Topic: Official FutureBit Apollo BTC Software/Image and Support thread - page 96. (Read 48628 times)

Given that this is a FutureBit thread, and I said the models LTC and BTC, I thought it was self explanatory what miners I'm using.  Given that the FutureBit power supplies are 200watts, I'm looking at an 800-1000 watt replacement.

It's from a batch 2 ltc unit, I think it's 2 years old or so. It looks like it had some kind of surge. The wires are toasted and the connectors on the power supply and miner are both destroyed.
Thanks for the recommend on the Dell power supply.  I tried a server one from HP, but it was just too loud for my environment.

Oh right, makes sense. I don't know - they (or someone from the community? not sure) made it sound like it was something quite involved, which relies on some dependencies that break when updating the OS. Since your software proves it's possible to implement in a way that isn't affected by security updates, it's quite weird that the Apollo OS behaves like this.

I'm myself not comfortable running a non-updateable machine connected to the internet and sitting in my home network. So I will either install a stock Linux OS and connect the miner via a short USB cable (maybe make a custom one so it's not too in the way), or put in the time to figure out if it's possible to modify one of your drivers to communicate with the thing via GPIO

I'm not sure we will get an official answer though; the NDA they have seems quite restrictive, since any low-level-ish questions in this thread in general, were never answered.

No, the bab driver is for early BitFury boards (circa 2013-14)
My point was that there's no special libraries for either of the SPI cgminer drivers I wrote, that run on the RPi,
so I'm wondering what is locking down doing linux security upgrades in the apollo ... which is especially concerning since it's an internet device.

I've been using Dell 03MJJP + a power supply adapter a guy makes on eBay. I'm running a full package + 2 standard units at around 370W on Eco with absolutely no issues. It does get a bit loud if you overclock to turbo+ it but still relatively quiet.

https://www.ebay.com/itm/234213412917
https://www.ebay.com/itm/294446940004

How long did you use the futurebit power supply for and how was the damage? I was planning on using my extra ones with some GekkoScience stuff.

Thanks you very much kano! So the assumption here is that this unit uses SPI as well? Once I get my unit, I'll try to get it running with GPIO and cgminer directly. We actually established these use BitFury chips right? So bab might even work out of the box? We'll see!

(bitfury info from here)

sigh....
What are the model numbers or better yet, how much power does each one need? We're not psychic and there are many different miners that all use different amounts of power...

Can anyone recommend a good PSU to use to run 2 LTC and 2 BTC miners? Right now I'm using the Futurebit power supplies and one blew up on me so I'm looking to replace and consolidate.

I can only guess they used some special library, that's completely unnecessary.

If you look at the 2 SPI GPIO drivers I've done in cgminer (bab and minion) there's no necessary special drivers at all.

They were done on ... RPi ... (and I moved the minion driver to another device with minor changes as can be seen in the code)

Edit: the first one (bab) was a BitFury board

Oh yes, sorry for the bad wording! I know what GPIO is, used those pins on Raspberry Pi projects a lot myself already. What I meant to ask was with which protocol data is transmitted over those pins in the case of the Apollo unit.
As you said, in theory, it could be anything; from I2C to SPI to USB or anything else really.

Just do a simple search using 'GPIO protocol' and you will find Think of it as a modern version of the LPT parallel port that used to be used for printers which is also pin addressable.

Very good info about it is here: https://www.ics.com/blog/introduction-gpio-programming  A key bit from that is

Thanks a lot! Is the GPIO protocol anything 'special' or just internal USB? Because if it was, maybe someone can come up with a kernel patch to apply to an arbitrary Debian or other Linux installation

No coms through GPIO or USB is the same. Unfortunately the GPIOs require kernel level changes to make coms work with the board, so its not an easy change. Main reason is for ease of use. Most people buying our device are new to mining so a plug and play all in one solution was the main design choice (we were debating having a separate controller (remember the KNC days , but this is way more cleaner and easier to use).

Thats what I thought. I’ll make up a gpio to fan connector lead and plug it in at some point when I get time.

EDIT:-

Both 5v GPIO pins are used by the proprietary plug so couldn't get access to it without cutting wires which I'm not prepared to do so I'll just put up with it.

Yes, I did that in mine, it's connected to deliberately lower 3.3V, I put it back to 5V. You can do that by re-crimping the wire or just forcing it through the hole with 5V. You can find 5V by checking Orange Pi 4 pinout or checking with multimeter, as you will already know where GND is, just find where 5V is. 5V it's right next to 3.3V, but check exactly if front or rear row somewhere.

Is it okay to have GPIO and USB connected at the same time and initialize only the preferred connection?

Is there any way to get the CPU fan to run at a faster speed?

I don't know for sure, but this must be for aesthetics only. Everything fits under miner skirt. No extra USB cable required.
Hashrate is the same in both cases, I've tried that, no differences in performance. Functionality also seems to same, hashrate readings, speed, fans, temps, power use.
So feel free to give your Full unit a USB cable so you can install system of your choice. You can always revert to Futurebit system if you don't like it, by reformatting SD card and removing USB cable.

There are a few differences between batches and even units within batches. We have multiple van vendors and their PWM curves are not exact match, so some fans will be faster.

FYI

After having units running from Batch 1 and units from batch 2 I have found a difference in temps and fans speeds from the two batches.
Although I was concerned I'm not to concerned now, just for fan longevity of batch 1 fans.

It all started when I added a batch one unit to my batch 2 setup
Batch 1 Setup https://imgur.com/a/A0KMm1H
Batch 2 Setup not as fancy, https://ibb.co/s9S5NVc

All batch 2 unit temps were 68c-69c and fan speeds 2800-2900rpm constant across the 7 units from batch 2
When I added a batch 1 unit to the other 7 units one unit was 62c-63c and fan speed 3000-3100rpm (so lower temp with faster rpm)

The question to myself was why is one unit so far apart from the others (originally didnt know it was the batch 1 unit yet)

Through process of trouble shooting I isolated the lower temp unit and swapped the fan for one of the others of the consistent temps and fan speed like the others. I found out that it was not the unit but the fan itself as the lower temp and faster rpm followed the fan from unit to unit.

I further looked at the fans closer and notice a distinctive difference.
Batch 1 (4 units) I got had these fans with smaller gauge wires and a different wiring pattern.

Blue yellow red black
https://ibb.co/GTGwJkH

Batch 2 units had larger gauge wires
Black red yellow blue
https://ibb.co/NZkKXWB

Both fans do have the same order at the plug to the board just different going into the fan. (different brand of fan I take it)

In conclusion the difference between the fans from Batches is quite noticeable so if your seeing this too, here is what I found.