HashFast BabyJet users thread

[SolarSilver]

Silicon for B2 and the complete full-pull MPP were allegedly ordered months ago at the same time.  Maybe they have massive yield issues?

from what I see in other threads:

https://bitcointalk.org/index.php?topic=269216.msg5058225#msg5058225

35% bad systems arriving (so pure wafer yield should be a lot worse) is about  the same results I got: 4 systems arrived and only 3 work fine (25% bad)

My RMA request is being postponed each time with the excuse "wait for the new firmware, it might magically fix your problem"
Great... a software solution for a hardware bug ;-)

[stan258]

I just sent in request.  Was working fine for 5 days.  Now starting stopping thermal overload.  Is somebody going to help me or am I on my own.  This just sucks.

[cedivad]

Maybe they have massive yield issues?

Maybe not?

[iambaboon]

Guys, let's be ontopic here.

[storm2k5]

Guys, let's be ontopic here.

Indeed. This is a technical discussion thread. Everybody here understands that people want answers to these questions, but this is not the place to ask them.
We don't want this to evolve into a 400+ page rage thread.

[perezoso]

so batch 2 gets delivered already. And batch 1 still waits for the upgrades. This is not right.

Batch 2 people are getting their units? darn it. Yea, I want my upgrade unit AND my 'miner protection program' too.

2nd call to Hashfast in over a week with no responses about it.

here the same.

I was overclocking to 652 Mhz on own build of cgminer 3.11 and it works perfect at 440 Gh/s since 29th jan.

It takes 11 seconds on hotpluging the usb device. I run minepeon version from HF Engineer.
No problems all is good at this moment.

You should get more than that... at 3.11.0 on Windows I get 450 GH/s at 606 MHz on one of my units.  These hash non-linearly with clock, bump your clock by 2 MHz at a time until it jumps.

I was going to say something similar.  If your machines behave like mine, I'll bet you can get the same performance, or damn close, at a much lower overclock rate.  I'm getting 430-435 gh/s at 600, and only a few gh/s more at 610.  For those few gh/s, I was "paying" with an additional power draw of about 3 watts per gh/s.  

If you are oc'ing to 652, I'll bet you are drawing over 500 watts at the wall, and may be paying quite a bit in electricity for only a couple more gh/s than you would get at a lower (more efficient and less risky) overclock rate.

[storm2k5]

I just got notification from Hashfast (72h wait) telling me they will send me a new board and they will update me with the tracking number. It looks like they don't even want me to send back my botched board (which I think is great)

So soon I will have 2 boards, hopefully one of them will mine at full speed. But I want to use that broken 250ghps board of mine, even tho it probably got a very bad efficiency power wise.

So the question is, what do I need to run both boards in the babyjet? I already know I need that cooling system, but it seems I also need a connector cable to loop the USB to the 2nd board. I can probably solder my own cable, but would need pin assignment. Or is it just straight through? Which connectors on the board do I have to use to loop to the 2nd board?

[mgio]

I just got notification from Hashfast (72h wait) telling me they will send me a new board and they will update me with the tracking number. It looks like they don't even want me to send back my botched board (which I think is great)

So soon I will have 2 boards, hopefully one of them will mine at full speed. But I want to use that broken 250ghps board of mine, even tho it probably got a very bad efficiency power wise.

So the question is, what do I need to run both boards in the babyjet? I already know I need that cooling system, but it seems I also need a connector cable to loop the USB to the 2nd board. I can probably solder my own cable, but would need pin assignment. Or is it just straight through? Which connectors on the board do I have to use to loop to the 2nd board?

How did you get in touch with them???  I've been waiting a week and a half for them to contact me regarding my broken babyjet (it was broken on arrival) and i have heard NOTHING yet. Did you call or email or something else? Did you harass them or where you just lucky.

[stan258]

So Far I have Replaced the PI, SD card and the cpu cooler.  Clocked back to 600 and getting 425 GH from the pool and getting about 1.25% stale or returns.  However Temp reading are all over the place in Cgminer.  (Yes I properly applied thermal paste). Any sugestions? 

[HF-Engineer]

I just sent in request.  Was working fine for 5 days.  Now starting stopping thermal overload.  Is somebody going to help me or am I on my own.  This just sucks.

Thermal overload is an easy fix.  It's either a loose cooler head (tiltled or insufficient pressure) or somehow not enough thermal compound.  Get a tube of Arctic Silver, Noctua, or Xigamatek.  Unscrew the 4 phillips cooler head screws (don't lose the washers!) and then re-apply a very small amount of paste (a grain of rice sized amount) right in the center.  Reinstall the cooler head, and tighten the screws evenly on each corner until the bottom out.  They should stop, if they keep turning, you haven't tightened them fully yet.

This should fix your thermal overload.   This will also usually fix a device with one or two hot dies on a side.

-Phil

[HF-Engineer]

I just got notification from Hashfast (72h wait) telling me they will send me a new board and they will update me with the tracking number. It looks like they don't even want me to send back my botched board (which I think is great)

My understanding is that the procedure is we send you a label for the return board FIRST.  Then you ship your board back, THEN we ship a replacement ASAP.  I am not aware of any procedure where we ship one out first.

-Phil

[HF-Engineer]

How did you get in touch with them???  I've been waiting a week and a half for them to contact me regarding my broken babyjet (it was broken on arrival) and i have heard NOTHING yet. Did you call or email or something else? Did you harass them or where you just lucky.

As I've said, we hired a new term for support, and are working through the Zendesk tickets.  I'm helping as much as I can (when I have spare time).  This is why I'm not posting in here as often.

The best thing you can do is send an email.  Be sure to include as much information as possible.  A 5-10 minute log file attached generated with "--verbose" is very helpful and needed with most cases. (Tip: Zip or TGZ before attaching.)   People with clear requests having adequate information and not being rude and irate or more likely to get prompt assistance.

-Phil

[HF-Engineer]

HOW TO GENERATE A TECH SUPPORT LOG:

You can create a detailed log by putting the following line in your "Extra cgminer parameters" which is located under the "Settings" tab in your RPI web interface:

Code:

--verbose --monitor 'cat > /opt/minepeon/http/log.txt'

Be sure to get the single quotes (apostrophes) and spaces correct.  Incorrect data entered in this box can break your system!  Note that this will now be logging a large volume of text to your SD card, which is a bad idea, as it will rapidly wear out if it has to log all that data.  So be sure to restore the original settings (default is blank) after your logging session.

If your log is getting overwritten too much because the session keeps dying, simply add an additional ">" to the existing one, with no extra spaces.  This will cause it to append the logs instead of overwriting them.

Once you have seen the problem occur, you can access the log file at: http://ip.address.of.your.minepeon/log.txt (so for example: "http://192.168.1.101/log.txt")

If you are not sure of your IP address, you should be able to find it by visiting http://setup.hashfast.com/.

If you can then ZIP or TGZ up the log file and either email it or upload it somewhere where you can share it, such as https://mega.co.nz/, We will then be able to take a look and see what's going on.

Again, DON'T LEAVE THE LOGGING RUNNING TOO LONG!  It will burn up your SD card!

-Phil

[starsoccer9]

I just got notification from Hashfast (72h wait) telling me they will send me a new board and they will update me with the tracking number. It looks like they don't even want me to send back my botched board (which I think is great)

So soon I will have 2 boards, hopefully one of them will mine at full speed. But I want to use that broken 250ghps board of mine, even tho it probably got a very bad efficiency power wise.

So the question is, what do I need to run both boards in the babyjet? I already know I need that cooling system, but it seems I also need a connector cable to loop the USB to the 2nd board. I can probably solder my own cable, but would need pin assignment. Or is it just straight through? Which connectors on the board do I have to use to loop to the 2nd board?

Wow lucky. What email did you send to? I might as well contact them and tell them my board broke and see if they will ship me a free one.

[storm2k5]

My understanding is that the procedure is we send you a label for the return board FIRST.  Then you ship your board back, THEN we ship a replacement ASAP.  I am not aware of any procedure where we ship one out first.

-Phil

Ah great, I just received a 2nd reply saying that they'll send me the form. If I have to ship back before I get the new board, I probably wont be doing it, except they do overnight. Waiting a week or two will mean at least one difficulty rise (or two) and I'll probably spend the next 2 months mining getting back what I lost during waiting for the board to return. I guess if they send me the label they also pay for the shipping at least.

What would you guys do? Mine away @250ghps@stock or @320ghps@634clock (and hope for firmware that fixes the issue) or send the board to Hashfast?

@Everbody saying I'm lucky:I knew it was a risky investment (even tho all seemed well at HF back then) and therefor I assumed the ~3k$ I paided were gone from start (at least, I was smart enough not to pay in bitcoins) Therefore I didn't engage in the monster thread where everybody let off steam.

On the 2nd of february or so I sent them my first support mail ever (support@hashfast.com), asking for delivery or I would request a refund. 24h later I had the tracking number. Another 2 days later I got a sub perf board, no idea if you guys call that lucky, cause now it looks like I'm basically stuck with it.
When sending the 2nd support mail about the low perf of my board, I included all the information I could possible give them about it, including updating the case with further infos. Not that I think they tried to analyze the problem.

I kept it civil tho, cause I know from first hand experience that the guy/girl sitting at HF support is probably not the one to blame for anything.
And I found it goes along way towards, you know, not pissing the people off who have to decide wheter they go have a smoke and leave for the weekend or just finish that one last friendly formulated case...

[storm2k5]

Another thing: I want to apologize to everybody who got their hopes up about getting a board without having to send it in the damaged one. Or not sending in a working board but claim it as broken. But this was the first mail I got:

Dear  ,
We apologize for the issues you are having with your BabJet. We are sending you a new board so to replace the one you currently have.
We will send you the tracking number so you know when you should be expecting your board.
Sincerely,
LD
Customer Service

They probably reading on this thread and somebody intervened.

[Melbustus]

Hi Phil,

I tried some very minor overclocking. Looks like there's only a smidge of leeway on my machine as it stands right now. Specifically, I went to 570 clock and everything was fine, but when going to 580 (also tried 600), I get the following errors, non-stop:



Note that since the last update-push from HF, bringing us back to cgminer 3.9.0h2, my system was very stable at stock clock. Hashing steadily just under 400GH/s at Eligius, with little variance. Never more than 3-4 degrees variance between the cores, miner only rebooted once in several days, etc...

So I figured I'd have more overclocking leeway than 570. Any suggestions?

Thanks.

Well, this is still happening when I overclock *at all*. 555 does it too. I even tried compiling the hash-fast cgminer build on an Ubuntu laptop to rule out *anything* having to do with the RPi; mined fine at the default 550, but gave the same above errors any higher.

Suggestions?

^^ Just in case this got lost a couple pages back ^^
??

[CurcO]

Hi Phil,

I tried some very minor overclocking. Looks like there's only a smidge of leeway on my machine as it stands right now. Specifically, I went to 570 clock and everything was fine, but when going to 580 (also tried 600), I get the following errors, non-stop:



Note that since the last update-push from HF, bringing us back to cgminer 3.9.0h2, my system was very stable at stock clock. Hashing steadily just under 400GH/s at Eligius, with little variance. Never more than 3-4 degrees variance between the cores, miner only rebooted once in several days, etc...

So I figured I'd have more overclocking leeway than 570. Any suggestions?

Thanks.

Well, this is still happening when I overclock *at all*. 555 does it too. I even tried compiling the hash-fast cgminer build on an Ubuntu laptop to rule out *anything* having to do with the RPi; mined fine at the default 550, but gave the same above errors any higher.

Suggestions?

^^ Just in case this got lost a couple pages back ^^
??

If you have a good Chip then you should be able to clock it at 600-625 stable (3>  watchdog /issue per day), also have you tried to do a manual reset and PSU power cycle? . The manual reset is on the left side of the board, the first small button from the left. Sometimes when OC it seems the chip might go into fail safe, at least it did it on mine when I went all out and did 900 OC and failed ,jajaja , I had to do the manual rest and was able to clock it at 600-625 stable. If not then guess HF Engineer might be able to help . Also have you tried another host like an Ubuntu setup instead of the dreadful RPi.

[btchedge]

Hi Phil,

I tried some very minor overclocking. Looks like there's only a smidge of leeway on my machine as it stands right now. Specifically, I went to 570 clock and everything was fine, but when going to 580 (also tried 600), I get the following errors, non-stop:



Note that since the last update-push from HF, bringing us back to cgminer 3.9.0h2, my system was very stable at stock clock. Hashing steadily just under 400GH/s at Eligius, with little variance. Never more than 3-4 degrees variance between the cores, miner only rebooted once in several days, etc...

So I figured I'd have more overclocking leeway than 570. Any suggestions?

Thanks.

Well, this is still happening when I overclock *at all*. 555 does it too. I even tried compiling the hash-fast cgminer build on an Ubuntu laptop to rule out *anything* having to do with the RPi; mined fine at the default 550, but gave the same above errors any higher.

Suggestions?

^^ Just in case this got lost a couple pages back ^^
??

CG v 3.11.0 makes it much easier to overclock, and has been very stable. As mentioned before you can see the "real" hashrate via the API. Also, you can easily change the clock rate via command line.

Your config file will look like this. Notice you can change clock and temp below the pool settings. Then when you run the full stats via the API it will show you actual [hash clockrate] => xxx

If you put in certain speeds then check the API it may not be the same. For what it's worth, 656 seems ideal for running a single unit on a controller whereas a controller that has 3 attached runs best at 604. With these settings the config and API show the same clockrate.

{
"pools" : [
            {  
                "url" : "xxxx",
                "user" : "xxxx",
                "pass" : "123"
        },

        {
                "url" : "xxxxx",
                "user" : "xxxxx",
                "pass" : "123"
        }
],
"hfa-hash-clock":"656",
"hfa-temp-overheat" : "89",
"api-allow": "127.0.0.1,192.168.xxx.xx"
}

[KaosMcRage]

Not sure if this is a pool thing (Eligius) or something else, but my BabyJet has been dying once/twice a day after running completely stable for over a week. What do you make of this?

https://i.imgur.com/FZeDgSZ.png

It will just keep updating every time it detects a new block, but it won't mine or switch pools. Restarting seems to fix it temporarily. Any ideas?