Topic: Mining rig burning smell! - page 2. (Read 632 times)

I stopped my rig as soon as i read your helpfull post!!
So do you mean i can use only 2 sata/molex cable i have per psu(so 4 in total) only for powering 2(so 4 total) gpus?
That is one cable per gpu, despite any sata/molex cable has 4 molex connectors?
So i should buy 3 of that sata to molex adapters, to use the other 3 sata connectors(6 in total with 2 psus) i have per psu, to adapt those cables for the risers molex connection?
Thanks for your quick salvation of my rig!!

no no no no no no.

a molex can only do 1 riser safely
2 risers are a risk
3 riser = meltdown

Let me look at that psu a bit more closely

and post a photo of your risers


here is your psu

https://www.evga.com/products/product.aspx?pn=220-GT-1300-X1




CABLE LENGTH

ATX Cable   1x 600mm (m)
EPS Cable   2x 750mm (m)
PCIe Cable   4x 750mm (m), 2x 750mm, 900mm (m)

SATA Cable   4x 550mm, 650mm, 750mm (m) let me find an adapter for this found it https://www.amazon.com/Power-Cable-Adapter-Female-8-inch/dp/B07BQFKTG7/ref=sr_1_3?

Four-Pin Peripheral Cable   2x 550mm, 650mm, 750mm, 850mm (m).    these will do 1 riser each so 2 risers for 2 cables
Floppy Adapter   1x Molex to FDD Adapter
AC Power   1x 1500mm (m)
Modular Cable Type   Full
Molex to FDD   1x
CONNECTOR QUANTITY

24 Pin ATX   1x
EPS (CPU)   2x 8pin(4+4)
PCIe   8x 8pin(6+2)
SATA   12x
Four-Pin Peripheral   8x
Floppy   1x
POWER SPECS

AC Input   100-240V~, 15-8A, 60/50Hz
Rail   +3.3V   +5V   +12V   +5Vsb   -12V
Max Output   25A   25A   108.3A   3A   0.5A
130W   1300W   15W   6W
Total   1300W
ADDITIONAL DETAILS

Operating Temperature   +50° C
Japanese Capacitors   100%
ECO Mode   Yes
MTBF   100,000 Hours
Net Weight   9 lbs
Size   86mm (H) x 150mm (W) x 180mm (L)
Approvals   TUV, CB, CE, FCC, RCM, cTUVus, WEEE, RoHS
Compliance with ErP Lot 3 2014 Requirement
Fan Size / Bearing   135mm Fluid Dynamic Bearing
Product Warranty

This product comes with a 10 year limited warranty with registration.
EVGA offers several warranty solutions for our customers to choose from. Please refer to the suffix of the part number and our Warranty Terms for details.



So you have 4 sata cables use one of these on 1 sata cable

https://www.amazon.com/Power-Cable-Adapter-Female-8-inch/dp/B07BQFKTG7/ref=sr_1_3?

so 3 of these packages are nine adapters  you have 2 psu's with 4 sata cables you can do all of those cables

so 4 sata to molex
2 molex    and first psu does 6 risers with 6 cables Should be safe

The first psu arrived and i spent some hours to assemble everything with only 5 gpus that work as ever, luckily.
Tomorrow the second psu will arrive and i will install the other 5 gpus.
Now i want to know approximately the power consumption of all the system in addition to the gpus, so the mobo, the risers, the ssd, the ram ecc..
So i can have an idea about the total power drain of my rig other than all the 10 gpus.
I took 2 of these:
https://www.amazon.it/gp/product/B09JP7QPSV/ref=ppx_yo_dt_b_asin_image_o01_s00?ie=UTF8&psc=1
I think these psus are perfect and now i have been able to arrange all the cables neatly and not tangled between the gpus.
I only have a doubt about the risers connections, because i used 3 molex connectors located along a single cable, but I have heard that it is better not to go beyond the two cable connectors per single cable.
Right?
However it's confusing because i found in the psu package, 2 molex cable with 4 connectors per cable.
Thanks again!

GPUs are powerful hardware products. These products do not deteriorate immediately. The routers you use to connect the GPUs to the system can cause problems. Therefore, enable each gpu one by one.

If any of the video cards is faulty, then you can get fireworks after turning on the mining farm. The motherboard for mining is more expensive, so I test video cards on a separate PC. If you think that all your video cards are working, then do as you please.

Yes, i took just this one, because the first buy had a problem with the crypto card and so i looked better down the page and i noticed just that one.
Moreover it's 1300Watt.
Thanks however!

Thanks, but which is the differences if, at first glance, i try to start my rig as is and i check if all of my gpus mine whitout issues?
After this easy and straight check, if i notice some differences in the hashrate of some gpus, then and only then i go to deepen, only for that specific gpu.
What do you think?

The burnt smell of the power supply warned you. You turned off the system before the system saw a bigger malfunction. I think that there are no problems with the remaining components of the system. When your new power supply arrives, you can control the system by running the GPUs one by one.

If you are doing a video capt check, then you should not start all video cards at once. When I buy a used video card, I always test it on another computer before installing it in a mining farm. I install new video cards without checking.
If you have doubts about the performance of video cards, then it is better to spend a few hours checking them on another PC.

My rig is on HiveOs, so i go for the mining check.
So the gpus can also turn on, but then have problems while working?

To check video cards you need a working computer. Run furmark for 15 minutes at maximum settings and after the test you can be sure that the video card is working. Or better run mining on each video card for 15 minutes.

Thank you!
I think the issue has been such as a forced power off, because there wasn't any strange sounds or smells.
Moreover, the mobo is working, so i think there aren't other damages.
I hope i'm right.
But is there a chance all my gpus were killed forever?
Can i check the gpus one by one, connecting them in turn to the working power supply, just to check they power up?
If i see the gpu fans start rolling and the leds light up, does is it a good proof of work?

maybe your gpus are good. my fingers are crossed for you.

Ok, my 2000W psu is dead!
Just during a reboot of my rig, anything turned off, but no strange smells.
Then i tried to temporary connect another psu to the mobo only and the system turned on, so it's surely the 2000W psu that stopped working.
Now i have to wait for the new psus to arrive, before restart my rig.
Do you think that some of the gpus may be damaged after this issue?
The model of my broken psu is:
"Crypto Mining Xpert 2000W" and it is Chinese because I see several writings with Chinese characters.

I have a Gigabyte H110-D3A mobo, but i'm using 2 pcie 1 to 4 splitters and i increased the pcie bandwidth to GEN3 in the bios, so i can use almost all the available pcie band for the splitters and so for the gpus.
I know i should get a new mobo for mining, with 12 pcie risers slot, so i can exploit all the power of my gpus.
If you can recommend me a good model to buy, would you give me another help.
Now i'm mining with a reduced gpus power limit, untill i install the new psus, so for now my hashrate has been decreased of about 150mh/s.
At full power, the hashrate of my rig, is at about 550mh/s, with the proper overclock settings for any gpu.
With the new gpus i plan to achive about 680mh/s, but the power consumption will be very high and i have already asked for a power increase of my elettrical system, from 3kw to 6kw, otherwise i wouldn't be able to live in my house anymore.

Are you running all these cards on a single motherboard? What type of motherboard you using? You need better air flow in your mining room or shed because those 3070ti generates high temp, should have bought 3070s instead, anyways two 1200watt will do well.

that is a good one.

it has six slots or jacks that will take pcie cables or cpu cables.

so 1 cable cpu goes in one slot or jack

and you can use five pcie cables for pcie

it will allow for 5 cables either single or double ended .

you can draw a steady 850-900 watts.

if you have two of them one can do the cpu and five cables for gpus
the other can do all six cables for gpus.

so you can run. 11 gpus each pulling about 150 watts or less.

I got a link for spare pcie cables I will look for it.


https://www.amazon.com/gp/product/B0979Y7F86/ref=ox_sc_act_title_1?smid=A2ZUCH4AFFNDHF&psc=1

they should be good for the hx1200

if i remember the psu will come with 2 cpu cables and 4 pcie cables

so you will need a few of these cables to fully use the psu’s

do you have a kill-a-watt-meter?


https://www.amazon.it/multifunzione-monitoraggio-apparecchiature-elettriche-sovraccarico/dp/B08V4RQZQL/ref=sr_1_16?

they help to know how you are drawing power.

the psu you purchased should stay under 900 watts on the meter for long term runs.

Ok i bought this:
https://www.amazon.it/dp/B06W2GJDRJ?psc=1&smid=A11IL2PNWYJU7H&ref_=chk_typ_imgToDp
What do you think?
Thanks again for your great help guys!

If you add the GPUs you plan to buy, these power supplies that you say you will buy will be enough for you. Since there will be 12 GPUs in total, make the distribution so that 6 GPUs are connected to each power supply. Also get from power supply multiplexers.

https://it.aliexpress.com/item/32822981755.html?spm=a2g0o.productlist.0.0.38914395g2xkRe&algo_pvid=d4f3b3f6-f922-4f19-84f5-fceba1602fe5&algo_exp_id=d4f3b3f6-f922-4f19-84f5-fceba1602fe5-0&pdp_ext_f=%7B%22sku_id%22%3A%2264914293098%22%7D&pdp_pi=-1%3B5.34%3B-1%3B-1%40salePrice%3BEUR%3Bsearch-mainSearch

I was thinking the same thing.  Typically you want to leave 15% is what I always heard.  That would mean you wouldn't want to run more than 1700 watts on that PSU.  With 10 GPUs and likely a motherboard running on that, I can't imagine there wouldn't be problems.  Not to mention all the loss from what I can assume is a mess of cables.  I'd definitely get a new GPU going, and it wouldn't be a bad idea to think about replacing the single 2000 watt PSU with 2x 1300 watt PSUs to take some of the load off.  I'm not sure I'd risk hooking the burned PSU back up in any capacity.  You don't want to risk it damaging something else.