T17/S17 malfunction: cases, solutions, remedies, RMA history

[philipma1957]

I wtb a couple of psu fans for my t17, two of then are not working

w40s12bmd5 from nidec

12vdc 0.64a

https://www.nidec.com/en/product/search/category/B101/M111/S100/NCJ-W40S12BMD5-5/

Anyone knows where to buy?

delta makes one that should work it is 40x40x28. same size. it is .81 amp which is close to your spec.

it is four pin. 

ebay item 283157645798

even if you dont order it from that seller you can look for part number.

[1714953279]

1714953279

[1714953279]

1714953279

[1714953279]

1714953279

[1714953279]

1714953279

[mikeywith]

Wait, you can actually get a coupon because you had to cover shipping expenses to their service location?   I haven't read that info anywhere.

You can, if you know how to negotiate, in fact, i asked for

1- A new control board
2- A compensation for my shipping fee
3- A compensation for the downtime caused by their irresponsible actions.

I got 1 and for 2 I got a coupon, and nothing for number 3.

You see in reality when you pay for a brand new mining gear and it arrives broken, they should be responsible for all 3 points,  but because of lack of competition and people not demanding their rights - Bitmain can keep treating their clients like shit.

I have bought hundreds and hundreds of gears directly from their website, some are mine and some are for clients, maybe I get slightly better treatment than those who bought only 10 or 20 gears, but this doesn't stop anyone from demanding their rights from Bitmain, in the end, all you lose is a few emails back and forth.

[Scorpyy]

Wait was it a DOA ( dead on arrival ) ? Or did they work for some-time?

[mikeywith]

It's complicated, so bear with me, the machine was DOA,  I replaced the control board and the gear was hashing away, before changing the control board I sent them an email describing the issue hoping they would give me the recovery files to fix it, but no luck with the recovery, so I stopped talking to them and simply went along with replacing the control board (took a while for me to get the replacement), and I was busy then, so I didn't bother about it.

Later on, in March I was like hey, I think Bitmain owe me a control board, so I sent them an email demanding all three points above, and this was their first reply

Mar 5, 18:20 CST

Dear *************

Thanks for your consultation!

why did you not ship it back in Feb we found you consult with us in Feb if you send back to us at that time it will meet the DOA
We checked your miner did not belong to the DOA now, due to you did not create the repair ticket  and send it back to us in time which we have advised you before
4. Please ship back them within 30 days from Order Ship out date, otherwise the devices cannot be treated as DOA.
we checked your SN which is exceed 30 days warranty
and for the DOA the 30 days is calculated from shipping date to creating repair ticket date ,but you still not created the repair ticket in our system

So to them and according to their rules, it was not DOA, so they wanted me to bear all costs, apparently, I didn't give a flying **** about that and went on with my daily negotiations, a few emails back and forth resulted in a new control board shipped from China and a coupon to compensate the shipping cost to the Netherlands.

If it was considered DOA (me creating the repair ticket within 30 days from when they shipped it out) would they compensate the shipping cost without having to send Jessica a few emails?

As I explained above, I don't know much about their warranty and DOA stuff, not even sure if what I got was a good deal or should I have asked for more?

[Biodom]

Hello, I have been reading every single answer to this interesting post. This is because I have also been framed by Bitmain with those S17+ antminers. I have some background in electronics and analyzing those intermittent failures that these miners have just got me a little confused, specially during a time that I checked control boards and data cables by disconnecting back and forth. At times the miner was full throttle mining just to drop after few hours. That in electronics is called a "cold soldering". Although there are not visible cold solderings I discovered that the soldering material that is being used on this S17 have 2 problems.

First, high amount of lead and low proportion of tin, and of course zero silver in the soldering material. This can be observed under the microscope used to soldering circuit boards. All of the solderings showed a greyish mate color and no shinning finish.
 
Second, below normal temperature of the soldering material, also leaves traces, specially on the hashing board chips. Also analyzed under the microscope there are traces that look like an sponge along with tiny cracks surrounding the chip's area at the surface of the hashing board.

So, this takes me into the conclusion that somebody at Bitmain's is trying to cut ends and by doing this they have compromised 50% of the S17, T17 series. The use of soldering with high percentage of plumb just tells me that they are throwing the cheapest crap they have to solder the components on those miners. Also they are trying to reduce electricity costs by not heating well enough the soldering material melted at the trays and this is specially true because the soldering material with higher proportion-ratio of lead instead of tin and silver takes colder temperatures to melt.

So taking all this into consideration I really recommend to all of those that are willing to throw the money into shipping costs, customs fees and no mining income, just don't do it because Bitmain still uses the same bad quality materials to fix it.

Since that soldering material is all over the hashing and controller boards is just a matter of time chips keep coming off the boards, the only and costly solution is to desoldering them all and soldering back with a good quality soldering flux, other than that those miners will keep failing over and over.

Basically most of the S17 series are doomed.

Since S9 had a pretty good reliability/longevity (on average), I wonder what is the cause of the apparently shoddy work done on these machines?

[philipma1957]

Well the s9. had a lot of bad batches say 4 to 11 way back say. sept 2016.

It came out in April 2016. batches 1,2,3 were decent. Then a drop off.

but in feb 2017 on they got their shit together.  They put out a lot of s9's they worked for 2 years or more.

The s15 and t15 came and gone quick under a year
The s17 and t17 lasted a year- 14 months . say April 2019 to  June 2020

Now the s19 are out.

So no time spent on bettering the gear has been spent.

Also. I am willing to bet that. a ton of s17 gear was burnt due to too much over clock.

But that is another guess.

[Scorpyy]

Here is the update on my situation.

T17e ordered, worked for 6-7 months. Dropped a board recently - out of warranty - will try to resolder chips as explained in this thread once i gather all knowledge and material.

2x T17+ - dropped almost all boards after 2 months of operation, sent to Netherlands for repair, shortly after we received info that replacement units have been sent out from China.

After 20 days of trouble with customs, we finally managed to get them in our possession. Customs inspected units and concluded that it were used ones and not new ones and asked for statement that we are importing operational units and not scrap (importing scrap is against the law). Both units were really dirty on 1 side, like it was stored in some warehouse for 3 years, i haven't inspected the inside but fans appeared to be clean. Its not just regular dust you can wipe it out, this appeared like it was glued to the miner. You could scratch it out only. Bitmain sent out replacement units within 2 days of receiving ours, and units arrived in our country in less then 2 days via DHL, so that was fast, but our customs delayed it a lot. Also one thing to notice is that Bitmain extended our warranty period for 18 days.

Shipping didn't cost us very much. I think total will be 80$ for shipping, and about 62$ for customs (for both units total). But the trouble we went trough has no price. Since we ordered as company we needed to provide a total of 12-17 documents ( statements, invoices, confirmations etc etc ) regarding export and import.

So, i was aware that Bitmain sent out same crappy gear and that it is just a matter of time before same issue re-occurs. I plugged in both miners. Both worked fine but then...

After 5 minutes = 1 miner dropped a board.
After 4 hours of mining = the other miner dropped 2 boards.

So conclusion is, replacement units were DOA, One miner works with 1 board, the other one works with 2 boards ( for now ). Both miners have exact same issue ( temp sensors ). I have contacted Bitmain and they told me to try and replace firmware  and if that didn't work they suggested me to send them again for repair but this time they will provide me with coupons for shipping fees. I don't mind sending the units back again for repair, but whats the point if we are going to bounce back-and forward this shitty gear until warranty expires.

[favebook]

Both units were really dirty on 1 side, like it was stored in some warehouse for 3 years, i haven't inspected the inside but fans appeared to be clean. Its not just regular dust you can wipe it out, this appeared like it was glued to the miner. You could scratch it out only. Bitmain sent out replacement units within 2 days of receiving ours, and units arrived in our country in less then 2 days via DHL, so that was fast, but our customs delayed it a lot. Also one thing to notice is that Bitmain extended our warranty period for 18 days.

If in addition to your statement, the miner was oily or dirt had oily/wet look to it. You can be damn sure they used it in immersion (either Bitmain (for testing or otherwise) or some other customer which returned it for repair).

Shipping didn't cost us very much. I think total will be 80$ for shipping, and about 62$ for customs (for both units total). But the trouble we went trough has no price. Since we ordered as company we needed to provide a total of 12-17 documents ( statements, invoices, confirmations etc etc ) regarding export and import.

So, i was aware that Bitmain sent out same crappy gear and that it is just a matter of time before same issue re-occurs. I plugged in both miners. Both worked fine but then...

After 5 minutes = 1 miner dropped a board.
After 4 hours of mining = the other miner dropped 2 boards.

So conclusion is, replacement units were DOA, One miner works with 1 board, the other one works with 2 boards ( for now ). Both miners have exact same issue ( temp sensors ). I have contacted Bitmain and they told me to try and replace firmware   and if that didn't work they suggested me to send them again for repair but this time they will provide me with coupons for shipping fees. I don't mind sending the units back again for repair, but whats the point if we are going to bounce back-and forward this shitty gear until warranty expires.

At this point (as you are a company ordering from other company), I'd politely ask for a big check of compensation for all the troubles (if they give you coupons, fight that as they are mostly useless unless you plan on supporting(buying from) them (which I wouldn't do if I had gone through all that) or ask to trade your DOA and broken machines for one of their new miners (if that works for you).
If politeness does not work, you could threaten them with a legal action for sending broken machines as "new ones" and then failing to repair them, but that is totally up to you and I do not know how would that benefit you as you will probably lose a lot of time if you actually go through with it. But a little threat could work out in your favour if politeness does not as I previously stated. But it's always POLITENESS>threat.

Also do not be afraid to politely tell them that even those units that are out of warranty need to be fixed. I paid only 40 USD or something silly like that for my S17 PRO (that broke twice in a span of 9 months). First repair was in warranty and was free ("only" 2 dead boards here) and second one was like 2-3 months out of INITIAL warranty (day when I got my first unit and not it's replacement) and as I said, I paid something silly while all three boards were dead (physically there was nothing wrong with them (I personally took them out, inspected them and filmed them from both sides and sent videos to Bitmain Support)). But they kept dropping at initialization 95% of the time. Also I should note that I tried with different control board but that did not fix or change the problem in any way. Only the PSU was untested as I didn't have a different one at that time.

[Scorpyy]

you could threaten them with a legal action for sending broken machines as "new ones" and then failing to repair them,

I can't as terms from Bitmain say:

1.6 “Replacement Product” or “replacement product” means the Product of the same type or specifications provided to replace the defective Product or Product of the same brand with hash rates no less than that of the defective Product. Replacement Product may not be a new Product. The warranty period of the Replacement Product will be the remaining warranty period of the original Product calculated from the date of dispatch of the original Product, except when the warranty is voided in accordance with the terms and conditions of this Policy.

https://service.bitmain.com/support/terms

As for the state of miners, they are not oily, they are dusty. I can show pictures if anyone is interested.

Regarding the first one T17e that dropped 1 board out of warranty - i think im gonna wait for it to drop more boards. But as i said whats the point in repairing if they are sending same crap back.

I have made 2 proposals to Bitmain so far:

1) We send you DOA units, you replace with new series once available ( we can pay difference in value )
2) We send or keep DOA units, you provide with coupons for new series once available. We expect coupon not to be below 60-70% of what we payed for 3 units, T17e 2x T17+.

This seem fair and logical instead of sending crappy gear back and forward. Also we are risking our company to get fined for up to 10.000 USD because importing scrap equipment is against the law.

[favebook]

Bad choice of words and word order on my side. What I meant is that they sent you NEW ones (ones you bought), and they were either DOA or broken in few days/weeks after NORMAL use (if that is the case, and I would assume it is). And about their so called "replacement product", they should work and not be DOA either, so that is another point you could make.

1) works much better for you imho
2) really bad idea if they give you 10% or 30% coupons. If they give you 50%+ coupons, that might be good for you.

*by % I do not mean % of what you spent, but rather the rule of coupon where you can use it to deduct only up to % of your new order*

[Scorpyy]

Ok so i decided to  re-solder chips, however instead of going for every single chip i want to start with one that is already faulty. Does anyone know a way i can determine which one is it if there are no visible damages on the board? I want to gather all required information before i dissamble the miner.


@The_binary_Observer we need your expertise on this one.

[The_binary_Observer]

Hello good friend Scorpyy, to begin with visual inspection is paramount. The most probably is that you will find at at least 1 chip falling off the hashing board, although I'm shure there is going to be more than one. The thing here is to touch, press and shake every heat sink on the hashing boards in order find those that are almost loose but not enough to fall by gravity. Most of the time if there is a complete hashing board not showing, if that is the case, chip number 1 and chip number 65 should be checked.

This is because the way the hashboards are designed, the chips circuitry works  in series so if the first or the last chip of the chain fails the whole chain fails.

Some people say that if the temperature sensors on the hashing board fail the controller board does not activates that hashing board, the thing is that if there is 1 out of 4 temperature sensor working the controller stills start that hashing board and is very unlikely that all 4 temperature sensors fail all together.

Now, if you want to take this to the next level in order to be absolutely and completely sure about which component is causing the failure on your miner then you have to buy an special diagnostic device called universal test fixture that connects in series between the miner and a laptop or pc, from there you can start to determine what is exactly causing the hashing board not to start.

Depending of how many hashing boards you expect to repair and assuming that you are going to use the universal test fixture device along with a top quality soldering paste, soldering flux and hot air solder you may be spending around $400 dollars. However, I put you this way, there is a place in the U.S. that claims they repair S17 series hashing boards, they charge for EVERY hashboard repair service $350, plus shipping.

If you send at least 1 complete miner to Bitmain in Hong Kong you may be end paying $150 or $200 on shipping costs plus customs fees when they send back a new crappy miner which is not guaranteed to be 100% free of failure within the next 15 days of first start. However, if there is more than one complete miner to be shipped to Bitmain then do the math between buying the repair/diagnostic equipment or paying either 3rd party repair or shipping/customs costs to and from Hong Kong, you may be ending paying more than $400 and if your miner or miners fail again you are ready to fix them right away on the spot.

Also, I have found that the S17 series are shipped from Bitmain overclocked and that is one of the reasons they fail so much. All S17 use the same hashing chips, what makes them different is the clock speed and the number of chips on hashing boards. I have been running some testings with this S17 and S9 and I have been able to reach consistent and almost zero HW errors overclocking to 100 Th on S17+ and 20 Th on S9. Don't ask how I do reached those numbers because there is time and money invested on those tests that I am not willing to share. All I can tell is that if the above instructions, specially    re-soldering the chips, are followed higher overclocking speeds can be safely achieved. By saying this I am really suspicious about the S19 series and their so called 110 Th. From my perspective it seems those S19's are nothing more than the same S17 with maybe couple of more chips per hashing board with an aggressive overclocking. Only time will tell if I'm wrong or not, I can't wait to see massive failure claims about those S19 in the coming weeks.

To end this post my advice to everyone ...save, your time, your money, your mental health and your soul STAY AWAY FROM BITMAIN PRODUCTS.

PS: Buy Kester soldering paste and soldering flux. Some people here maybe thinking... "this person works for Kester and that is why is recommending it" I don't work for that company but in my years of experience Kester soldering products have proven to be highly reliable.

[mileys1990]

I did order a new one and it ships whiting 20 days so i might end up with at least a month of an expensive gear sitting there doing pretty nothing but loses value. mobdro
tubemate

[The_binary_Observer]

I wish you the best of luck with that my dear friend.

[danieleither]

I've recently taken delivery, and installed the following:

12 x T17+ 58T
24 x S17e 60T

These were all installed 1 week ago, and much to my surprise - all 36 units are working perfectly so far (they've been running 7 days). I was expecting at least a few to be DOA or drop hashboards within a few hours. Of course I realise that they can still fail, but anyone that understands the 'bathtub curve' will know that if a new piece of equipment is going to fail, it is most likely to do so very shortly after powering up (e.g. within a few hours).

It is of course early days, but it's a good sign that perhaps they have made improvements with the manufacturing / soldering process? These units are all from the Shenzhen factory.

It would be interesting to hear from some S19 owners about reliability so far, DOA units etc?

[philipma1957]

I've recently taken delivery, and installed the following:

12 x T17+ 58T
24 x S17e 60T

These were all installed 1 week ago, and much to my surprise - all 36 units are working perfectly so far (they've been running 7 days). I was expecting at least a few to be DOA or drop hashboards within a few hours. Of course I realise that they can still fail, but anyone that understands the 'bathtub curve' will know that if a new piece of equipment is going to fail, it is most likely to do so very shortly after powering up (e.g. within a few hours).

It is of course early days, but it's a good sign that perhaps they have made improvements with the manufacturing / soldering process? These units are all from the Shenzhen factory.

It would be interesting to hear from some S19 owners about reliability so far, DOA units etc?

A lot depends on room temperatures.

What is the hosting room at 25c or more?

[Scorpyy]

but anyone that understands the 'bathtub curve' will know that if a new piece of equipment is going to fail, it is most likely to do so very shortly after powering up (e.g. within a few hours).

My brand new units:
2 T17+ died within 2 months.
1 T17e died within 6 months.

And as for the environmental temperatures it all happened during the winter so no worries on that side. I wish you good luck with those honestly.

[danieleither]

Hard to say as I've gone for a passive cooling approach with this setup. Each miner has a double fan-duct on the rear, connected to 150mm ducting (approx 0.5m lengths) which in turn exhausts the hot air directly to the outside. The intake air temperature will be dependent on the air temperature outside, since there are large air vents on the intake side (drawing air from outside). Outside temperature has ranged from 10deg to 25deg over the past week. There is a small build-up of heat from the psu's which circulates back through the miners, and depending on wind conditions - it's possible some of the heated exhaust air may get drawn back into the miners from outside.

[The_binary_Observer]

It would be interesting to hear from some S19 owners about reliability so far, DOA units etc?

Tell you something, my parts supplier for bitmain miners just sent me a message last night related to the Universal Test Fixture software... They are working on an update to attend the new S19. So, what this tells me is that S19's are about to be the same story of the deceiving S17 series.

Also. I am willing to bet that. a ton of s17 gear was burnt due to too much over clock.

You are absolutely correct !!!

They were shipped overclocked straight out from Bitmain. Among the several tests I have been performing the last 10 months with those S17 series I have discovered that they are overclocked for several reasons but I am going to mention a couple of them.

1- The HW error factor is unusually higher than average, if there are no problems with the internet connection the next cause for this to happen is aggressive overclocking

2- The working temperature factor of the miner is too high even on controlled environment ( 23 degrees celsius  ) or winter time, plus the fan speed is also too high 90% of the time.

[mikeywith]

My brand new units:
2 T17+ died within 2 months.
1 T17e died within 6 months.

While I don't personally own many of those T17 series but you are not alone, EVERY single person I know that has those gears have had a very high failure rate, the only exception so far is S17pro, there is something special about those gears, maybe they are way underclocked and that's the reason why the solder paste is still holding well.

my parts supplier for bitmain miners just sent me a message last night related to the Universal Test Fixture software... They are working on an update to attend the new S19. So, what this tells me is that S19's are about to be the same story of the deceiving S17 series.

Let me guess, was it Kevin from www.zeusbtc.com? if he tells you that the S19 are as terrible as the 17 series I would believe him without a single doubt, but the fact that they are developing a tool/firmware to fix those gears doesn't really indicate anything bad about them, the proof is, the first fixture tool they made was for S9 and we all know how robust those gears were. I hope that nobody is expecting Japanese or German standards from Bitmain whereby nothing will go wrong at all, I will be happy with something that is as good as S9, although I highly doubt I will ever touch Bitmain gears ever again.