I am now leaning towards the website code as possibly being the cause of the heap memory leak...

That's the same as when you pointed me in the direction of merging the correct CVS-file data into the factory firmware when uploading to the LV07. This updates some info in the NVS area (Non Volatile Storage) about what miner is being used. I fully understand this now.

Up until now, the highest skot version I've been able to install in my LV06 has been v2.1.8. Any later version showed incorrect data on the dashboard and the ASICs were not mining! I understand now my error was not updating the NVS with the appropriate data.

As of now, what is the latest skot firmware that will work on both LV06 and LV07 when the correct CVS-file is merged with its factory firmware?

I use the original ESP-Miner firmware from skot for the LV06. You only have to change the board version in the CVS-file, then you can use the firmware unchanged.


Glad to hear I didn't see any ghosts. I have also observed such behavior. With my iPhone, I also have recurring problems to the point where the miner is no longer available via the GUI.

I'd like to point out that I have 3 x LV06 miners all running on (heavily modified) v2.1.1 skot firmware and all 3 miners heap memory is at a fairly constant 164K over periods of days to a week without any issue.

Once again, most pages of the main menu work fine except for the RealTime Logs page. That causes intermittent reboots/restarts.

Here's another interesting thing that I haven't seen anyone else talk about:
Whenever I use my phone or my tablet (both are Android devices), after a certain amount of time accessing my LV06 and LV07 miners I get "Page not accessible". I've tried using different browsers even in Desktop mode on the portable device but the miner is still not accessible. So then I turn on a computer and I'm still unable to access the miner. The only way I can access the miner again is to physically disconnect power to the miner and restart the device. Once the miner has booted up I can then access it again with my portable devices and the PC.

But if I use my phone or tablet again to access my LV06 and Lv07 miners again, after a certain amount of usage, their pages become unavailable. It's only on portable devices like phone and tablet that they become inaccessible but they then cause the miners to be inaccessible on the PC as well. In other words, if I access the miners on PC only there is no issue.

Anyone else have this problem?

Yes I am aware that displaying the logs for a certain amount of time can cause your miner to reboot/restart. Sometimes you can view the logs for quite a while without any issues. And sometimes viewing the logs only for a short time can cause a reboot/restart. It's unpredictable.

The memory leak happens over time, even when not displaying any logs. If I only go to the logs page briefly from time to time to see the heap memory available, the reduction in heap memory is still occurring.

EDIT:
The memory leak occurs regardless of what page of the website you're on.
After 3.5 hours of up time, all 3 of my LV07s are down to 105K of heap memory.
After 4.33 hours, heap memory is around 95k.
After 6.5 hours, heap memory is at 68K only.
After 8 hours, heap memory is down to 46K.
After 9.5 hours, down to 26K of heap memory left available!
15 minutes later, 2 of the LV07s Dashboard were not updating so I decided to restart all 3 miners.

The original Lucky Miner LV07 firmware 1.0.0 and 1.1.0 also showed 8MB of memory. I was aware of this but thought it was incorrectly shown.

Does the v2.4.4 of skot work on the LV07 without any issues or modifications?

My LV07 ran for 7 days without any problems with my firmware. Then I left the GUI open in the browser for too long and the LV07 rebooted.
There is a known issue here, but it seems to be solved already: https://github.com/skot/ESP-Miner/issues/199

If you watch the GUI for too long, it dies because you are watching it. Great.

Does the memory leak also happen when you watch /api/system/info?

Incidentally, I have not changed anything in the memory management compared to the TCH version. I am currently testing a customized version v2.4.4 of skot. There are 8MB more RAM available.

I think this custom firmware has a memory leak bug!

I've spent the last day going through the code but have not found it yet.

When I restart my LV07 heap memory (observable on the Realtime Logs page) is around 148k to 150k. But something is slowly reducing it at a rate of about 6k to 8k an hour. Eventually the LV07 will do a system restart on its own when the heap memory gets too low. This will happen about 15 to 20 hours from a system reboot/restart.

I've been specifically looking at the code on these files: "create_jobs_task.c" and "stratum_task.c" but everything appears to be okay.

I believe the problem probably lies with a call to the malloc() function that does not have a corresponding call to the free() function thereby allowing the heap memory to be slowly depleted.

Is anyone else having this issue or even aware of it?

EDIT:
I've had my 3 x LV07 miners running for 2.5 hours and all them show heap memory under 120K. So it appears heap memory is disappearing at higher rate than I thought. I's closer to 12K reduction hourly!

Thank you for your compliments. Of course I'm happy if someone can benefit from my firmware customization!

At the moment I am only planning to adapt the TinyChipHub version of ESP-Miner to the LV07. Everything else is too time-consuming for me. Apart from that, I want to provide a reliable and stable firmware. I don't have the time or inclination for regular testing and releases.

Great work, unpainted, much appreciated.

Are you planning to synchronize your codebase with skot/ESP-Miner:master?

Hey, I'll try anything as long as I know I have a backup that works!  

Plus there's really no way to brick the device if you have a jTAG on hand and know how to use!  

Yes I am aware that the factory file bin contains everything including the NVS data that is read from during boot up. I was wondering how the setup information is installed in the unit. Now I know how to do it thanks to you! Incidentally, this is where the All Time Best Difficulty is stored. So when you install the factory bin file, you erase your highest difficulty found to date. So you have to start from scratch...

My first 2 LV07s both had firmware 1.0.0. My third and last LV07 had 1.1.0. I don't know if there are any hardware changes between the two firmwares. As far as I can tell nothing appears to have changed. All boards have the same manufacture date. I have noticed only one difference. When all three LV07s are running on the same settings (1.25v @550MHz), the last one I bought is showing 4 to 5 more watts of power usage at 3c to 4c cooler temperature. Go figure that one out!  

I have all 3 LV07s connected to a 200 watt Mean Well PSU as well as 8 large PC fans and 6 individual fan controllers (it's summertime now here in AUS). The PSU has the same voltage on all 3 terminal pair connections. But I'm getting 3 slightly different voltage readings from the LV07s. That's not a problem unless the voltage drops down closer to 11 volts. There is however a graphical issue on the web interface when the input voltage is at 12v or less the dial is skewed to the left the lower it is. At 12.1v or higher the issue goes away. It's obviously a programming error. I might be able to fix this myself...

https://ibb.co/pPJ4zhq   https://ibb.co/dKtpLWy   https://ibb.co/3f8kSY5

I'm glad to hear that! And how is my firmware running on your LV07?

Congratulations: as far as I know, you are the first one who dared to install my firmware!  BTC

I've already thought about all the questions I need to ask you so that we can get to the bottom of the problem! Above all, I was afraid that there might be different hardware versions of LV07.  


You seem to know more about flashing an ESP32 than I do. I don't know how you could create a single factory binary that you could then flash with the esptool.

I'm just glad the bitaxetool exists and the forked github repository can build the binaries on its own.  


If you flash the factoy binary (approx. 15MB) and the CVS file with the bitaxetool, then you already have everything you need on the ESP32.
The smaller binaries are only for the over the air update of the firmware and the website.

So I used the bitaxetool (through VS Code) and typed the following:

bitaxetool  -c config-lv07.cvs -f esp-miner-merged.bin

Flashing proceeded as normal and took about 164 seconds.

Then I Plugged the power in and the LV07 booted up fine. Except it's still in low power mode. Requested voltage is 1.2V but actual voltage is 0.04v. And the power reading is 6w only. Also the fan running at full speed is not as fast as it should be. Data is being received every 30 seconds and is being queued. Everything works in the web interface as well.

I also installed the esp-miner.bin and www.bin files but it made no difference.

Oh, I forgot to mention the input voltage is 5.3V, so it appears the TPS546 voltage is not at the right level.

Something is still not right...

EDIT: I found the problem. It turns out I grabbed the wrong PSU for my work bench to do the testing. The 5.3 volts I was getting was correct as I was using the 5 volt PSU from my LV06! Doh!

No wonder the ASICs were not switched on.

Boy, do I feel like a clone right now...

Okay, I think I understand it now.

The BitaxeTool can be used to to overwrite some data in the NVS partition with the CVS-file info. This only gets overwritten when flashing the complete factory bin file. Flashing the esp-miner.bin and www.bin files don't write to the NVS partition which is at 0x9000.

Up until now, I've only been using the esptool to flash the bin files.

I was curious how the data in the CVS-file gets written to the NVS partion. You actually have to merge it with the factory file.

Is this correct?

I don‘t know the procedure with direct flashing using the esptool. I use the bitaxetool, as mentioned in https://github.com/un-painted/ESP-Miner-TCH/releases/tag/v2.3.3-TCH-LV07:


The correct CVS-file is crucial. Without it, the LV07 does not know who he is and what he should do.

Hello again unpainted.

I tried flashing the binaries, both the esp-miner.bin at 0x10000 and www.bin at 0x410000 with no luck. The LV07 powers up, no screen with fan at full speed. power consumption is only 4 watts. It should be around 25 to 30 watts.

I then flashed the esp-miner-merged.bin with the same result as above. So then I flashed back the backup I made only moments before and the LV07 works fine.

So I fired up VS Code and compiled the source code producing the esp-miner.bin and www.bin in the build directory. This time I only flashed the esp-miner.bin and results are same as above.

Am I missing something here? Should the config-lv07.cvs file be copied to the config.cvs file with the correct parameters for the LV07 in the root directory of the source code during the build?

Take a look at the small but important changes:
https://github.com/un-painted/ESP-Miner-TCH/commit/29f843211120ce2f9011f56eafbd612ee4d9bd62 and
https://github.com/un-painted/ESP-Miner-TCH/commit/5b4569361a259e1825060c09ce631f12c4e0329f

I treat the LV07 internally as an Ultra-Hex with only two BM1366s. I only had to bring the supply voltage up to the level of a simple Ultra.
In the code, the Hex voltage is three times higher because the Ultra-Hex addresses three BM1366s in one chain.

Sorry, my bad!

I should have realized you flashed your LV07 as you said you tried different firmwares yourself!

I got my first LV07 about 5 months ago and I've been looking for firmware (or the source code) to update my LV07s for months. I'm glad you were able to modify the TCH source code and get the LV07 to respond. About a month ago, I tried about 15 factory firmware, including some of the TCH files. These were the only two I could get to boot up from:

esp-miner-factory-v2.1.10-TCH-All-in-one.bin - This one was receiving the stratum data on regular intervals but the ASIC chips were not powered up.

esp-miner-factory-v2.3.1-TCH-All-in-one-303.bin - This one was reconnecting to wifi every 5 seconds but the ASICs were also in low power state. I know this is for the HEX.

Just one question. Was it any easy fix to turn on the ASICs?

Finally, thank you very much for getting the alternate firmware to work. I'll definitely use your source code now as I have a few tweaks of my own I can add to my 3 LV07s!

Hello JohnnyBitGoode

I know the procedure for flashing an ESP32 MCU. This is how I got my firmware onto the LV07 using the bitaxetool.
Yes, exactly: my own firmware! I have explained everything in the German thread.

In a nutshell:
- The LV07 with its 2x BM1366 is practically identical to the Bitaxe Ultra-Hex, only with 2 instead of 6 ASIC chips.
- I adapted the firmware from TinyChipHub and made it work for the LV07. So far my LV07 works perfectly.

You can find my firmware at:
Repository: https://github.com/un-painted/ESP-Miner-TCH/tree/unpainted_2.3.3_lv07
Binaries: https://github.com/un-painted/ESP-Miner-TCH/releases see under “Assets”

Warning: I take no responsibility for my firmware. So far everything works fine. But I am not a hardware specialist. And if you break something when flashing the ESP32, it's not my fault.

Enjoy!


Thank you for your offer! Unfortunately, I was unable to restore my backup of the 1.1.0 firmware. And since I only have one LV07, I couldn't make a second dump of the firmware.

But: I no longer need the LuckyMiner firmware as I'm using my own firmware. I leave it to others who want to go back to the LuckyMiner version to contact you.

Hello unpainted.

Are you familiar with using a jTAG device with a 6 pin probe attached?
If so, I can send you a copy of my LV07 v1.0.0 backup factory firmware and you can flash your LV07 back to life with the jTAG and probe.
If you take off one of the metal panels on the LV07, it's the one on the left if you're looking directly at the LCD screen, you'll see the jTAG probe points.
I bought both the jTAG and 6 pin probe from AliExpress.

I did the same thing as you and tested several other firmwares on the LV07. But because I learned how to flash my Bitaxe MAX (which has no USB port) and LV06 using the jTAG, I was able to revive them! Luckily I had 2 other LV07 miners and I saved a copy from one of them and flashed it back to the non working one without any problems. The only mistake I made, was not making a backup of my LV07 first before I flashed it as it was my newest LV07 with the newer 1.1.0 firmware!

Regards johnnyBytes...

Hello everyone

The same post in German: https://bitcointalksearch.org/topic/lucky-miner-lv07-firmware-gesucht-5521980

I took the risk and flashed my LV07 with open source firmware from https://github.com/bitaxeorg/esp-miner-multichip and https://github.com/TinyChipHub/ESP-Miner-TCH. Had to make a few adjustments because of 2 instead of 6 chips as with the Ultra-Hex. The LV07 identifies itself as board 302.
This worked with both firmwares and the bitaxetool as far as the LV07 came online and showed a web portal. Unfortunately there must be something wrong with the voltage regulator TPS546, because the ASIC chips do not start and do not mine.

Does anyone know a way back to a working LV07? If necessary with the manufacturer's firmware, which is nowhere to be found?

Thank you