Unofficial Spondoolies SP20 thread

[LASERminer]

Newbie question if I may:
I am trying to set SP20E to work night shift only, so I set Voltage Scheduling for each day to 0V before 8am and default 0.69V after 6pm. Saved - looks good.
When refreshing Setting page I got this:

only Monday is as set every next entry changed to out off range values, same on two machines.
Firmware  2.6.14, cgminer 4.7.0
Any idea?

[krisgt30]

Im having a problem with my SP20 that I just bought off of ebay.

I have tried getting my SP20 to start and it wont. When I turn on the power supply the fan will come on run for about 5 seconds, then shut off and the yellow light will come on. Then the yellow light goes off and the fan comes on again for about 10-20 seconds then the fan shuts off again. The yellow light then begins blinking 6 times every few seconds. The ethernet lights are both going but it is not showing up on my router and going to myminer.io shows me nothing.

I have tried to sd boot it = nothing
sd reset = resets but then when I take out the sd card and restart it.....nothing
......I even tried the blow dryer trick......nothing

The old owner has been a huge help trying to get it going but so far nothing

Any help would be majorly appreciated!

What PSU are you using? Model and watts.

I would start at looking at that.   I never had to do the sd boot luckily. 

Im using a Thermaltake Toughpower Grand 1200W

Garbage psu multi rail so


Here is my guess.

One rail is 36 amps that is good for one pcie plug not 2
Second rail is 36 amps that is good for one pcie plug not 2
Third rail is 20 amps that is only good if your gear is set to 200 watts per pcie plug

So if your gear is set to 200 watts per pcie plug you can only do three plugs not four.

I would connect two pcie plugs

C.       Empty
C.        Empty

See diagram.

If it starts let me know

If it does not start

C.       C.
Empty empty

If it starts let me know
If it does not start

Empty.  C
Empty.  C

If it starts let me know

If it does not start

Empty  empty
C.         C

It should start on one of those patterns

With 3 rails of 36 36 20

You could get it to work on careful setup and big under clock.

Since the most a 36 can do is 18 x 12 = 216 watts.  A setting like 150 watts would be okay.

So .61 volts. And 150 watts would be good.

But you need to start it to downclock it to thay

Ok tried all the combinations....same thing.
I even brought up another 1300w that I have and tried them both together....Still nothing
Is it dead?
The guy I bought it from mined until he shipped it and it worked for him...

I was the original owner of this machine, had my sp20e hooked up to 2 corsair rm850s, then underclocked it down to 1. Worked both ways. I reset to factory settings, turned off the device and shipped it out.

[Finksy]

Ok tried all the combinations....same thing.
I even brought up another 1300w that I have and tried them both together....Still nothing
Is it dead?
The guy I bought it from mined until he shipped it and it worked for him...

Try a different ethernet cable, and setting the IP address by DHCP using something like tftpd32 and connecting directly to your computer.  I had some that were difficult to connect to initially using DHCP, but most of my issues were how I was wiring the ethernet cables (not swapping the blue and green whites) once i had more than 6x SP20's in use.

[QuintLeo]

only Monday is as set every next entry changed to out off range values, same on two machines.
Firmware  2.6.14, cgminer 4.7.0
Any idea?

 The web software is bugged for setting voltages.
 Use it to set initial info, then SSH into the machine and manually fix the entries in the chrontab

 It worked - mostly - but when I did daily voltage schedualing sometimes something would hang and cgminer would never restart.

[LASERminer]

only Monday is as set every next entry changed to out off range values, same on two machines.
Firmware  2.6.14, cgminer 4.7.0
Any idea?

 The web software is bugged for setting voltages.
 Use it to set initial info, then SSH into the machine and manually fix the entries in the chrontab

 It worked - mostly - but when I did daily voltage schedualing sometimes something would hang and cgminer would never restart.

Thank you. Where I can learn about SSH and chrontab?

[klondike_bar]

only Monday is as set every next entry changed to out off range values, same on two machines.
Firmware  2.6.14, cgminer 4.7.0
Any idea?

 The web software is bugged for setting voltages.
 Use it to set initial info, then SSH into the machine and manually fix the entries in the chrontab

 It worked - mostly - but when I did daily voltage schedualing sometimes something would hang and cgminer would never restart.

this. it has some issue with adding zeros to every successive cron entry and is easy to spot and fix.

1) download putty ssh.
2) use it to "ssh" to your miner IP, port 22
3) login is admin/admin (IIRC)
4) type cron -e and you should pull up cron in editor mode
5) from tere, read up on a few google sites, crontab is quite simple and takes only a minute or two to figure out

[LASERminer]

Thanks so much!

[LASERminer]

this. it has some issue with adding zeros to every successive cron entry and is easy to spot and fix.

1) download putty ssh.
2) use it to "ssh" to your miner IP, port 22
3) login is admin/admin (IIRC)
4) type cron -e and you should pull up cron in editor mode
5) from tere, read up on a few google sites, crontab is quite simple and takes only a minute or two to figure out

3) works with root not admin - got #1 prompt - or I'm doing something wrong? PuTTY from putty.org has nothing of "IIRC"
4) -sh: cron: not found

help please

[QuintLeo]

there should be a crontabs file in /etc If I Remember Correctly (IIRC).
You should be able to manually edit that.

[LASERminer]

solved: crontab -e (not just cron -e)

Thanks guys for help again!

[nicklello]

Does anyone know what the right tool(s) are to tighten up the heatsinks on the SP20E hash boards; I have a second hand unit with extremely loose heatsinks.

Also, is it safe to drill out the rivets that are holding the intake end plate in order to remove the hash boards --- does anyone have any experiences to share (good or bad) doing this ?

[alh]

Does anyone know what the right tool(s) are to tighten up the heatsinks on the SP20E hash boards; I have a second hand unit with extremely loose heatsinks.

Also, is it safe to drill out the rivets that are holding the intake end plate in order to remove the hash boards --- does anyone have any experiences to share (good or bad) doing this ?

I am confused by the reference to the end plates and board removal. I have at one time removed both hash boards from my SP20, and did it all from the "top" (i.e. where the controller is on top). I did have to removed the controller to do this, but I didn't have to touch the end plates. I did vacuum the end plates while the hash boards were out.

[tlhIlwI]

Does anyone know what the right tool(s) are to tighten up the heatsinks on the SP20E hash boards; I have a second hand unit with extremely loose heatsinks.

Also, is it safe to drill out the rivets that are holding the intake end plate in order to remove the hash boards --- does anyone have any experiences to share (good or bad) doing this ?

I am confused by the reference to the end plates and board removal. I have at one time removed both hash boards from my SP20, and did it all from the "top" (i.e. where the controller is on top). I did have to removed the controller to do this, but I didn't have to touch the end plates. I did vacuum the end plates while the hash boards were out.

Second this.  Disassembly is all done from the top... 6 screws get you to the controller, disconnect that and the 4 fan screws, 6 more get you to the blades.  Pull the spacer and then they come straight out the top.

Regarding the heatsinks-- tightening them will not help.  The springs just do not hold enough tension on their own and are nearly useless.  The heatsinks almost entirely rely on the epoxy to keep them down.  If that epoxy is broken/gone then you'll have to get rid of the existing epoxy stubs, clamp the heatsink down as level as possible (any gap will ruin it), apply new epoxy, and let it harden.  If you are going that deep you will almost certainly mess up the paste, so you may as well do a repaste while in there.  If it is working good enough then I would recommend not bothering since a lot can go wrong without experience.  If it really needs it then it is possible and can be done.

I shot a bunch of video almost two months ago now about a SP20 repaste project which covers this, but I haven't gotten around to editing and posting it yet.  I haven't been all that motivated to edit it since the SP20 is so outdated, but if there is interest I can find the time.  My repaste was better than new though-- I even managed to get one of the back ASICs able to hit 0.75 while still only 100C (I didn't even know it was possible to get one of those over 0.71 ever).  What was my worst SP20, a flaky eBay special that topped out at 1400 under ideal conditions, became my second best able to hit 1640.  Of course, these days I never run them past 0.64 anyway-- no profit otherwise.

[edgar]

i think the spacer is the confusing bit, mine took a serious bit of levering to get up. i actually sliced my finger doing it. minorly

[alh]

My two cents (unsolicited) on the SP20 and it's speed. I would only push above about 1300GH if I had VERY cheap electricity. While it's originally rated at 1700GH that's more of a mythical number that few folks achieve, and only at great electrical and efficiency cost. I think that the SP20 is a WAY more manageable miner at about 1250 GH with good efficiency, reliability, and temperature control. That will also help in managing any "hot chips" and reduce the need for new paste and such.

One other suggestion: It's possible to lay the SP20 on it's side, so that the boards are horizontal, with the heatsinks above the chips. This tends to help level them out and make better contact with the chip. It's really unfortunate that Spondoolies only used 2 screws per heatsink. It's easy have them just a bit "cocked" and make less contact with the chip.

Just my thoughts on using a great miner, though fading quickly in terms of efficiency.

[nicklello]

i think the spacer is the confusing bit, mine took a serious bit of levering to get up. i actually sliced my finger doing it. minorly

That's what I'm trying to avoid doing.... the way the end pieces fold over prevents you from cleanly removing the boards; however I did not remove the spacer....

I assume the spacer referred to is the one that the fan sits above at the outward end of the unit ?

(My inner 'mad scientist' is making me think about simply immersing the unit in a liquid coolant rather than screwing around with it all... has anyone done this I wonder ?)

[LASERminer]

Yes - under the fan.
Should be loose, free to pull it w/o tools.

[nicklello]

Yes - under the fan.
Should be loose, free to pull it w/o tools.

No joke, I stupidly did not try that !!

 

[LASERminer]

I shot a bunch of video almost two months ago now about a SP20 repaste project which covers this, but I haven't gotten around to editing and posting it yet.  I haven't been all that motivated to edit it since the SP20 is so outdated, but if there is interest I can find the time.  My repaste was better than new though-- I even managed to get one of the back ASICs able to hit 0.75 while still only 100C (I didn't even know it was possible to get one of those over 0.71 ever).  What was my worst SP20, a flaky eBay special that topped out at 1400 under ideal conditions, became my second best able to hit 1640.  Of course, these days I never run them past 0.64 anyway-- no profit otherwise.

WOW! 0.75V at 100c? I'd love to see your video! What magic you have done. I'm looking to maximise while staying under 100c.
Maybe SP20 is outdated but when you run miners in working environment, new noisy ones are out of question. I run SP20 at 0.61 to keep ASIC reporting 85c (I guess it is actually less and 85 is software lower threshold). This way fan stays at 6 with "auto" stetting and hash goes up to 980GH/s taking under 500W. Show me miner you can down-clock so far! And bought it under $150 in EU.

[tlhIlwI]

WOW! 0.75V at 100c? I'd love to see your video! What magic you have done. I'm looking to maximise while staying under 100c.

Not magic-- I just got extremely lucky on that one ASIC.  The others perform more typical for their position on the board.  Getting the heatsink perfectly level over the ASIC to close the gap is the key.  That is extremely tricky to do given the small surface area.  I used a thin coat of AS5 for the ASIC-to-copper heat spreading plate junction (where there is a lot of heat in a small surface area), and ordinary Thermalcote for the copper to aluminum junction (the heat is spread out there so a lower grade paste works just as well).

IMO, Spondoolies used way too much paste on the copper-aluminum junction.  Also, several of my copper plates had distinct nicks on them that indicated they were never balanced properly on the ASIC to begin with, and the aluminum heatsink bottoms are way too rough.  They had the right idea with a copper heat spreader before the aluminum heatsink, but the execution of it was poor.  Had it been done better (4 sided with correct tension and smoother aluminum) these probably would have met their 1700GH spec.

Maybe SP20 is outdated but when you run miners in working environment, new noisy ones are out of question. I run SP20 at 0.61 to keep ASIC reporting 85c (I guess it is actually less and 85 is software lower threshold). This way fan stays at 6 with "auto" stetting and hash goes up to 980GH/s taking under 500W. Show me miner you can down-clock so far! And bought it under $150 in EU.

Interestingly, I find mine to run a bit more efficient W/GH around 100C than at 85C (my power meter reads about 25W per miner less for the same hash), but mine are all down to 85C now too due to profitability.  The one I repasted shut down once from being too cold (10C intake resulted in 42C exhaust at 0.62V/6% fan and it didn't like that-- it handled 0.64V/6% better).  Winter is gone now, so that is no longer a problem, but I'm still running 0.64V for the time being since the profit difference is miniscule and I'm cautiously optimistic.  

Except for the eBay one, I bought mine in January 2015 during the fire sale and they are still mining 15 months later (on 0.0875/kWh power).  I more than ROI'd those!  Their profit is tiny now, but they will remain on until no longer profitable (which could happen any diff change now).  I'll probably toss them solo for a month after that for a little fun before halving then I'll retire them.  I almost sold them, but I guess there are some tax implications since I used them for more than a year which make it less worthwhile to sell them now (I don't completely understand it so don't ask me to explain-- that's what a CPA is for).

I'll see if I have some time in the next week or two to edit that video.  It's been a bit chaotic here, so I'm behind on at least a dozen projects.  I'm not sure why I'm on a forum right now.....