Topic: [Guide] Dogie's Comprehensive ASICMiner Blade Setup - page 37. (Read 580761 times)

 Mountain, Please add some pictures to aid in understanding what you mean please. Pictures are worth more than a 1,000 words =P

Please post some pictures of your config menu when you are connected.

Also, are you using two different IP's on each blade? or are you using 1 specific proxy for each individual blade? For example:
Vlad, No. I didn't need to touch the reset pins at all, unless you are actually trying to reset them by "shorting" them out by applying some conductive material across 1-2pin(reset) 2-3pin(factory reset).

Can you also supply me with a link of the PSU or a picture? Hard to help you out man without knowing what PSU you got =) Thanks

Other Info:
How many blades do you have?
Model / Specs of your PSU?
Anything else I might need to help troubleshoot?

Its the Rev 2 board, so integrated NIC. I just bought a new PSU, I can use the 6pin PCIe power right? What about the 4-8pin CPU power? Its 600W, 2 x 12v rails at 18A and 16A, 18ga wire. Does the hardware reset need to be done with power on? The reason I ask is that I've bricked a desktop mobo resetting its CMOS while powered (early days of PC building), and I don't know if this is the same.

Having some issues with my 3 new blades.
Have them set to connect to 2 IPs on my network running stratum proxies.  I've set the proxies to connect to both slush's pool and btcguild but they're still having this issue.

For a little while yesterday all 3 blades were running great at 10.7ghash each.  At this point, however, they display much lower hashes from 4-7 k and are not returning a large amount of shares after receiving them.

When I monitor my stratum proxies, it appears that the blades go on with their work, then at a random time after asking for work or submitting a share, the miner pauses, quits, and connects to the other pool in its config for no apparent reason.  The miners keep doing this and keep switching between the 2 IPs I have in their configs.  Apparently, they work for a little while on one computer's stratum proxy, stop sending back shares, then connect to the other computer's stratum proxy and repeat the process.  

Another weird thing is that when I set the miners to connect to one proxy by repeating the IP address in their config, they don't really work at all.  They appear to receive 8 shares then don't really do anything for a few minutes then reset.  Once I specify 2 different pool addresses in the config, the miners properly connect and begin to do work.

Not sure what is going on here, wondering if anyone could give some advice?


Edit: After reading vesperwillow's post from yesterday, it sounds like we could maybe be having the same issue?

Question: Is there any way to physically determine if the chips are bad? I can digitally see I have 3 x's in a row and want to make sure the chips are not bad. If they aren't bad, then its my PSU. I have two chips running exactly same config using a 350W 12V rail / 12amps per rail. So if the new blades are already OC'd, then each module should have 1.2V and also have 12V at the power module of the ASIC board. Only thing different is one has 3 x's & the other has O's across the board.

Pictures:
Board 1: https://www.dropbox.com/s/1q6q3jtgualykbf/253Config_1.jpg (10.6v @ ASIC pwr Adapter)

Board 2: https://www.dropbox.com/s/s74yxlfncwfi2ft/254Config_1.jpg (10.6v @ ASIC pwr Adapter)

Note: Voltage verifed with cheapo multimeter / voltmeter

Possibilities:

Crappy PSU
Wiring could be more efficient. Possibly use another rail for fans / cooling. (currently have 1 rail for 120mm, ASIC board, & 1 5-12V 60mm fan)
Bad chips on the ASIC board.... There use to be 4 X's, but when i moved the rig, somehow it changed it from 4 - 3 X's...<--- Why I think its a power issue, rather than a broken chip..

Vlad, if you can't even reach the NIC, I wouldn't worry about the PSU being a big problem at the second. The NIC should still respond to your requests. I would do a hardware reset on the blade, short jumpers 2-3 on the NIC (it'll say hardware reset or something like that). You'll know it's reset when you get brigh green LED flashes.

Run a network cable from the blade through a switch/router/etc which is also on a 192.168.1.x/255.255.255.0 network, and make sure your computer is also on the same network subnet. Wait like 10 seconds, try to hit the blade's web interface. This SHOULD work even if your PSU sucks donkey ass.

If it doesn't, try swapping the NIC of the blade with another blade's NIC (if you have another). Otherwise, contact friedcat.

Also, just to be safe, bind all of the yellow wires you have coming out of that PSU, and all of the black ones, and run them to the blade. This will help overcome the 22ga wiring. I would suggest at least 4 of each.

Ok awesome. Now take them two extra 5V (red) wires your not using and take 1 ground from each molex adapter & pair them in series with the 5-6V Fan. Turn on PSU after the adapter is plugged into the ASIC board. That should work.

Pictures: (Going to build a custom rig for the cards to slide in eventually) Just happy it works now =)

• Side Angle: https://www.dropbox.com/s/nq3fad6fn4capv3/2013-09-19%2014.16.19.jpg
• 6V Fan / Molex Adaper (5V red wires): https://www.dropbox.com/s/rj279bvln0tb2id/2013-09-19%2014.15.59.jpg
• Molex ASIC adapter: https://www.dropbox.com/s/5unirtfub0438ox/2013-09-19%2014.15.33.jpg
• Molex Adapter: https://www.dropbox.com/s/018f41b2w0ojuyt/2013-09-19%2014.15.21.jpg

Hopefully them pictures help =) I took two grounds from one slot of the ASIC main green adapter & added em to the 5V red wires for the 6V fan.  (Note: the 5V fan is on top of the ASIC chip blade)

That should fix your problem. If it doesn't send me a PM or post a reply here. If you decide to purchase a new better PSU. Buy one that is at least 800 Watts & has two 12v rails that supports at least 12amps per rail or more. Its important to have about 12v at asic power adapter & 1.2v on each module.

Here is what I do when I need a part TODAY but cant wait for shipping.

Staples has a 14 day return policy.

go buy a PSU from staples, order the one you want online. Once the one that you ordered comes in, take the staples one back.

Worse, I just looked at the wire and its 22ga.

That is a nice PSU, and I would probably be better off with that, but I really need something I can pick up today, since I'm loosing money every day this blade isn't hashing.

Maybe it's just the photo or something, but those wires look like 20ga. They look really thin.

As for a suggestion, honestly I'd go with this to make things easy, it's what I'm doing with my new blades:

http://www.amazon.com/gp/product/B00D7CWSCG/ref=oh_details_o00_s00_i00?ie=UTF8&psc=1

Then just grab some 12 or 14ga wire per blade and you're rock solid baby.

It is an old-cheapo PSU,  300W max. I can't find a molex splitter, as I just moved to a new house and its packed up somewhere in the mountain of boxes. What I did was hack off 2 Molex connectors on seperate rails and spliced the yellow 12v and black ground to the molex connector. Actually, now that I look at it, it appears as though one rail is 1.5A on 12v max while the other is 11.5A on the 12v.





Actually, I'm going to go to Best Buy today and pick up a new PSU for use with this and any future miners I get. Can anyone suggest me a PSU?

I'm not at home atm, so I can't look, but I am pretty sure my blades don't have any jumpers on those headers.

If you are on linux, or Windows with Cygwin installed, you can use 'arp -a' to see if anything else is on 192.168.1.254.

Vladimir - Can you share your PSU specs and take a picture of the modified molex adapter. I had the same problem when working on the blades. Sounds like its a PSU problem and also sounds like your molex modification is correct, but if you have an old or cheapo PSU you will have to use the 5v (red wires) from the molex adapter to power a 5-6v fan. Hope that helps =)


And, no you don't need anything for the reset pins. The 2 pins closest to network adapter factory reset when you short them out and the other two reset the blade and keep the configs from resetting back to default.

Alright, well that's a relief knowing that the blade is functioning. I disconnected my AP just in case it was on 192.168.1.254, still no response from the web configuration. I noticed the 3-pin reset header says "1-2 Normal 2-3 Factory reset" Does this require a jumper on 1-2 to run?

Yes.

My network is 192.168.1.x, and I have been trying to connect to 192.168.1.254. Actually, I think one of my APs is sitting on 192.168.1.254. Is it normal operation to flash the green light, then stop?

Update 5:

• Got both ASIC Blades up and running
• The second ASIC i have, only runs at about 8gh/s... There are currently 4 x' when I connect to the config page. (Where do i check the voltages on each module to see if its a voltage problem or a power module problem?
• Connected em to proxy & everything looks like its working for the most part =)
• Now, I gotta take all the equipment and organize it =P
• More pictures soon To Come. Also will be creating a better tutorial / Guide... I found this one very very outdated & not specific at all.
  

Make sure you are connecting to 192.168.1.254.
Make sure your PC/router has an IP on the 192.168.1.x net.  192.168.0.x won't work.  A lot of people seem to miss that.

I just received my Rev. 2 Block Erupter Blade, and I've wired everything up and connected to my network. When I power the board, a green led behind the ethernet port flashes for about 5s, then shuts off. I can't access the network config page.

Wrong.

Just thought I'd toss in a few pennies of input to this thread. 1 is a proxy issue, 2 is on the 24v conversation.

Odd proxy issue? Stratum proxy died out of nowhere, only getwork functions now

I've got some blades running, have been hashing away like champs 11gh+ for awhile now, going through stratum proxy. These are the original OC'able blades, but running low clock. One stratum proxy server is running on Ubuntu, another is on XP. Everything has been working fine, been getting overly-lucky returns. Something to note, they've always used the unbuntu proxy and have never swapped to the backup one on the XP server.

Suddenly lastnight, they just up and stopped and dropped to 0 hash and then got caught in the 140 second reboot cycle (reboots when its not hashing). Heat is not an issue, power is not an issue (500w psu per blade, all 12v lines tied together), physical network connectivity is not an issue (1ft cables between physical devices).

Keeping an eye on network packets to the proxy, the blades will send a burst if requests to the proxy right as soon as they boot up, the proxy immediately responds back with work, but then the blades are silent. Nothing returned to the proxy, not even bad packets. This is happening on all blades. I have tried letting only one blade at a time work, still the same issue. I have reset the proxy servers.. no change.

I find this odd considering nothing has changed.

They will, after about 10-15 seconds, switch to the backup proxy. Same thing, right upon switching they will talk to the proxy, the proxy returns data, the blades go silent. This continues endlessly. They keep switching between proxies, unable to process anything.

I tried setting up more proxy hosts using different ports. Same issue. Tried different pools, same issue. I compiled midstate and used the proxy with that flag, but that changed nothing either.

HOWEVER, if I bypass the stratum proxy entirely and simply tell it to connect to api.bitcoin.cz (which I've never done before), the blades come back on, 100% efficient, full speed.

WTLF ? Nothing changed. Stratum proxy server / software / settings remained the same. Physically everything is the same. Blade setup remained the same. out of nowhere it simply died and refuses to work via stratum now. I'm going to try bfgminer in a few to see if that's different.

Having examined things at the network layer level, something tells me the blade firmware has a bug in it which causes a conflict with the proxy software at some level--or the proxy software has some bug causing the issue. It gets an initial burst back from the proxy, and then poops. Getwork is working flawlessly. I went back to stratum after this just for giggles, and again 0 hash. Once or twice I saw a single return.

To further prove to myself I wasn't just seeing things, I put the getwork server in along with the stratum (getwork,stratum). Sure enough, it almost halved the efficiency, with nothing being accomplished via stratum, but all the work being done via getwork.

Has anyone else had this issue? I see one other person had something similar happen about the same time as my issue, not sure if it's related.

24v discussion

Regarding the comments made about the regulator and 24v input, what doggie seems to be saying is because the OEM specified an input voltage, regardless of anything else you shouldn't use anything other than 12v--though his tone was "cant" not so much shouldn't. This is the safest bet, however, the fact remains as long as the components between the power port and the regulator are rated for 24v or more, AND the regulator is capable of handling 24v input, then you could use 24v.

As one poster mentioned, this requires verifying input range via knowing what regulator(s) being used, as well as verifying the component ranges mentioned.

Again, the safe bet 99% of folks should take is simply use 12v. If I have time, out of curiosity, I'm going to see what regulator(s) being used, if at all possible.