Pages:
Author

Topic: How to get a 220-240 k-watt-meter in the usa (Read 602 times)

newbie
Activity: 2
Merit: 0
February 04, 2018, 09:03:56 PM
#46
To make life easy, I took some pictures for everyone - it's going to vary depending on what kinds of cables you've got, but you get the idea:

http://www.analogx.com/images/gpu/img_1664s.jpg

And as a little bit of rig porn, here they are in production:

http://www.analogx.com/images/gpu/img_1663s.jpg

I mount them with nylon spacers on one of the PSU screws - works like a champ.

Hey Mark,

That's a damn fine setup you got there.
I'm wondering where you manage to source your D1800s and their respective cases? I'm wanting to migrate my three 6-card rigs from open-air to enclosed boxes for co-location.

Cheers,

Markurian
sr. member
Activity: 700
Merit: 294
February 03, 2018, 07:11:10 PM
#45
To make life easy, I took some pictures for everyone - it's going to vary depending on what kinds of cables you've got, but you get the idea:


Jesus, that's a lot of work. Why not just get a switched PDU?



Ahem...   I wrote up a solution....  Puwaha's Poor Man's Networked PDU: Using Smart Plugs and Awesome Miner

https://bitcointalksearch.org/topic/how-to-puwahas-poor-mans-networked-pdu-using-smart-plugs-and-awesome-miner-2866608
hero member
Activity: 687
Merit: 511
February 03, 2018, 07:02:35 PM
#44
Have you looked at 415/240? You can push the savings by not having step down transformers and getting transmission voltage to 415/240 service. Then you can get 24kw (usable) tripplite PDUs for around $43.75/kw. This give you switched per outlet monitoring and you don't need a step down to 208/etc (415 -> neutral = 240v). The only issue is the ISC, you need to use in-line fuses to reduce ISC at the PDU. Still, you're only talking another $50/kw for wiring, 480 distribution panels, fuses and a custom transformer at the multi-mw level.

Planning 415 for my next build.

Hmm, I haven't heard of that service level, but maybe they're calling it something different here.  For all of our mines we've been running 208v, primarily because it's MUCH easier to find 208v/1600a buildings than anything larger, and so our building cost is much cheaper.  The issue we ran into the few times we've looked into bigger buildings is that usually the power hasn't been used a few years, and the allocation at the power company is gone, so they won't/can't actually deliver that anymore (no capacity left at the closest substation).  We've kind of setup most of our buildouts so as much of the infrastructure as possible can move with us, as we've moved from buildings before - so generally speaking we add a couple panels 400a panels, we use 30a breakers, and those are wired directly to L6-30R.  Then we have high quality extensions that we run out to the racks.  On the PDU side right now we have relatively expensive ones, because we've only been using the Sonoff's for a few months, but in that time we've been sold on them.  Right now our case per Sonoff is basically $30 (that's the unit, wired with C13/C14 extension, reflashed with our modded firmware), and each of our machines is about 1kw.  We also run TED units on the panels themselves, so we're monitoring both at the miner, and also at the panel, so we have a really solid grasp of how power is being used.  We also even go through everything with a thermal camera once a month to make sure that nothing is running abnormally hot, etc - super easy to do, and can really save your bacon.

I do know one other large miner who is running 480v, since each leg is 277v he's able to run directly off of those, as most of the server PSU's are tolerant up to that voltage.  Pretty slick and he's been doing it for years without issue.
hero member
Activity: 1118
Merit: 541
February 03, 2018, 02:13:19 PM
#43
Jesus, that's a lot of work. Why not just get a switched PDU?

Actually, I've got a local friend who I helped setup a jig and a bunch of the tools who does them for me - I've been pushing him to offer them up on the forums for other people who are interested, so hopefully he will at some point.

The big advantage over a traditional PDU (and you can't see them, but we have Tripplite and APS ones in the background), is that it also gives you fast watt level per port monitoring.  Cost-wise it means we can skip buying an ethernet enabled port-metering PDU (which are expensive), and instead use a bunch of these and cheap distribution PDU's.  Plus with the port-metered ones I've worked with are generally low resolution (like .1 amp) or have slow update rates (like once a minute) - and they also tend to use SNMP instead of JSON/HTTP, which is really much easier for us to work with.

What we've made instead is like a single distributed PDU - so we can monitor per outlet live all the device, and define more advanced rules.  I also use the TH16 as well, which is their switch that you can add a temperature sensor onto, and scatter them around to give us a realtime view of the temperatures in the building, plus can use these to automate all of the cooling.  For sub $100 you can completely automate your evap system - one TH16 for the dump pump on the evap cistern, one TH16 on the recirculating pump, one TH16 on the fan.  Then I can do rules that you wouldn't typically be able to easily do - like I can program the system to dump water when it goes above a certain temperature, so really like a partial dump as the city water will then cool down the remainder in the cistern.  Really it's pretty amazing what you can cobble together using these things.

It seems like you have a pretty substantial setup..

Have you looked at 415/240? You can push the savings by not having step down transformers and getting transmission voltage to 415/240 service. Then you can get 24kw (usable) tripplite PDUs for around $43.75/kw. This give you switched per outlet monitoring and you don't need a step down to 208/etc (415 -> neutral = 240v). The only issue is the ISC, you need to use in-line fuses to reduce ISC at the PDU. Still, you're only talking another $50/kw for wiring, 480 distribution panels, fuses and a custom transformer at the multi-mw level.

Planning 415 for my next build.
hero member
Activity: 687
Merit: 511
February 03, 2018, 02:01:21 PM
#42
Jesus, that's a lot of work. Why not just get a switched PDU?

Actually, I've got a local friend who I helped setup a jig and a bunch of the tools who does them for me - I've been pushing him to offer them up on the forums for other people who are interested, so hopefully he will at some point.

The big advantage over a traditional PDU (and you can't see them, but we have Tripplite and APS ones in the background), is that it also gives you fast watt level per port monitoring.  Cost-wise it means we can skip buying an ethernet enabled port-metering PDU (which are expensive), and instead use a bunch of these and cheap distribution PDU's.  Plus with the port-metered ones I've worked with are generally low resolution (like .1 amp) or have slow update rates (like once a minute) - and they also tend to use SNMP instead of JSON/HTTP, which is really much easier for us to work with.

What we've made instead is like a single distributed PDU - so we can monitor per outlet live all the device, and define more advanced rules.  I also use the TH16 as well, which is their switch that you can add a temperature sensor onto, and scatter them around to give us a realtime view of the temperatures in the building, plus can use these to automate all of the cooling.  For sub $100 you can completely automate your evap system - one TH16 for the dump pump on the evap cistern, one TH16 on the recirculating pump, one TH16 on the fan.  Then I can do rules that you wouldn't typically be able to easily do - like I can program the system to dump water when it goes above a certain temperature, so really like a partial dump as the city water will then cool down the remainder in the cistern.  Really it's pretty amazing what you can cobble together using these things.
hero member
Activity: 1118
Merit: 541
February 03, 2018, 10:48:43 AM
#41
To make life easy, I took some pictures for everyone - it's going to vary depending on what kinds of cables you've got, but you get the idea:


Jesus, that's a lot of work. Why not just get a switched PDU?



https://www.cyberpowersystems.com/products/pdu/switched/

Switched PDU's are too expensive and usually is not for huge loads, 10a, 20a and 30a for so much money is not worth. Sanoff for $10 for 16a is cost benefit.

I guess I'm looking at it more long term. Being able to auto reboot down miners, send notifications when necessary to physically inspect certain miners for issues, optimize your miner without needing to stand near it to do so, and monitor 3 phase balances.. Ya, mining hardware changes, but since i've been around in 2011, mining as been a thing, and it appears it will continue being a thing. Might as well build decent facilities and minimize manpower requirements.
sr. member
Activity: 2142
Merit: 353
Xtreme Monster
February 03, 2018, 10:39:20 AM
#40
To make life easy, I took some pictures for everyone - it's going to vary depending on what kinds of cables you've got, but you get the idea:


Jesus, that's a lot of work. Why not just get a switched PDU?



https://www.cyberpowersystems.com/products/pdu/switched/

Switched PDU's are too expensive and usually is not for huge loads, 10a, 20a and 30a for so much money is not worth. Sanoff for $10 for 16a is cost benefit.
hero member
Activity: 1118
Merit: 541
February 03, 2018, 10:10:46 AM
#39
To make life easy, I took some pictures for everyone - it's going to vary depending on what kinds of cables you've got, but you get the idea:


Jesus, that's a lot of work. Why not just get a switched PDU?

sr. member
Activity: 2142
Merit: 353
Xtreme Monster
February 03, 2018, 09:07:44 AM
#38
TP Link HS110's say they only work on 120v on the back of the units, but they do work on 220v as well.

Only the EU version 230v and its shucko socket, for ground to work you will need an adapter or you do yourself. That sanof is better cause you can wire them no adapter is required, however I wonder what is the maximum cable size you can fit on that sanof, 16 amps at 230v requires at least a 2.5 cable line size.

I do not trust Chinese products, so to see if 16amps will work 24/7 only with tests, yeah you can use with up to 10 amps, most things function without any problems, now a constant 16amps, that is what I really like to see a product to do.

......

Have you tried a 16 amps constant with them?

As far as I could see you use one for each computer, have you tried let's say 3 computers? That will be close to 16 amps.
newbie
Activity: 50
Merit: 0
February 03, 2018, 07:51:20 AM
#37
I've had multiple people tell me the kill a watt US version will work 240 with no issues, so I'm interested in this conversation.  I found this link https://electronics.stackexchange.com/questions/90805/how-to-calibrate-a-kill-a-watt-meter-and-understand-its-hidden-service-menu
in that article it looks like the hidden service menu has an option to recalibrate the kill a watt to 220 instead of 110.  Maybe philip should try to change the setting mentioned and redo his test to see if it then shows the correct voltage? 
One of the people that told me just to use adapters and the kill a watt on 240 mentioned he has been using them in his mining farm for over a year with no issues on 240.  confusing...
hero member
Activity: 687
Merit: 511
February 03, 2018, 01:41:00 AM
#36
In this case, wattage is pretty much irrelevant. They're direct pass through devices that don't 'use' power and could overheat/melt. Similar to the way that wire is rated in current/voltage and not wattage.
13a at 120v would be just as hazardous as 13a at 240v to the kill-o-watt.
I tested up too 12a. I wouldn't run them long term at that. Amperage = heat = bad.

All of my kill-o-watts were used at a friends house, who is a master electrician. He setups the power, space, racks, etc. I help with the PC side of things and actually finding cards...

You're definitely wrong about wattage being irrelevant and the Kill-a-Watt being an indirect current sensor - it's most definitely a resistive - here's a detailed teardown of it, at the time he talks about how it measures power:

https://youtu.be/zZFrRiCdfT8?t=6m30s

You'd be right if it was an indirect sensor like what you're describing (that uses a current transducer and measures the magnetic field of the power moving through the wire), but those are generally much more expensive to make than the simpler ones.  Your buddy just probably assumed it's reading current the same way all his equipment does - but if you show him this video I can assure you he'll change his tune - an electrician just wouldn't typically ever use resistive measurement tools since they're the cheap way of doing things.

Really, all of the consumer ones I've taken apart (including the Sonoff's) use the resistive approach to measure current - it's just cheap as hell and has nice linearity which makes it really easy to implement.

Although from what philipma says, it sounds like there are multiple variations of the Kill-a-watt, with potentially very different internals, but I would still bet they're all resistive.
legendary
Activity: 4326
Merit: 8950
'The right to privacy matters'
February 03, 2018, 12:31:41 AM
#35
It looks like any of the meters with the 'Intertek' branding on the back works, while the other ones don't.


All of my p4460.01 meters were bought within the past year from various places, all have the intertek branding on them and read 240v accurately.



And this thread has solved the 240 volt meter issues I had.  You were very helpful thanks.
member
Activity: 126
Merit: 10
90*c is good, right?
February 03, 2018, 12:24:18 AM
#34
It looks like any of the meters with the 'Intertek' branding on the back works, while the other ones don't.


All of my p4460.01 meters were bought within the past year from various places, all have the intertek branding on them and read 240v accurately.

legendary
Activity: 4326
Merit: 8950
'The right to privacy matters'
February 03, 2018, 12:03:16 AM
#33
So the p4400 doesn't work and the p4460 does, Undecided

no really weird

but I found all three usa ones.

one is model p3 4400  it does not work

one is model p3 4400.01 it works a little low but works.

I found an intek  looks just like the UK model but usa plugs and it is spot on

I am uploading 2nd video.

So it shows us some usa work some don't.


Your video will be live at: https://youtu.be/qmPwbXRORaA

so testing  shows this to work best


https://www.amazon.com/Voltage-Display-Monitor-Analyzer-Electricity/dp/B0755FFLP1/ref=sr_1_fkmr0_4?

within ½ volt of my UK model and of my DMM

A good money saver for all to know that want to measure 240 volts.
member
Activity: 126
Merit: 10
90*c is good, right?
February 02, 2018, 11:58:50 PM
#32
So the p4400 doesn't work and the p4460 does, Undecided


Or is it because your p4400 melted from over current?  Tongue
legendary
Activity: 4326
Merit: 8950
'The right to privacy matters'
February 02, 2018, 11:51:00 PM
#31
Okay first video up. Your video will be live at: https://youtu.be/ytQZXXviQZw


I found a second usa meter  same label  it does  240 volts and 120 volts

wait for second video
member
Activity: 126
Merit: 10
90*c is good, right?
February 02, 2018, 11:38:18 PM
#30
whitrzac, I don't doubt what you're saying that it works at 240v - my issue is how accurate it is.  At 120v it was like +/- 5% (so 10% error), at 240v it might be 2x of that, and +/- 10% is better than nothing, but it isn't terribly useful depending on what you're trying to do with it.

A better test would be to put it under some known load, like a heater or something (resistive load) that is rated at like 1000w - generally speaking those manufacturers are pretty close in their loads (the higher the better).  I also usually like to test things against inductive loads as well (like a fan or really any motor), then you get a sense if they run into any issues with different types of devices.  With a 10% margin of error, you might be looking at 160+ watts error in what's being displayed... If you're interested in just ballparks that it's close to what you're after, then it's probably fine - but where it becomes more problematic is the different ones might could end up returning different values.  

Another interesting thing would be to put all 4 of your Kill-A-Watts in series, and look at the reported watts... That also would give you some idea of the variance you're going to see.  I'd be interested to see that if you're so inclined to do it - I only ever had one Kill-a-watt, so I never really looked at the device variance - but I've literally got hundreds of Sonoff's and once calibrated they're all pretty consistent.

gotminer, believe it or not, it's probably safer to run these things at 240v than 120v, depending on how they're doing their current sensing.  In a wildely general statement, the higher the voltage, the better...  That's why I said it's probably more likely to be an accuracy issue than a safety issue running these things out of spec.


The 1st one we tested was ~2% off on amperage, IIRC, after that we never really checked. They were accurate enough for us.

IIRC they can be calibrated, but I haven't looked on how to do it.

I forgot the exact terminology he used, but they use the same 'method' as a clamp meter does to read current/voltage.

Since I've added more rigs, we've gone away from kill-o-watts and we setup a sub panel with a meter socket... A bit overkill, but they're cheap and work.
member
Activity: 126
Merit: 10
90*c is good, right?
February 02, 2018, 11:34:02 PM
#29
I've tested them up to ~10-12 amps at 240v without issue. I do not run them long term at those current levels.
I have 3 of them that have been used at ~6 amps and 240v for over 6 months without issue. The KWH recording, ect are all accurate.

Yikes!  Be super careful with that - I can 100% assure you that they will not support safely higher than their reported wattage.  12a/240v is 2880w, that's more than 1000w above the rated specs - although your 6a one is totally within safe margins... I would treat any of these devices with the same rules as wiring, meaning like any electrical wiring don't exceed 80% of rated values, but your 6a/240v is spot on for that and you should be able to safely run at that indefinitely.

In this case, wattage is pretty much irrelevant. They're direct pass through devices that don't 'use' power and could overheat/melt. Similar to the way that wire is rated in current/voltage and not wattage.
13a at 120v would be just as hazardous as 13a at 240v to the kill-o-watt.
I tested up too 12a. I wouldn't run them long term at that. Amperage = heat = bad.

All of my kill-o-watts were used at a friends house, who is a master electrician. He setups the power, space, racks, etc. I help with the PC side of things and actually finding cards...
legendary
Activity: 4326
Merit: 8950
'The right to privacy matters'
February 02, 2018, 11:30:10 PM
#28
Standard USA spec Kill-A-Watts read 220v just fine. Just dont try to pull more than 8 amp through them.

They work fine on 240v up to 12+a...

Kill-a-watt's on page shows the operating range is 115v-125v:

http://www.p3international.com/products/p4400.html

I looked at all their other 'energy saver' products, none support > 130v.  Philip is on the money - plus he wouldn't have gone through all that effort to get EU ones if he could easily get US ones.

Some people are just that simple.

I have 4 p4460s that have been used for over 6 months on 240v without issue.

Post pictures when your shit burns to the ground.  Lol ... Kidding, but I'm not joking when I say that I would take Phillip's word over most people who post in this forum.


Okay. I am good but why would someone say his work?

I am looking for all my USA meters.
I have a few of them.

So far the I have from reads normal for 120 volts and reads crazy dumb numbers using 240 volts.

With no load it gets 173 volts. The Uk reads 238 .

Under load it gets more screwed up but it works fine for 120 volts.

I need to find other USA meters. I have them.
hero member
Activity: 687
Merit: 511
February 02, 2018, 10:52:39 PM
#27
I've tested them up to ~10-12 amps at 240v without issue. I do not run them long term at those current levels.
I have 3 of them that have been used at ~6 amps and 240v for over 6 months without issue. The KWH recording, ect are all accurate.

Yikes!  Be super careful with that - I can 100% assure you that they will not support safely higher than their reported wattage.  12a/240v is 2880w, that's more than 1000w above the rated specs - although your 6a one is totally within safe margins... I would treat any of these devices with the same rules as wiring, meaning like any electrical wiring don't exceed 80% of rated values, but your 6a/240v is spot on for that and you should be able to safely run at that indefinitely.
Pages:
Jump to: