Rotary Phase Converters, Transformers, Subpanels and Grounding

I know this is going to be a can of worms, but I think that this could be a generally useful thread if people provide good responses. So here goes...

I have some questions about grounding with regards to using a rotary phase converter, a transformer fed by a subpanel.

In my case, I have a subpanel with a grounding rod that is tied/bonded to the ground line run from the main panel. The neutral is only bonded to ground in the main panel and is run to the subpanel, but not bonded to ground in the subpanel.
From the subpanel, I have a 240V (about 246V on average) single phase run to my RPC (rotary phase converter). The output from the RPC is 3ph 240V, which I run to my low voltage 3ph machines, but also I run the 240V into a 3ph delta-delta transformer to step it up to about 440V 3ph.

The questions I have are in regards to grounding for the RPC and transformer.

Should I provide another grounding rod that I tie/bond to the ground from the subpanel?
Should I bond the low voltage side and high voltage side ground wires all together and bond it to the transformer frame and the RPC motor frame and bond it to a local grounding rod?

I know a number of people that are suspicious of grounding and never attach the ground wires for 3ph machines.

As a rule, you do what the NEC says, UNLESS your local jurisdiction says to do something different.

Usually local grounding is reserved for agricultural outbuildings served from a remote panel (particularly milking sheds and the like). But this is a place where local jurisdictions vary.

Generally, the equipment grounding conductor comes from the service panel and goes all the way to the machine. I cannot think of one good reason to not ground a machine, 3 phase or otherwise. It is an NEC and UL requirement, unless the item in question is "double-insulated".

Trying to separately ground the machine can make the grounding partly or totally ineffective for preventing shocks. The purpose of the equipment grounding conductor is two-fold: One, to prevent a dangerous voltage from developing on the machine. Two, to open the branch circuit protector if a fault develops.

A local ground rod may or may not be an effective ground. If the machine case becomes "hot" with mains voltage, there may be a significant voltage between the machine and the earth somewhat away from the machine. Ground rods may be up to 25 ohms resistance to earth.

The resistance between the machine and the "ground" or earthing means at the main panel may be too high to allow enough current to open the breaker. So the fault condition may continue for some time before other failures occur, or someone is injured by the voltage to ground.

Using the "green wire" prevents these problems to the best possible extent.

Your post is a bit unclear, but at third glance it sounds like your question involves the secondary of the step-up transformer, after the
rotary converter. If that is so, then the answer would be the secondary circuit is a separately derived service and would need to be 'bonded'
to the groundING (green wire) conductor in some way. Not clear if this would need its own ground rod, or if the one present at the sub-panel alone complies
with your local codes.

The NEC requires all the ground rods be tied together. The easy thing to do is ground/bond everything to the service panel and then out to the ground rod(s). The only point the nut. and the ground meet will be at the service panel after that the nut. and ground wire never join. Nut wire will be White and the ground Green. This means you need 4 wires for single phase and 5 wires for 3 phase if you need the nut.I have seen local inspectors that couldn't come up with the proper grounding.If in doubt contact a good industrial electrician...Phil

Phil in Montana said:

The NEC requires all the ground rods be tied together. The easy thing to do is ground/bond everything to the service panel and then out to the ground rod(s). The only point the nut. and the ground meet will be at the service panel after that the nut. and ground wire never join. Nut wire will be White and the ground Green. This means you need 4 wires for single phase and 5 wires for 3 phase if you need the nut.I have seen local inspectors that couldn't come up with the proper grounding.If in doubt contact a good industrial electrician...Phil

Click to expand...

Nut?

I’ve got 4 wires going from my subpanel to my RPC: hot, hot, ground, neutral
But, I only use the neutral for my 120V contactor, and I don’t have any neutral wire for any of my 3ph lines, just 3xhot+ground.

From a NEC perspective.

It appears from your post that your existing building wiring may be correct. You equipment wiring not so much.

The grounded conductor (neutral) and the equipment grounding conductor (ground) should only be bonded together at one point, and that point is to be at the first over-current device (main overcurrent device) at the service entrance.

At this bonding point it is also required to bond the required earth electrodes (ground rods, ufer, building steel). Interior metal piping systems and entering other utilities need to be bonded here as well, (phone, cable, antenna )

Downstream of this service bonding point, there shall be no other connections between the grounded conductor and the equipment grounding conductor. Such as your subpanel, the grounded conductor (neutral) should be isolated from the equipment grounding conductor (ground)

The purpose of the grounded conductor (neutral), is to carry normal current, unbalanced current from the center tapped supply transformer.

The purpose of the grounding conductor (ground), is to only carry non normal, fault currents and tie all bonded items to the same potential. There should not be any normal current flowing on the grounding conductor.

If your subpanel is located in a different building, it is required to have additional earth electrodes (ground rods), bonded to the grounding conductor (panel ground). The grounded conductor (neutral) shall not be bonded to the equipment grounding conductor (panel ground bus). No bonding screw in the neutral bus to the enclosure.

For the rods at the outbuilding, if you can prove that one rod has a resistance of 25Ω or less, one rod is sufficient. If you have no proper way to measure earth resistance, then two rods are required, spaced apart 8'. And with two rods, you do not have to meet any earth resistance standard.

The RPC system is directly connected to the 1Φ building supply system, and as such should be supplied and connected with an equipment grounding conductor. The output of the RPC should be extended with this same equipment grounding conductor, to all of its loads.

For a isolation transformer connected to the output of the RPC, the equipment grounding conductor should be extended to the transformer frame and enclosure. This will handle any fault currents from the primary side (supply side) of the windings.

The secondary side of the transformer (output side) is totally isolated from the input side and its grounding system. And therefore is required to be grounded. How it is to be grounded is dependent on the output coil arrangement. It is to be bonded to the building grounding system, and all downstream connected loads. Any faults from the output coils or its connected loads need a low impedance path, back to it's source, the output coil of the isolation transformer. If no direct low impedance path exists, there will be no way for the fault current, to return to the output coil. Breakers or fuses will not be able to interrupt this fault current, and the equipment load frame could stay energized. This is against the NEC, and could result in an unsafe condition.

If your isolation transformer output coil is a delta and only has 3 connection points, your only choice is to make it a corner grounded system. One of the output lines bonded to the building grounding system, just like the 1Φ supply system. It's called a separately derived system and has the same bonding and earthing requirements as the service. A corner grounded system is a rare bird today, and most utilities will no longer supply this type system. Most electricians have no familiarity with them and they have their own particular requirements, and for this reason not usually used.

A wye output coil is much more preferred, the center point of the wye would be the connection that gets grounded, instead of a corner, and would produce an output that is symmetrical with respect to the grounded conductor. This is the standard type system that most are familiar with.

A corner grounded system (non symmetrical) is normally detrimental to most any VFD. Electro-mechanical controls typically will have no problem with this type system.
And there are NEC requirements that are different from standard wye grounded systems.

SAF Ω

jim rozen said:

Your post is a bit unclear, but at third glance it sounds like your question involves the secondary of the step-up transformer, after the
rotary converter. If that is so, then the answer would be the secondary circuit is a separately derived service and would need to be 'bonded'
to the groundING (green wire) conductor in some way. Not clear if this would need its own ground rod, or if the one present at the sub-panel alone complies
with your local codes.

Click to expand...

You surmised correctly, the 240V 3ph output from my RPC feeds 240V 3ph machines and is input into my step up delta-delta transformer.
The 440V 3ph output from the transformer feeds my high voltage machines.

Ok, so it sounds like I have it right.
I have everything grounded.

I guess I was wondering if adding another grounding rod would be useful/good.

Thanks!

SAF said:

From a NEC perspective.

It appears from your post that your existing building wiring may be correct. You equipment wiring not so much.

The grounded conductor (neutral) and the equipment grounding conductor (ground) should only be bonded together at one point, and that point is to be at the first over-current device (main overcurrent device) at the service entrance.

Click to expand...

Thanks for your response!
My neutral is bonded to ground only at the main panel.
There is a neutral line from the main panel to the subpanel, but it is not bonded to ground in the subpanel.
My main panel has 2 grounding rods and the subpanel has a grounding rod.

I am running 4 wires from the subpanel to the RPC: 2x hot+ground+neutral (being used just for the 120V contactor).

SAF said:

At this bonding point it is also required to bond the required earth electrodes (ground rods, ufer, building steel). Interior metal piping systems and entering other utilities need to be bonded here as well, (phone, cable, antenna )

Downstream of this service bonding point, there shall be no other connections between the grounded conductor and the equipment grounding conductor. Such as your subpanel, the grounded conductor (neutral) should be isolated from the equipment grounding conductor (ground)

The purpose of the grounded conductor (neutral), is to carry normal current, unbalanced current from the center tapped supply transformer.

The purpose of the grounding conductor (ground), is to only carry non normal, fault currents and tie all bonded items to the same potential. There should not be any normal current flowing on the grounding conductor.

If your subpanel is located in a different building, it is required to have additional earth electrodes (ground rods), bonded to the grounding conductor (panel ground). The grounded conductor (neutral) shall not be bonded to the equipment grounding conductor (panel ground bus). No bonding screw in the neutral bus to the enclosure.

For the rods at the outbuilding, if you can prove that one rod has a resistance of 25Ω or less, one rod is sufficient. If you have no proper way to measure earth resistance, then two rods are required, spaced apart 8'. And with two rods, you do not have to meet any earth resistance standard.

The RPC system is directly connected to the 1Φ building supply system, and as such should be supplied and connected with an equipment grounding conductor. The output of the RPC should be extended with this same equipment grounding conductor, to all of its loads.

Click to expand...

I have a ground line going to my RPC.

SAF said:

For a isolation transformer connected to the output of the RPC, the equipment grounding conductor should be extended to the transformer frame and enclosure. This will handle any fault currents from the primary side (supply side) of the windings.

The secondary side of the transformer (output side) is totally isolated from the input side and its grounding system. And therefore is required to be grounded. How it is to be grounded is dependent on the output coil arrangement. It is to be bonded to the building grounding system, and all downstream connected loads. Any faults from the output coils or its connected loads need a low impedance path, back to it's source, the output coil of the isolation transformer. If no direct low impedance path exists, there will be no way for the fault current, to return to the output coil. Breakers or fuses will not be able to interrupt this fault current, and the equipment load frame could stay energized. This is against the NEC, and could result in an unsafe condition.

If your isolation transformer output coil is a delta and only has 3 connection points, your only choice is to make it a corner grounded system. One of the output lines bonded to the building grounding system, just like the 1Φ supply system. It's called a separately derived system and has the same bonding and earthing requirements as the service. A corner grounded system is a rare bird today, and most utilities will no longer supply this type system. Most electricians have no familiarity with them and they have their own particular requirements, and for this reason not usually used.

A wye output coil is much more preferred, the center point of the wye would be the connection that gets grounded, instead of a corner, and would produce an output that is symmetrical with respect to the grounded conductor. This is the standard type system that most are familiar with.

A corner grounded system (non symmetrical) is normally detrimental to most any VFD. Electro-mechanical controls typically will have no problem with this type system.
And there are NEC requirements that are different from standard wye grounded systems.

SAF Ω

Click to expand...

This is one of my concerns.
My CNC machines (440V CNC lathe, 230V VMC) have VFDs in them.
Empirically things have been working so far.
I just want to do a good, safe job.

Thanks!

rpseguin said:

Thanks for your response!
My neutral is bonded to ground only at the main panel.
There is a neutral line from the main panel to the subpanel, but it is not bonded to ground in the subpanel.
My main panel has 2 grounding rods and the subpanel has a grounding rod.

I am running 4 wires from the subpanel to the RPC: 2x hot+ground+neutral (being used just for the 120V contactor).




I have a ground line going to my RPC.




This is one of my concerns.
My CNC machines (440V CNC lathe, 230V VMC) have VFDs in them.
Empirically things have been working so far.
I just want to do a good, safe job.

Thanks!

Click to expand...

Not bad. But about that "Delta"...

Mindful of what SAF just summarized rather nicely for us all.. for "a while".. as-in writing the manuals on it, Northrop-Page for US Army Signal Corps IWCS FPTS sites, late 1960's ... then a "mostly" Telco career. "Site Bonding and Grounding" is a subject VERY dear to the collective hearts of those among us as had to keep a remote island's LNR satellite link and over-water microwave shots operational - routine lightning strikes and all - whilst perched at the highest point - or near-as- dammit - of a hunk of rock surrounded by saltwater!



Soooo ..the MAIN thing I did differently from your rig was... "minimize" presence of "Delta" to RPC (array..) out and/or Phase-Perfect out. Short runs - very - and well-protected. Both are Delta-out, there's nothing else for it.

"Howso" involved a used-but-good 27 KVA EGS Hevi-Duty Delta-Wye "drive isolation" transformer [1] - so I COULD "re-derive" a "Local Neutral".

It will all "soon" be relocated to share quarters and supplementary ground rods - with the recycled NATO Diesel gen set. Acoustically quieter that way, the lot of them. "Wire exists", etc.

Not that I actually USE leg-to-Neutral for any 1XX loads at all. 2XX across, avoiding the "generated" leg, wherever single-phase is needed as primary for control transformers, task lighting, or direct power to single-phase juice pumps. ELSE separately sourced "out of the wall" off the common split-phase.


As with Best Current Practice, most powercos?

I just didn't want any high-leg/corner-grounded Delta "potential surprises" anywhere under my roof, no matter how likely they'd be protected inside an equipment enclosure.

Page Two:

Whilst some of my switchgear IS "600 Volt Class", I also decided I didn't want to HAVE to have that requirement to cover. So NO as in "NOT ANY" 4XX loads, either! 2XX is already more than lethal enough, thanks.

And then I have stick-and-rectumfry Dinosaur Current @ up to 275 V or so. "IN machine enclosure only", but still..

Highly lethal s**t. Keeps yah paranoid. In a good way!

2CW


[1] The differentiator is that these are nominal "1:1", so 220-240+ VAC Delta-in comes out as 220-240+ VAC Wye-out, and-not 208 Wye.

As that puts the hot leg-to-local-Neutral @ 133 VAC or better, not 120-plus, all the more reason to avoid USE of it, directly.

The other reason the mass of Iron and wire in that transformer is "there" is that when the transfer switch is selecting 3-P off the Phase-Perfect ("dirty" Sine Wave) instead of the RPC (very clean, smooth Sine-Wave), the massive lumped-inductance of that "Isolation" and-not "auto" transformer helps finish the clean-up of wotever fast-movers might have gotten past the TCI Sine-Guard.

Old Iron's Old Motors live longer that way, and a whole building full of router and 'puterish s**t, VFD'ed fridge-freezers, "inverter" microwaves, yadda, yadda and growing, every newer era of household appliances, even washers and dryers.. enjoys the enhanced peace and quiet on the wire.

SAF said:

..........
A corner grounded system (non symmetrical) is normally detrimental to most any VFD. Electro-mechanical controls typically will have no problem with this type system.
And there are NEC requirements that are different from standard wye grounded systems.

SAF Ω

Click to expand...

The subpanel for a 480V corner ground is normally different, since a 480 wye has 277 to neutral/grounded conductor, but the corner ground has 480V. That pops you to a new level of "voltage to ground" which the regular panel may not be rated for. Then there is the issue of the "grounded conductor", which cannot have a fuse etc.

if it is going to one machine, that may not be an issue, but otherwise, it may be.

The VFD issue is generally related to the transient protectors on the input. The rest of the VFD should easily be good with the 480V, given the way a VFD works it has to be insulated well above that for UL508.

JST said:

.. given the way a VFD works it has to be insulated well above that for UL508.

Click to expand...

Problem is ... that all manner of household - or shop office / lunchroom - APPLIANCES are incorporating switched PSU, inverters, VFD'ish stuff .. that is MEANT to work in a nominal 130 VAC world.

UNlike even "economy grade", but nonetheless "industrial" VFD, household goods are built to the least-cost to survive in THEIR market off the back of cheap-enough warranty cost or charge so much for field repair they are de facto "disposables".

"Can't protect them all", but

... the extra expense I've gone to for my 3-Phase, both downline and input feed filtering is not even a quarter of what ONE (out of two..) shattered "linear inverter" fridge-freezer - and frozen goods IN it - could set me back.

That spend is a lot less "overkill" than might first appear. Good meats vs rotten meats, and "BTDTGTTS" thing.

JST said:

The subpanel for a 480V corner ground is normally different, since a 480 wye has 277 to neutral/grounded conductor, but the corner ground has 480V. That pops you to a new level of "voltage to ground" which the regular panel may not be rated for.

Click to expand...

Too bad he can't reconfigure the secondary to be star connected. Or centertap grounded on one leg.

rpseguin said:

This is one of my concerns.
My CNC machines (440V CNC lathe, 230V VMC) have VFDs in them.
Empirically things have been working so far.
I just want to do a good, safe job.

Thanks!

Click to expand...

Sounds like your running the 480V machines ungrounded, it must be since they are still working fine.
You may have a ground wire connected to them from your RPC supply, but that will not get fault currents back to the transformer output coil. Grounding that delta coil on its corner, is likely to cause issues for your VFD's. Allen Bradley drives are the only ones I'm aware of that will let you disconnect the MOV's on the front end, and allow you to run it on a corner grounded system.

The best and safest bet would be to obtain a transformer with a wye output, 277/480V. Ground the center of the wye, and supply that ground reference to your machine. They are made, but pretty rare used.

This is a sample of what you would be looking for, but likely too big for your application.

Used Square D 75T112H 75KVA 240-480Y/277 Transformer | eBay
SAF Ω

SAF said:

The best and safest bet would be to obtain a transformer with a wye output, 277/480V. Ground the center of the wye, and supply that ground reference to your machine. They are made, but pretty rare used.

Click to expand...

That one you linked isn't "wrong".. but it is gawdawful expensive @ $2,200.... plus transport. EGS-Hevi-Duty and other "Drive Isolation" 1:1, 1:2, and 2:1 types are not as rare as all that, either. Just not as readily "reversible" as Delta-Delta usually are, so DO pick "the right one" at the outset.

By strapping my loads for 240, not 4XX, I got by with higher amps, larger wire, but a lot lower net spend AND stayed out of the 600 Volt class needs.

Lest we forget that STARTS at 300 Volt + and TOPS OUT at 600 Volt -, so 4XX goods fall under the 600 Volt class rules.

From my eBay purchase history & notes:

Hevi-Duty DT691H27S 27KVA 3PH Isolation Transformer
Pri 230 Delta
Sec 230Y/133
ITEM PRICE: US $499.99. (Bullseye Industrial Sales),

Invoice with freight included was $729.99


No "accident" that my Wye is 230/133 ++, not 208/120 - 416.

No accident I don't USE it for 120 V goods, either, given my utility mains is 246 VAC, not 230 VAC.

The motors don't see the fourth, "local Neutral" conductor, last "inch", at all, are meant to think it's Delta!

The motors don't see the fourth, "local Neutral" conductor, last "inch", at all, are meant to think it's Delta!

Click to expand...

It's still a centre-grounded supply and that's what's important. Loads don't otherwise care or know what the supply is and it doesn't really matter.

I know larger ABB & Siemens drives have screws to remove for non-grounded or corner grounded supplies and most other industrial-grade ones do too.

Running a larger system ungrounded without a ground fault monitor is not super smart.

SomeoneSomewhere said:

It's still a centre-grounded supply and that's what's important. Loads don't otherwise care or know what the supply is and it doesn't really matter.

I know larger ABB & Siemens drives have screws to remove for non-grounded or corner grounded supplies and most other industrial-grade ones do too.

Running a larger system ungrounded without a ground fault monitor is not super smart.

Click to expand...

Yah welll..

Running a lowly 120 VAC "Insinkerator" hot tea-water dispenser w/o a GFCI wasn't super smart, either.

Not when it failed and meant starting with the morning's coffee "wake up" was a tad firmer of an eye-opener that usual ... after it volunteered to knock yah on yer azz!

Some of us get "anal" about bonding and grounding 'coz that's where the calluses off the hard floor are?

Ground fault monitors for ungrounded systems are very different to GFCIs.

Isolated systems are single-fault-safe. That is, you can touch any one wire and not get a shock, earth any one point and not get a bang. But when the second occurs... well, the two of you are in series across whatever voltage.

A ground fault monitor would either disconnect or alarm when that first fault occurred, before anything even gets a spark.

Great for critical systems, ships, and that kind of thing because you don't have anything lose power on the first fault. Or great for safety if you tell it to cut power on that first fault, because no-one even has to get a shock.

But they're not used for systems with lots of small loads, because it's hard to tell where on the network the fault is. Can you imagine being a linesman and having to find the one faulty hairdrier fed from one transformer, before you can turn people's power back on?

SomeoneSomewhere said:

Can you imagine being a linesman and having to find the one faulty hairdrier fed from one transformer, before you can turn people's power back on?

Click to expand...

It's several times a year, every year, thirty years already. Most times, they send enough trucks to find it in three hours.

They walk the area in several teams of two. Prolly so one can call 911 if his mate is shot by irate homeowners.

Last few goes I've just teased them by asking if I can sell them a few spare KVA off my surplus NATO "tactically quiet" diesel gen set.

Takes out the entire community for lack of proper isolation, compartmentalization, fault alarm sanity.

The krews tell me they find the fault with their nose. Not "hair dryers". Snakes and rodents getting into the transformer enclosures for DIY barbecue piss-ups.

Ouch. Average around here is an outage of an hour every two years...

But no, this is 'whoops my fluoro ballast failed' type issues - the kind that will trip a GFCI but not a breaker.

SomeoneSomewhere said:

Ouch. Average around here is an outage of an hour every two years...

But no, this is 'whoops my fluoro ballast failed' type issues - the kind that will trip a GFCI but not a breaker.

Click to expand...

More than 35 years ago, now.. but "GFCI" duplex outlets must still have cost the very Earth.. as the PO of this home put a grand total of THREE in a 3 to 4 bedroom three-level split with about 1,200 SF of 2 1/2 car garage / workshop / annex.

Talk about insane wiring runs? Kitchen fridge, lighting, or appliances go dark, trek upstairs and try to remember which bathroom's GFCI is the one daisy-chained.. etc.

Several ten-pack boxes of new GFCI duplex outlets later, the only "daisy chain" NOW is in the quad outlet boxen I've added for "puter & printer/scanner convenience, master & slave side-by-side, same box - no confusion likely.

All others are stand-alones.

Even BETTER NEWS?

The "clean-up project" had me discover where the pure fool had "saved money" four branch circuits, by ignorant "dry" wire-nutting many feet of shiney-wood cable so that Proper-Copper appeared at box end and load center end.

HIDING that silly "house ignitor" shit in the middle potentially far the more dangerous lie than running shiney-wood end-to end so yah at least could see "at once" it had been f**ked-up.

Utility mains gurus, OTOH, know which shiney-wood alloy actually WORKS and how to terminate it properly for VERY long and trouble-free life.

About 35 years and counting on the Square-D QO 200 A service entrance panel, here, and nary a shiney-wood issue with the mains feed at all.

I did mention it pays to know yer s**t?