Lucky Miner LV08 technical help required

[fromport]

Has been running stable for
Uptime:   8 hours, 46 minutes, 20 seconds
Average hashrate : 4.91Th/s
asic temp 55C
VR temp 67C
freq: 600Mhz
Core voltage: 1205 mV

So far so good!
Thank you

[fromport]

Use additional cooling to bring the temperature down on both the ASICs and VRMs. The cooler you can get them the more efficiently they will run. My first LV08 is still cracked open with a 120mm PC fan on top of the heatsink. Otherwise, put a fan up against (or close to) the intake vent and you'll see the temps drop.

At 600MHz and 1.25v on the core of the ASICs, my LV08 power is  showing 109 watts with the ASICs at 47c and the VRMs at 53c. Power at the wall is 129 watts. The hashing rate sits right on 5TH/s!

1.200v is too low for the LV08 at 600MHz in my opinion...

EDIT:
If the TPS546 current approaches 30 amps, it will start to throttle. The initial setup for the TPS546 is OC_WARN_LIMIT = 25 and OC_FAULT_LIMIT = 30. So maybe need to increase these parameters. The TPS546 is capable of up to 40 amps of current!

I am pretty sure I previously ran into the 30 Amp limit.
With the new beta firmware this no longer seems to be the case.
I am keeping track of temps/freq/voltage in a spreadsheet but I just kept hitting a ceiling without anything overheating.
The PSU inside the LV08 is rated 180watt MAX on the load side with 10CFM airflow x 80% = 144 watt continuous.
But since I am in the USA and only able to feed it with 120volt AC it has to be de-rated with 60% :
https://www.dth.net/crypto/miners/lucky/lv08/LV08-PSU-label.jpg
I feel I am pushing the PSU more than the LV08 itself atm.

I have just increased the voltage on the asics to 1215mV and now averaging 5.03Th with 120.7Watt in the gui, 53C asic temp, 64 VR temp

**update**
Just as a quick test, just pushed it to 615/1250
Asic temp 54C, VR 70C
Gui watts 134.5, AC watts 164
Efficiency in Gui: 25.2 J/Th
Average hashrate: 5.16Th/s

Code:

₿ (360919) TPS546: Got Vin: 12.125 V
₿ (360919) TPS546: Got Vin: 12.125 V
₿ (360919) TPS546: Got Vin: 12.141 V
₿ (360919) TPS546: Got Vout: 1.242 V
₿ (360929) TPS546: Got Iout: 31.531 A
₿ (360929) TPS546: Got Vout: 1.246 V
₿ (360929) TPS546: Got Iout: 31.406 A
₿ (360939) TPS546: Got Vout: 1.244 V
₿ (360939) TPS546: Got Iout: 30.625 A
₿ (360949) TPS546: TPS546_0 Temp: 68
₿ (360949) TPS546: TPS546_1 Temp: 70
₿ (360959) TPS546: TPS546_2 Temp: 61

[JohnnyBitGoode]

But since I am in the USA and only able to feed it with 120volt AC it has to be de-rated with 60% :

I don't think that that's what it means.
Going from 240v AC to 120v AC will only make the efficiency of the power supply drop by 1 or 2%.

Assuming the GUI is reasonably accurate, then at those settings your power supply is 82% efficient (134.5 / 164 = 0.82).
Most modern day power supplies are closer to 90% efficient. So I'd say on the DC side of the power supply it's probably closer to 145 to 150 watts. By adding a few watts of extra cooling, you may drop the overall power consumption by 10 watts or so. This will cool down the power supply a little as well and make it more efficient.

[fromport]

I don't think that that's what it means.
Going from 240v AC to 120v AC will only make the efficiency of the power supply drop by 1 or 2%.

Assuming the GUI is reasonably accurate, then at those settings your power supply is 82% efficient (134.5 / 164 = 0.82).
Most modern day power supplies are closer to 90% efficient. So I'd say on the DC side of the power supply it's probably closer to 145 to 150 watts.

As you say: I interpret this different.
The label clearly says:
"85-176VAC Load power 60%"
I read this as : with input voltage 85-176 Volt AC the load power should be limited to max 60%
I don't see anything that says 1-2% less efficient on that label like you interpret it.
The WEB GUI uses the amps/volts of the VR  x 3 strings to calculate the ASIC power usage.
Then it adds a guessed amount for the ESP32 & the fans to come to a WEB GUI power usage number.
It is not 100% measured.

[JohnnyBitGoode]

I don't think that that's what it means.
Going from 240v AC to 120v AC will only make the efficiency of the power supply drop by 1 or 2%.

Assuming the GUI is reasonably accurate, then at those settings your power supply is 82% efficient (134.5 / 164 = 0.82).
Most modern day power supplies are closer to 90% efficient. So I'd say on the DC side of the power supply it's probably closer to 145 to 150 watts.

As you say: I interpret this different.
The label clearly says:
"85-176VAC Load power 60%"
I read this as : with input voltage 85-176 Volt AC the load power should be limited to max 60%
I don't see anything that says 1-2% less efficient on that label like you interpret it.
The WEB GUI uses the amps/volts of the VR  x 3 strings to calculate the ASIC power usage.
Then it adds a guessed amount for the ESP32 & the fans to come to a WEB GUI power usage number.
It is not 100% measured.

Take your pick from a quick Google search:
Most modern day power supplies are switch mode.

https://www.google.com.au/search?sca_esv=c6b583333c988112&sxsrf=AHTn8zpv35ivkZJIuH2b8zXNelsyFL3F6Q:1739831817974&q=120v+vs+240v+power+supply+efficiency+for+pc&udm=2&fbs=ABzOT_CWdhQLP1FcmU5B0fn3xuWp5u5rQsC2YJafWTbyNSy6GwyMGdsHB0d4chDhHh_xW6AFFIH8LbBL39tQziqwJqn4z-wAXy39LfSETRrBOJ8GjWbzIS3MMwkaSWb2P77hvH2rdMUhD35JUxs42fZWv44Sne4SW2TdYHIbht8G3WYZzUlj7UlD9Q-Ja7AheA4nK4mEgCimfe-hyITZ8b_NRXLstjZCyA&sa=X&ved=2ahUKEwiAwOrL4suLAxUKwzgGHb9CG-IQtKgLegQIDhAB&biw=1280&bih=618&dpr=1.5

[fromport]

https://d-central.tech/product/apw3-psu/


With a rated power of 1600W (on 200-240V) and 1200W (on 110-120V)
So 75% available power on 120 volt compared to the max power on 240 volt.

I will follow the labels/specs of the manufacturer.
You can follow the results of some google searches.

[JohnnyBitGoode]

https://d-central.tech/product/apw3-psu/


With a rated power of 1600W (on 200-240V) and 1200W (on 110-120V)
So 75% available power on 120 volt compared to the max power on 240 volt.

I will follow the labels/specs of the manufacturer.
You can follow the results of some google searches.

Yes, it probably is wise to stick with what the manufacturer states about their own products specifications!
Couldn't agree more...

[JohnnyBitGoode]

**update**
Just as a quick test, just pushed it to 615/1250
Asic temp 54C, VR 70C
Gui watts 134.5, AC watts 164
Efficiency in Gui: 25.2 J/Th
Average hashrate: 5.16Th/s

Code:

₿ (360919) TPS546: Got Vin: 12.125 V
₿ (360919) TPS546: Got Vin: 12.125 V
₿ (360919) TPS546: Got Vin: 12.141 V
₿ (360919) TPS546: Got Vout: 1.242 V
₿ (360929) TPS546: Got Iout: 31.531 A
₿ (360929) TPS546: Got Vout: 1.246 V
₿ (360929) TPS546: Got Iout: 31.406 A
₿ (360939) TPS546: Got Vout: 1.244 V
₿ (360939) TPS546: Got Iout: 30.625 A
₿ (360949) TPS546: TPS546_0 Temp: 68
₿ (360949) TPS546: TPS546_1 Temp: 70
₿ (360959) TPS546: TPS546_2 Temp: 61

Hey fromport, I'm still confused about one thing.
If this label: https://www.dth.net/crypto/miners/lucky/lv08/LV08-PSU-label.jpg
de-rates the power supply to 60% of 180 watts (that's only 108 watts), how are you able to draw 164 watts from it?

[fromport]

Hey fromport, I'm still confused about one thing.
If this label: https://www.dth.net/crypto/miners/lucky/lv08/LV08-PSU-label.jpg
de-rates the power supply to 60% of 180 watts (that's only 108 watts), how are you able to draw 164 watts from it?

Remember I said I felt I was pushing it by means of the PSU load?

If you think logically, the higher the input voltage, the more power you can get from the PSU to a max of 180watt DC

It will not be binary situation where the load can only be 60%
If the input AC voltage is between 176-264 volt you will get max output =180 watt DC
Let's assume that at lowest voltage = 85Volt AC input it can still to the 60% of the max power of 180 watt DC
I would assume the power scales from 85 volt (60%) to 176 volt (100%) and would be a linear line.
At the moment my input voltage is 122 Volt AC (measured with multimeter)
so from 176-85 Volt = 91 volt  drop will result 40% loss of DC power.
176-122 volt = 54 volt drop, and it this would be a linear function it would be about a 24% loss of DC power

My guestimate is that with an input of 122 volt AC the max DC output I can draw from this power supply will be 76% (24% loss) of the 180 watts DC) = 137 watt DC

We are talking about output power DC in the above calculations
The 164 watt I talked about is AC power measured of the input.
There is a thing called power factor on the AC side that might make the measured AC numbers higher than what they really are.
Complete new discussion.
Difference between AC power in and DC power output == efficiency of the PSU.

[JohnnyBitGoode]

We are talking about output power DC in the above calculations
The 164 watt I talked about is AC power measured of the input.
There is a thing called power factor on the AC side that might make the measured AC numbers higher than what they really are.
Complete new discussion.
Difference between AC power in and DC power output == efficiency of the PSU.

Well fromport, it appears you were right about the power reduction in output operating between 85v to 176v AC input. At the lower AC input the PSU is rated for 100 watts no additional cooling. Or 150 watts fan assisted cooling (which the LV08 does provide). I bought this PSU because soon I'll have 3 * LV08s and I wanted a backup in case one of them died (the power supply that is). It's very tiny!

So yes, I think you're close to your limit at 164 watts from the wall! The PSU is about 90% efficient, which means 150 watts on the DC side will be ~110% of that on the AC side (110% of 150 watts = 165 watts). Close enough. If you try to draw more power than the PSU can deliver, there's a chance it will shut down.

This particular PSU comes in 4 models, (12v, 19v, 24v and 48v).

Plenty of specs on this link below (scroll down):
https://www.aliexpress.com/item/1005006504788335.html

[fromport]

So yes, I think you're close to your limit at 164 watts from the wall! The PSU is about 90% efficient, which means 150 watts on the DC side will be ~110% of that on the AC side (110% of 150 watts = 165 watts). Close enough. If you try to draw more power than the PSU can deliver, there's a chance it will shut down.

This particular PSU comes in 4 models, (12v, 19v, 24v and 48v).

Plenty of specs on this link below (scroll down):
https://www.aliexpress.com/item/1005006504788335.html

Great find!
I am going to order a few for other projects I have in mind.

[JohnnyBitGoode]

Hello JohnnyBitGoode,

I had no idea someone was working on the LV08 too… I have a modified version of skot’s firmware working on my LV08.

https://bitcointalk.org/index.php?topic=5530594.0

Hello Bonkerz

Yes, I managed to get my LV08 working fully a couple of days ago on v2.5.1-TCH which is forked from skot as well. While the modifications you made to the TPS546.c file communicate to all 3 TPS546 regulators at the same time, it doesn't allow you to query each regulator independently. I went for this approach:

Code:

static i2c_master_dev_handle_t tps546_dev_handle[3];    // For up to 3 TPS546 devices (LV08)
static uint16_t TPS_I2C_ADDR[3] = {0x7F, 0x24, 0x14};   // [1] (0x24) is only referenced for single TPS546 devices

And so I had to add another variable to all the relevant functions:
TPS_IDX = {0, 1, or 2}

Code:

/* public functions */
int TPS546_init(TPS546_CONFIG, int TPS_IDX);
void TPS546_read_mfr_info(uint8_t *, int TPS_IDX);
void TPS546_set_mfr_info(int TPS_IDX);
void TPS546_write_entire_config(int TPS_IDX);
int TPS546_get_frequency(int TPS_IDX);
void TPS546_set_frequency(int, int TPS_IDX);
int TPS546_get_temperature(int TPS_IDX);
float TPS546_get_vin(int TPS_IDX);
float TPS546_get_iout(int TPS_IDX);
float TPS546_get_vout(int TPS_IDX);
void TPS546_set_vout(float volts, int TPS_IDX);
void TPS546_show_voltage_settings(int TPS_IDX);
void TPS546_print_status(int TPS_IDX);

When I need to update the dashboard temperature of the TPS546 devices on the LV08, I query all three but only display the highest temperature. Also power in watts is an accumulation of all 3 TPS546 device's power (voltage * current), with an offset of 18 watts added (6 watts for each regulator). This exactly matches the results I got from the original Lucky Miner v1.0.0 firmware...

Nice work! I like this approach better, I am still trying to implement a similar code to what you have done.

Unfortunately I've just come across an issue that's really bizarre!

Yesterday I received my 4th LV08 and set it up  with my details and mined for several hours at 3TH without any problems. Usually I run my miners with the original loaded firmware for a day or so before I open them up, backup the firmware and install my own modded firmware. However, I decided after only a couple of hours to install my modded firmware on this my 4th LV08.

No matter what I've tried, only 6 out of the 9 ASIC chips are mining.  I used the same modded esp-miner.bin and www.bin files on this miner as with my other three LV08s but still this 4th LV08 has only 6/9 ASICs working. I have confirmed that all 3 TPS546 regulators are added to the i2c bus and that they have been initialized with the same settings. However, it's the middle voltage regulator that is not supplying power to the middle 3 ASIC chips.

I've even used a full firmware backup (v2.5.1 TCH) from one of my other LV08 miners on this new 4th LV08 but the results are the same, only 6/9 ASICs are mining. When I reinstall the original firmware that I backed up before I installed the updated firmware (luckily), all 9 ASIC chips are mining again!

All four LV08 miners appear identical. Looking at all the serial numbers, my first two LV08s are lower than my 4th LV08. But my 3rd LV08 serial number is higher than my 4th one!

Anyone else experience this issue?

[fromport]

Yesterday I received my 4th LV08 and set it up  with my details and mined for several hours at 3TH without any problems. Usually I run my miners with the original loaded firmware for a day or so before I open them up, backup the firmware and install my own modded firmware. However, I decided after only a couple of hours to install my modded firmware on this my 4th LV08.

No matter what I've tried, only 6 out of the 9 ASIC chips are mining.  I used the same modded esp-miner.bin and www.bin files on this miner as with my other three LV08s but still this 4th LV08 has only 6/9 ASICs working. I have confirmed that all 3 TPS546 regulators are added to the i2c bus and that they have been initialized with the same settings. However, it's the middle voltage regulator that is not supplying power to the middle 3 ASIC chips.

I've even used a full firmware backup (v2.5.1 TCH) from one of my other LV08 miners on this new 4th LV08 but the results are the same, only 6/9 ASICs are mining. When I reinstall the original firmware that I backed up before I installed the updated firmware (luckily), all 9 ASIC chips are mining again!

All four LV08 miners appear identical. Looking at all the serial numbers, my first two LV08s are lower than my 4th LV08. But my 3rd LV08 serial number is higher than my 4th one!

Anyone else experience this issue?

I have experienced the same.
And it is mostly 1 of the VR's that stops working but not always the same. eg VR1 stops, VR2&3 keep working. [powercycle] all 3 working, and after a while VR 1&2 working, VR3 no longer produces output power.

I wish someone would include syslog client into bitaxeos so all logfiles could be collected on a syslog server in a LAN so we could see the log files that triggered the event.
That would tell us a lot more probably

[iambonkerz]

Hmm, I’ve had a similar experience before, after I replaced the thermal paste and may have over-tightened the heatsink screws. Loosening them seemed to fix the issue. If you have swapped the thermal paste, you might want to try loosening the screws.

[JohnnyBitGoode]

I have experienced the same.
And it is mostly 1 of the VR's that stops working but not always the same. eg VR1 stops, VR2&3 keep working. [powercycle] all 3 working, and after a while VR 1&2 working, VR3 no longer produces output power.

I wish someone would include syslog client into bitaxeos so all logfiles could be collected on a syslog server in a LAN so we could see the log files that triggered the event.
That would tell us a lot more probably

Hmm, I’ve had a similar experience before, after I replaced the thermal paste and may have over-tightened the heatsink screws. Loosening them seemed to fix the issue. If you have swapped the thermal paste, you might want to try loosening the screws.

Well after quite a bit of experimentation, it does appear to be an issue with one of the TPS546 regulators as fromport has suggested.
What's interesting though is that it does not occur with the original Minerfixes firmware of v1.0.0!

This log snippet shows the minimum settings I could use for a 3TH/s low power usage with firmware v1.0.0:
Notice the iout for the middle regulator.

Code:

Freq: 375
Voltage: 1.085
Watts: 62.3
--------------
₿ (2659126) power_management: 7F vout: 1.08, iout: 14.47
₿ (2659126) power_management: 24 vout: 1.07, iout: 12.45
₿ (2659126) power_management: 14 vout: 1.08, iout: 14.06
₿ (2659136) EMC2302: GET-Fan Speed[0] 989 = 1987 RPM, MSB 0x1E
₿ (2659136) power_management: Board Temp: 40, 44
₿ (2659146) power_management: TMP546_0 Temp: 45
₿ (2659146) power_management: TMP546_1 Temp: 45
₿ (2659156) power_management: TMP546_2 Temp: 44

Unfortunately using the TCH 2.5.1 firmware this particular LV08 only has 2 of the 3 TPS546 regulators working if it operates below a certain frequency.

And once the problem has occurred, the only way to resolve it (it's not permanent though), is to unplug the LV08 (cold boot if you will) and restart it. But before doing that, the LV08 has to be set to no lower than 425 MHz and a high enough voltage so that all 9 ASIC chips are mining smoothly (a minimum of 1.150 volts).

At 425 MHz and 1.150 volts, the LV08 is mining at 3.5 TH/s showing about 79 watts on the dashboard. If I drop the voltage down to 1.125, the hashing rate appears to be stable and the power level drops 4 to 5 watts. But if I drop the the frequency down to 400 MHz, the power level shoots up about 15 watts and even though the middle TSP546 has increased to about 22 amps (from about 15 amps), only 6 out of the 9 ASIC chips are now mining!

The LV08 stays at this higher output level for a short period of time with only 6/9 chips working. Usually within a minute or two, the voltage on the dashboard drops to zero when the middle TPS546 regulator shuts down also bringing down the power usage by a third to around 52 watts.

I'm not sure if it's a parameter than needs to be modified on the TSP546 regulators as I said above, as it does not happen with the original factory firmware. Later this evening I will disassemble the unit so I can apply a better quality thermal paste and check the tightness of the heatsink mounting screws as iambonkerz has suggested...

[fromport]

I had to help 2 persons who flashed your factory 2.61.-1.0.4 image on a virgin LV08 with luckymineros.
Resulted in only 33% of hash power, only 1 of 3 VR's enabled.
In order to confirm, I had one virgin LV08 left that I converted my self and I can confirm

It turns out it has the config.csv from the LV07 is included in the LV08 factory image.
The LV07 has only 1 VR gets initialized and you only get 1/3 of the hash rate.
downgrading to 2.6.0-1.0.3 and upgrading www.bin & esp-miner.bin to 2.6.1-1.0.4 works great


Might want to look into that

Thank you for all your efforts

[iambonkerz]

I had to help 2 persons who flashed your factory 2.61.-1.0.4 image on a virgin LV08 with luckymineros.
Resulted in only 33% of hash power, only 1 of 3 VR's enabled.
In order to confirm, I had one virgin LV08 left that I converted my self and I can confirm

It turns out it has the config.csv from the LV07 is included in the LV08 factory image.
The LV07 has only 1 VR gets initialized and you only get 1/3 of the hash rate.
downgrading to 2.6.0-1.0.3 and upgrading www.bin & esp-miner.bin to 2.6.1-1.0.4 works great


Might want to look into that

Thank you for all your efforts

Thank you! I’ve fixed the factory image issue in v2.6.1-1.0.5. Could you redirect any firmware-related bugs back to the lv08 firmware thread, as there is no firmware released here.