Yes, the 230V/133V is your 3-phase output.

You should connect your miners like this.

A,B legs to miner 1
B,C legs to miner 2
C,A legs to miner 3

Then continue to rotate like this until maximum power is reached. I would not use the neutral. I would not use the single phase output.

What is the purpose of a PDU if not for remote switching?

I just use a 20 amp 2 pole breaker to drive a duplex 20A 250V outlet for each pair of S9s.
I use 12/2 with ground wire cable. It has a yellow jacket conforming to the newest jacket color code. 

All the parts are cheap and readily available from Home Depot, Lowes or ebay.

If you need to add more S9s, just add a sub panel. These are also cheap and off the shelf available.

I usually measure .97 PF on APW++ PSUs from Bitmain.

You can avoid almost all harmonics issues by avoiding non-resistive loads that use the neutral.

So, run all your power supplies, pumps and fans on 240V, but the coffee pot is fine on 120V.

It certainly does sound like you're loosing powe

If you have a voltmeter you can measure the 12VDC power lines. Just put your black lead into the back of any of the power connectors with a black wire and the red meter lead into the back of any power connector with a yellow wire. You should measure >12.0 VDC. Just leave the leads in and keep an eye on the voltage shown on the meter. You can set the meter to measure the minimum voltage if that feature is available.

Let it run until it shuts off, then see if the 12V power supply voltage dropped at any time below 12.0 V.

If it did. You can try to run the miner with two of the three hash boards powered. Just remove all 3 of the power connectors from one of the hash boards and try again.

If you don't have a voltmeter, now is the time to go get one. 

Length doesn't matter. haha

If you think about it, a fuse is very short yet it will burn in half when it's rated current is exceeded. Never, ever use a conductor under the size required for the protector.

I too ran out of positions in my panel. I just got several of those double breakers to consolidate my home loads onto fewer positions. So each single pole breaker has two output circuits. This opened up enough stab positions for me to install my 20A, 2-pole breakers. You have to buy two of those dual breakers to make room for one 2-pole breaker.

If you try to use 10-gauge wires they probably will not fit into your receptacles.

I'd suggest you either consolidate using dual breakers as described above or install a sub panel. I added a 125 amp sub panel once I grew past the original panel.

---

(Moderator's note: This post was edited by frodocooper to trim the quote from mineintx.)

The numbers shown in your original post do not look like measured data. I was guessing these were calculated values instead.

If you can, actually measure the voltage and the current and perhaps the circuit breaker and conductor temperatures. This information will help me determine if this should be a concern or not. Once I know some measured values, I can determine if this condition is typical or if you should be looking for other problems.

So a clamp on ammeter with volt meter and an IR temperature meter would be very useful.

Your description does have me concerned. "All my breakers get warm to hot to the touch" is a description that is abnormal. My first method of diagnosis is to touch things to feel how much heat exists. All my cords and breakers are just barely warm enough to feel any warmth if I really concentrate on it. If I ever feel more than that, I measure, double-check, inspect and test everything.

I'm guessing again the calculated current is based on the specs shown on Bitmain's site which can result in underestimation of current. Which will likely result in an overloaded circuit. If you have over 80% load on the circuit, it is overloaded by definition.

When you measure the current on the three phase breaker, please be sure to measure all the current going through each leg of the breaker. There are 3. Also measure the voltages between a-b, b-c and c-a. Three phase can be tricky. If you have 3 miners connected, one to each phase, the current per leg at the breaker needs to be known to understand the system loading.

Any 30A circuit loaded at only 21A should be very close to ambient temperature. It should be very difficult to feel any warmth on that circuit breaker. The mention you do feel warm to hot means something is not correct.



 

May I ask how you measured the current on your breakers?

If that doesn't work, I'd swap the controller with a known working controller.

New controllers with the latest firmware do handle both fixed frequency and variable frequency boards, even mixed on the same controller. The fixed boards will just use whatever you set as the initial frequency and the variable boards will go through the adjustment scan process.

The noise issue is very annoying, especially for neighbors. I was doing some calcs for my neighbors then ended up with this chart I thought may benefit others.

There will be differing opinions on this subject so I'll present my assumptions. I used the A-weighted Sound Pressure Level method to measure and calculate. The SPL scale is useful to determine how the human ear perceives sound. A-weighting aligns well with sounds under 45dB. I used that because while neighbors are outside to enjoy a quiet evening, ambient noise is likely around below that level so that's the relevant area. And my city uses this scale despite that fact their limit is really in the B-weighting range. It's clear they did not have a qualified consultant help them to write the ordinance.

My local code limits noise to 55dBA but they are unclear how, or where they measure. Since they didn't specify, I'm guessing they will measure at my property line. This leaves me vulnerable to fines based on the opinion of whomever shows up to measure. I measured one of my S9s to be 73dBA at 1 foot.

Sound energy, like all unfocused energy, dissipates spherically in all directions equally. When this is measured linearly, like from yours to your neighbors house, the sound energy is reduced by the inverse proportion law, 1/r, where r is the radius of the sphere. So if you measure the SPL from 1 foot away, it allows us calculate the SPL by knowing how many feet away your neighbors ears will be. When threatened with sound complaints, it's good to know your city's ordinances.

Each time you double the number of miners, the SPL level increases by 3dB. Each time you double the distance it decreases by 6dB. So to use the chart, the quantity of miners is in the left column and the distance in feet is in the top row. These numbers increase logarithmically to match sound energy characteristics.  

If you start by knowing your sound level tolerance and the distance to whatever you need to limit the noise to, just go down the distance column to find the dBA level tolerable to see how many miners can be installed. Some interpolation may be required but you can easily get an idea of the sound level. Once the sound level diminishes down to ambient, the miner noise is "lost in the noise", literally.
                           
55   dBA Limit                                 
73   dBA at 1 foot                                 


This simple chart is useful but could be misleading. Be sure to consider anything would could reflect some of the sound energy toward the adversary instead of dissipating as assumed. Of course you could reflect sound energy away from the offended neighbor too.

Also note if you have an existing farm, you can measure the noise level to establish a baseline. If you double the miners you should see a 3dB increase. If you have a noise problem, you can reduce the level 3dB by halving the miners to satisfy the limit.

---

(Moderator's note: This post was edited by frodocooper to format the chart as a code blob.)

Oh, well that's different.

The NG generator will require an earthing ground which will be connected to your PSUs via the power cord. Just be certain all earthing connections are welded. This system will be sufficient to provide surge protection without an MOV protection scheme. I'm assuming there is no utility power supplied so there is no switch over between generator and utility.

The surge suppressor will not hurt anything if it makes you feel better.

Yikes!

Breakers don't trip and fuses don't blow when the wire served is too resistive to provide enough fault current. The wire inside the extension cord is the fuse in this circuit. Although the conductor size is rated as a 15A circuit, it is very long which increase the voltage drop. This added resistance may limit the total current available to below the trip rating of the protector. Voltage drop increases as current and conductor temperature increases. This circuit description sounds like it is unable to supply the full voltage when continuously and substantially loaded.

Many NFPA rules are broken here. The input of the transformer must be supplied by a circuit capable of meeting the maximum input power rating of that transformer times 125%. It does look like it needs a 30A, 120VAC circuit.

Extension cords can never be used for continuous loads outdoors. You cannot bury the cord because it cannot dissipate enough heat underground. You cannot leave it in the sunlight because it is not rated and approved as UV resistant. It is also not "protected". All feeder circuits must be protected from damage, like within a wall, conduit or overhead, or buried.

You could just swing by Home Depot or Lowes, grab a 20A 2-pole breaker for your panel and 100' of 12 gauge, 3/12 w gnd NM-B direct burial cable and a 250V, 20A duplex receptacle and associated hardware to get a 240V circuit to your miners. That would be far easier than what you've done with that transformer.  That thing would sell on ebay well enough.



 

Without network connectivity there's no way to get the kernel log.

Here are a couple actions you can take.

First, what you are describing is a typical start up process. The firmware does many self tests during the initial startup. It is testing many systems and adjusting for optimum performance. It typically takes about 20 minutes to start a new miner.

However, you should be able to connect to the network. I suggest you login to your router and write down every IP address that is reported as active with the miner off. Then turn on and connect the miner to your router, wait 2 minutes, then check your router for the new active IP address. Type the IP address in your browser's address line to open the miners GUI interface. Enter root and root for username and password.

Once you're in there you should be able to follow the instructions provided to get it setup.

If you're still stuck;

Do you have a multimeter? If not, get one. If you do, please measure the DC voltage between any yellow and black wire from the back of a hash board power connector. See if the voltage is >12.0VDC throughout the attempted startup cycles.

Try to start miner with only one hash board and the controller powered. If that fails, power off, disconnect that hash board and connect a different hash board. Attempt to start miner with only that hash board and the controller connected. If that fails, try only the third hash board.






 

I noticed they added a fix in the latest firmware to correct an error. They were looking for a middle temp sensor which doesn't exist on these hash boards. The fix applies an offset to that missing chip to take the variable to minimum instead of maximum as before.

Oh, the warranty sucks so I fully disassemble every new miner before I power it up. Failures are just too costly to risk it.

If you're going to be a miner you'll need some spares.

I ordered several APW3++, a few control boards, cables, and fans as spares. The shipping per unit is better if you buy a couple items at a time. Shipping is very high if you only order one unit of anything.

They said it would take 20 days but I had them in a week.

MM, this is awesome. Thanks for doing this.

I looked up one of the components in their latest PSU so I'll give you the data. There are two of these thermistors in it. It's labeled 5D-15.

HERMISTOR NTC 47D-15

Abstract: NTC 5D-15 MF72 Power NTC Thermistor MF72 Power NTC Thermistor. The MF72 series Power NTC Thermistors provide inrush current suppression for sensitive electronics. Connecting a MF72 in series with the power source will limit the current surges typically created at turn on. Once the circuit is energized the resistance of the MF72 will decrease rapidly to a very low value, power consumption can be ignored and there will be no effect on normal operating current.

Surge suppressors are typically MOVs.

Utility lines carry many surge opportunities with them. As you can imagine, the substation and typically the overhead distribution circuits are all outside. They are susceptible to lightning strikes and very commonly nearby lightening events. They also suffer phase to phase arc over events which cause the protector to open. Those protectors have a re-trip feature which will reconnect the line after an arc over occurs. Distribution circuits are also switched to another source and voltage adjustments are done by a switching event. Lots of opportunity. These surge events present a high voltage, high frequency but low energy signal on the utility circuit. Well, all except direct lightening strikes.

An MOV shunts when the potential difference exceeds it's voltage rating. An MOV is only a small component wired between the two input power lines. It can and will shunt (short) any voltage presented above the rating. However, it can only handle a very small amount of energy. This makes it a good device to mitigate the effects of utility surges, again except lightening strikes.

Surge suppressors typically have many MOVs to increase the amount of energy they can shunt. These are sized based on the electrical storm frequency and severity of the area. MOVs do not last forever so the type with failure indicators is nice to have. A single surge suppressor cannot be effective protection against near or direct lightning strikes. I typically specify one at the service entrance, another at each distribution panel and the final MOV contained within the equipment PSU . If there is not one in the PSU, that device gets a small surge suppressor. The result is three MOV devices in series to protect each critical load.

Almost all electronics have an MOV in their circuit. I took apart a Bitmain PSU and looked but I didn't find one. There are two thermisters for start up surge protection but I didn't find MOVs.

If they did have an MOV in the circuit I would say no you don't need one unless your genset is exposed to potential direct or nearby lightening strikes. Obviously, you need one because the PSU is not equipped just to cover the mile away lightening surges. If you want to protect against nearby strikes, you'll need 3 surge suppressors. This would be good to do if you are in an area prone to frequent electrical storm activity. You can study the "let-through" voltage and energy specifications to see how to properly size each suppressor.

Can you just put in whatever you found... LOL no. That will not protect your system from lightening induced surges.

Since you are already running miners on the circuit, you can just get a clamp-on ammeter to measure each leg at the 150A breaker. The limit is 120A. If any two legs are less than 113A you can add one miner between those legs.  

People are taking liberties with the specs. 13.5 TH/s S9 spec is 1323W +10% at 25C ambient assuming the PSU runs at 93% efficiency. These specs are Bitmain's calculation model, not done by the mean of statistically significant sample measurements. Please also consider the power increase required to operate in 40C ambient conditions for both the miner and the PSU.

Note their power number is power with tolerance of +10%. It does not say + or - 10%. The 0.098 J/GH is 10% lower than the mean. That's why they stipulate to "expect" +10%. They really wanted to be under 0.1 J/GH because that sells more miners.

When I read all this I do not expect to measure 1455W maximum per miner. And I don't. I measure 1530W mean, at 25C. I don't know how much power to expect at 40C since I haven't been there to measure. I am certain it'll be higher.

All that above to share a point. The 80% circuit limitation is there partly to cover for temperature and specification maximums. Because Bitmain chewed up 10% and didn't account for 15C over normal temps, the circuit is actually at its practical limit if calcs show 80%. All that said it's no big deal in my mind as the breaker will just trip if you overload it. That is, assuming you follow all the other rules.

I'm not sure but I can try. Go ahead and post some pics.

Hmm, your HVAC engineer will struggle a bit with this. I calculated 32 Tons sensible cooling. I'd probably spec a 40 Ton nominal unit to handle the heat load. Cooling units over 5 tons are typically 3-phase powered. And if you found one, I'd not want it as it would likely fail too soon. You'd need 2x 20 ton units to cool this beast. Usually 20 ton units also have a reheat circuit to control humidity which requires more power than the compressor circuit when it's on.

When designing a closed system, where there is no air exchanged with the atmosphere, I start with 2X the IT load as the input power requirement for the site. This rule of thumb guess is a decent starting point to know about how large the input circuit needs to be.

So in rough numbers, I'm thinking about 800A, 3-phase @208.

Interestingly, I don't think your sound issue is as bad as some imagine. It's bad, but not as bad. Sound energy follows a logarithmic scale not linear. So 2x 30dB sound sources = 33dB and 10x 30dB = 40dB, 100x 30dB = 50dB. I didn't see the sound specification on their website so I can't do the actual calc. but that will not really matter much. It's based on the bel scale which is 10 times the sound energy = +1 bel over baseline, so 10 decibel is 1 bel.

Also, sound energy, like all unfocused energy, dissipates spherically in all directions equally. When this is measured linearly, like from yours to your neighbors house, the sound energy is reduced by the inverse proportion law, 1/r, where r is the radius of the sphere. So if you measure the SPL from 1 foot away, it should be a snap to calculate the SPL by knowing how many feet away your neighbors ears will be. When threatened with sound complaints, it's good to know your city's ordinances. SPL measured at the property line is really the calculation you should be determining before investing in a substantial electrical system.