277/480 WYE 4-Wire Woes

winnie said:
.
Click to expand...

Dear Jon,

Thank you so much for writing.

I know Ohm's Law pretty well. Most assuredly amperage goes down when voltage goes up; that's why generating stations send electricity out at such high potentials. I was just going by the readings I was taking. Each machine is pulling about 10 amps at 277 volts; that's 2700 watts.

And too I know one isn't supposed to max a 20 amp breaker, much less a 400 amp breaker. Schools in on the harmonics thing; I have no experience with that, but I'm a quick study. Hell, I've been working on gravity for 30 years and I've got half the secret, but this isn't the forum for that.

Email me and I will reveal it. bluescarecrow2000@gmail.com


Best,
Mike
 
Ahh, I didn't see that the power changed.

How are you measuring power? Because if the power consumption is going up with increased voltage that is a huge problem right there.

Since the job of the power supply is to deliver constant voltages to the miner, if power consumption is going up that basically means extra power being lost as heat in the power supply.

However there is a good chance that your current and power readings are being confused by harmonics, thus my question about how you are measuring current and power.

If real power consumption is actually going up, than that is stronger incentive for your customer to fix the problem, since they are paying for watts not going to hashing.

Jon
 
winnie said:
Ahh, I didn't see that the power changed.

How are you measuring power? Because if the power consumption is going up with increased voltage that is a huge problem right there.

Since the job of the power supply is to deliver constant voltages to the miner, if power consumption is going up that basically means extra power being lost as heat in the power supply.

However there is a good chance that your current and power readings are being confused by harmonics, thus my question about how you are measuring current and power.

If real power consumption is actually going up, than that is stronger incentive for your customer to fix the problem, since they are paying for watts not going to hashing.

Jon
Click to expand...
Agreed here.
My original suspicion was phase imbalance and high harmonics is causing bad readings on his “x”(whatever he is using)
 
winnie said:
.
Click to expand...

Dear Jon,

You asked how am I measuring power? I'm just doing a reverse pie... amps times volts.

All I have at my disposal is a Fluke, an Amprobe Ultra and a Cen-Tech Infrared Thermometer.

If Navopache can't re-tap, then that throws a really big monkey wrench into the works. If Herb (the boss) goes Delta, then all of the breakers will have to be replaced. The neutrals will have to be removed and replaced. 2200 divided by 240 = 9.16 per machine x 20 (machines) = 183 amps per panel. That's 91.5% and that's high.

There are 16 panels.

I was drawn into this by my friend Terry as he knows I'm an electrician; and he has about 40 mining machines in the pool of his own. I haven't done electrical work in years, but it's like riding a bike I guess.

The piping in this place is a maze 3" pipe with four and sometimes five (OMG) 90's and that's where you'll find the 1000 kcmil along with other smaller conductors. I haven't checked the fill table for this, but I'm doubtful that it is ok. The electricians scraped off some of the insulation of the 1000 kcmil and repaired it with black duct tape. They used their car instead of a Greenlee capstan. That speaks volumes. I would have loved to see what they used as a pulling head; and I didn't see any evidence of a pulling lubricant either. Yikes.

Thanks for writing Jon.


Best,
Mike
 
Would 100% rated breakers help? When I did some data center work several years ago, they specified 100% rated breakers on the services. Don’t know wether the engineers wanted extra cushion on the services, or it was to compensate for the harmonics.
 
I've never been involved with buck-boost at these current levels, but that seems a reasonable approach. You might talk to one of the transformer manufacturers and see it that is a possibility.
 
HB1,

Thanks for writing.

Maybe a little of both. Data Centers want ro keep you as a customer. Not that you could go elsewhere. There are only two Data Centers in Phoenix.

Didn't know they even made 100% rated breakers. I'll keep that in mind if I stay on this job.

Thanks!


Best,
Mike
 
bluescarecrow said:
2200 divided by 240 = 9.16 per machine x 20 (machines) = 183 amps per panel.
Click to expand...
Mike thats not how you do load calcs on a 3 phase panel.
You add up all the VA, add 25% as its a continuous load, then divide by 480* SQRT 3, so if say we have 100 units that draw 2600VA a calc would look like this if they were a 277V load:

VA Each​
Quantity​
VA​
Amps​
2600​
100​
260000​
312.74​
​
​
65000​
78.19​
​
​
325000​
391​


But you need a 415Y240 system so that would look like 88 miners per 400A panel:

VA Each​
Quantity​
VA​
Amps​
2600​
88​
228800​
318.32​
​
​
57200​
79.58​
​
​
286000​
397.90​

Just my guess, cheers
 
It is not uncommon for data centers to operate the IT equipment at 416/240V 3 Phase. Depending on if the transformers are utility owned or owner owned, will control your options. A typical 240V rated power supply should not (if I was writing with my PE pen, I would say "shall not") operate at 277V.
You need some holistic design going on here, not Band-Aids.
 
winnie said:
In any case, the real problem is 240V power supplies at 277V, and you are on the right track to fix that.

Jon
Click to expand...
An Electrical Contractor friend of mine does the work for another one of these mining sites. Using 277 to supply this equipment is a common practice in that industry.
 
don_resqcapt19 said:
An Electrical Contractor friend of mine does the work for another one of these mining sites. Using 277 to supply this equipment is a common practice in that industry.
Click to expand...

Good info.

Is this a general comment that the industry plays very loose with proper device specifications (using 240V rated components at 277V), or a comment that these power supplies generally function at 277V and that the OP should be looking elsewhere for the problem?

I could imagine that 240V rated devices might usually work well enough at 277V, but be more prone to failure because they are right at the edge of their actual component tolerances.

Maybe the OP should just get the transformers adjusted to the minimum voltage they are capable of and see if things improve.

Thanks
Jon
 
don_resqcapt19 said:
An Electrical Contractor friend of mine does the work for another one of these mining sites. Using 277 to supply this equipment is a common practice in that industry.
Click to expand...
At this point might be a toss up as to cost,
If 900 new 240V power supplies = $117,000.00, I imagine 277 ones would cost more.
Then with shortages of semiconductors and electronic parts...

If the POCO has no other customers, on 3 of those 1000kW transfromers and if the tap changers can get it down to 415Y240, I bet those miners could run off three of the transformers. then keep one at 480 for the other stuff,
Still might cost a ton, but your not changing 900 computer power supplies, which who knows the labor on that.

EDIT: Wrong term, I guess its called 416Y240
 
Last edited:
don_resqcapt19 said:
An Electrical Contractor friend of mine does the work for another one of these mining sites. Using 277 to supply this equipment is a common practice in that industry.
Click to expand...
If the nameplate has a top end of 240V, I would suggest that no one should be recommending running them at a higher voltage then they are intending or are listed.
 
ron said:
If the nameplate has a top end of 240V, I would suggest that no one should be recommending running them at a higher voltage then they are intending or are listed.
Click to expand...
Agreed as winnie said:
winnie said:
the industry plays very loose with proper device specifications (using 240V rated components at 277V)
Click to expand...
277 is only widly used in the states now, most other places have standardized on 220/230/240
 
tortuga said:
At this point might be a toss up as to cost,
If 900 new 240V power supplies = $117,000.00, I imagine 277 ones would cost more.
Then with shortages of semiconductors and electronic parts...

If the POCO has no other customers, on 3 of those 1000kW transfromers and if the tap changers can get it down to 415Y240, I bet those miners could run off three of the transformers. then keep one at 480 for the other stuff,
Still might cost a ton, but your not changing 900 computer power supplies, which who knows the labor on that.

EDIT: Wrong term, I guess its called 416Y240
Click to expand...
The mining center my friend works at is not have issues with power supply failures.
 
ron said:
Don, it is a code violation to use the equipment outside of its nameplate.

Just because my brother in law down the street has flying splices in the wall without a box, and hasn't had a fire, doesn't mean that's what we should be suggesting.
Click to expand...
It would be for installed equipment, but in general cord and plug equipment is outside the scope of the NEC, even though Article 422 tries to make you think otherwise.
 
Don the AHJ for the NEC may be a government employed electrical inspector during construction up to the CO.
People whom work for the local fire marshal, OSHA, banks (loan underwriter), and an insurance inspector/investigator that I deal with all apply the NEC to entire facilities.
And very much look at cords, especially on jobsites.
Edit: but I dont doubt there are listed 277 computer power units made, it just that buying 900 of them seems uneconomical.
 
tortuga said:
Don the AHJ for the NEC may be a government employed electrical inspector during construction up to the CO.
People whom work for the local fire marshal, OSHA, banks (loan underwriter), and an insurance inspector/investigator that I deal with all apply the NEC to entire facilities.
And very much look at cords, especially on jobsites.
Edit: but I dont doubt there are listed 277 computer power units made, it just that buying 900 of them seems uneconomical.
Click to expand...
No matter who the AHJ may be, if they are citing the NEC, they can only cite what the NEC actually covers.
 
You must log in or register to reply here.
Top