Three Phase site - multiple electrical issues.

[gar]

130928-1608 EDT

Sparty D:

Do the RS485 paths have isolation?

So you have a homemade 12 V DC power supply. What is the maximum load current on the 12 V DC output? What is the regulator chip? Are you using full wave rectification? I assume a capacitor input filter. What is the size of the filter capacitor? When the power supply is at full load and input AC voltage is nominal, then what is the average DC voltage across the filter capacitor, and the ripple voltage across the same capacitor? What is nominal AC input voltage at the power supply? What is the outside surface temperature on the regulator where it attaches to its heatsink at full DC load, and nominal AC input voltage?

At what input AC voltage does the minimum filtered DC voltage at full load output current drop below 16 V (this allows for a minimum voltage across the regulator of 4 V)? At full DC load what is the capacitor voltage when AC input voltage is at its maximum likely value?

If your load current is 1 A and the regulator drop is 4 V, then regulator power dissipation is 4 W. If at maximum AC input voltage the capacitor rises to 20 V, then regulator dissipation is 8 W.

You referenced 97 V as a minimum dip, and possibly 1 second long. If your filter capacitor is not large enough, then you could loose 12 V regulation. You may want a rather large filter capacitor, or a backup battery integrated into the power supply.

Are you getting data errors from the RFID door units, or do they just go into "never-never land"?

What is a PTAC?

.

 

[Sparty D]

In addition to measuring voltage to remote earth (and a valid "remote earth" may be hard to find across the street from a substation!), the EC should look for high current in any EGC or GES conductors. These will be indications of a high gradient between local surface earth potential across the building site.
It would take as little as 1A to cause a 10 volt differential across a 10 ohm ground electrode resistance, but a 10A GEC current could cause only a fraction of a volt difference within the GES wiring but correspond to a difference in local earth potential between the two electrodes of 100V, a real life safety hazard as well as potentially damaging connected equipment.

Since the first year of the site had a ground system that was not to code the second EC installed the three new rods 6-8 feet apart outside in the yard and ran a ground wire overhead (due to slab of building). I believe the original two ground rods are also attached. A Fall of Potential test was done (after a rain BTW) and showed both to be within NEC standards. 100' had 3 ohms on original and 2.5 on the new. Of course, the test is only a snap shot for that moment. The problems can disappear for a while and then seem to come in packs and then go away.

 

[Sparty D]

130928-1608 EDT

Sparty D:

Do the RS485 paths have isolation?

So you have a homemade 12 V DC power supply. What is the maximum load current on the 12 V DC output? What is the regulator chip? Are you using full wave rectification? I assume a capacitor input filter. What is the size of the filter capacitor? When the power supply is at full load and input AC voltage is nominal, then what is the average DC voltage across the filter capacitor, and the ripple voltage across the same capacitor? What is nominal AC input voltage at the power supply? What is the outside surface temperature on the regulator where it attaches to its heatsink at full DC load, and nominal AC input voltage?

At what input AC voltage does the minimum filtered DC voltage at full load output current drop below 16 V (this allows for a minimum voltage across the regulator of 4 V)? At full DC load what is the capacitor voltage when AC input voltage is at its maximum likely value?

If your load current is 1 A and the regulator drop is 4 V, then regulator power dissipation is 4 W. If at maximum AC input voltage the capacitor rises to 20 V, then regulator dissipation is 8 W.

You referenced 97 V as a minimum dip, and possibly 1 second long. If your filter capacitor is not large enough, then you could loose 12 V regulation. You may want a rather large filter capacitor, or a backup battery integrated into the power supply.

Are you getting data errors from the RFID door units, or do they just go into "never-never land"?

What is a PTAC?

.

A PTAC is a HVAC unit in the residents room. Heater/AC unit.

 

[Sparty D]

A PTAC is a HVAC unit in the residents room. Heater/AC unit.

The power supplies are quality units and off the shelf. There are fifteen of them and I have used them for years with no similar (or even close) problem. I have measured them with a logging meter and over a three day period are rock solid. Due to the high harmonics we installed a Tripp Lite Power Conditioner and still had problems. If you assume (and I know what that means) that the AC was filtered by the power conditioner and the DC power supplies was stable then the focus would seem to shift to the ground connection used to dissipate the static shock someone could give the unit when they walk up and enter a code or....RF. Both could be split second anomalies that appear and go away for days or months. Again, thanks for the brain storming.

 

[gar]

190929-1920 EDT

Sparty D:

You are making qualitative statements about your DC supplies when you need some quantitative data. I doubt the power supplies are your problem, but you need more information on their characteristics.

What is the load current on the output of the 12 V DC supply? Is it constant? If not, then what is the minimum and maximum load current?

If you put a constant load on the power supply equal to the maximum current your load requires, then what is the minimum AC input voltage where you just loose regulation?

If you apply maximum load on the supply, apply the nominal, sort of steady, minimum AC input you might expect, and then instantaneously drop the input AC to zero, then how long does the DC output remain regulated after the loss of AC input?

.

 

[hurk27]

The power supplies are quality units and off the shelf. There are fifteen of them and I have used them for years with no similar (or even close) problem. I have measured them with a logging meter and over a three day period are rock solid. Due to the high harmonics we installed a Tripp Lite Power Conditioner and still had problems. If you assume (and I know what that means) that the AC was filtered by the power conditioner and the DC power supplies was stable then the focus would seem to shift to the ground connection used to dissipate the static shock someone could give the unit when they walk up and enter a code or....RF. Both could be split second anomalies that appear and go away for days or months. Again, thanks for the brain storming.

Just a curious thought, in experience when I had a lot of data failures from installations done by others, several times I found shielded cable was being used in the wrong manner that was providing parallel grounding paths with the buildings grounding system by bonding both ends of the shield, have you looked to see if this was done like this?, 485 buss networks kind of have this problem built in because of the common ground reference in the cabling between points in the network, in many cases this can introduce allot of noise into the data streams that can cause allot of hiccups as well as system crashes, it can also be a problem when you have systems fed from more then one SDS transformers where the X0 is derived to ground at each transformer causing circulating X0 (neutral) currents in the grounding paths that can be imposed upon the shields or the ground reference in the 485 buss, not sure if this is the problem but like Gar pointed out you could be fighting many problems that may have to be looked at individually and eliminated one at a time, in the case of the 485 network you might want to look at isolation if you have more then one transformers that the systems are tied to or even if not you could have accidental neutral to ground connections causing a dirty ground path that is paralleled with the 485 buss, Gar is one of the sharpest on this subject and has developed many devices to offer isolation for 485 systems.

One of the hardest things is finding a good EC that understands how to find and fix these possible neutral to ground connections. but this is just one part of what I'm thinking that is happening as I also think you are experiencing some other power problems that could be coming from the utility.

Compound problems can make you pull your hair out when trying to trouble shoot them because in a way you have to isolate them and attack them one at a time, a good recording O-scope to look for the noise can be a good tool to have in cases like this so you can identify what kind of noise your dealing with.

 

[Sparty D]

Just a curious thought, in experience when I had a lot of data failures from installations done by others, several times I found shielded cable was being used in the wrong manner that was providing parallel grounding paths with the buildings grounding system by bonding both ends of the shield, have you looked to see if this was done like this?, 485 buss networks kind of have this problem built in because of the common ground reference in the cabling between points in the network, in many cases this can introduce allot of noise into the data streams that can cause allot of hiccups as well as system crashes, it can also be a problem when you have systems fed from more then one SDS transformers where the X0 is derived to ground at each transformer causing circulating X0 (neutral) currents in the grounding paths that can be imposed upon the shields or the ground reference in the 485 buss, not sure if this is the problem but like Gar pointed out you could be fighting many problems that may have to be looked at individually and eliminated one at a time, in the case of the 485 network you might want to look at isolation if you have more then one transformers that the systems are tied to or even if not you could have accidental neutral to ground connections causing a dirty ground path that is paralleled with the 485 buss, Gar is one of the sharpest on this subject and has developed many devices to offer isolation for 485 systems.

One of the hardest things is finding a good EC that understands how to find and fix these possible neutral to ground connections. but this is just one part of what I'm thinking that is happening as I also think you are experiencing some other power problems that could be coming from the utility.

Compound problems can make you pull your hair out when trying to trouble shoot them because in a way you have to isolate them and attack them one at a time, a good recording O-scope to look for the noise can be a good tool to have in cases like this so you can identify what kind of noise your dealing with.

You make a good case and Gar's advice is much appreciated. I also agree that High Voltage guy will need to know his stuff. I think I have one, but the low voltage guy fixing the high voltage stuff the challenge. Also, one of my problems is the end user uses a paper clip to reset the units and I have not been on site to see what is truly at fault. The 485 in the door units will isolate itself when a transient is seen, but some of the units, by description, are more than off line...their memory is scrambled. So I think both is happening. I have replaced all the power supplies. Normally a switching power supply is used (Altronix SMP3) and after a year I put new ones in (Altronix LPS3). The door units draw less than 1amp and these put out 2.5amps so current draw should be more than full filled. Voltage surge or sage was my first guess months ago. I have logged for days the output of the DC and it does not fluctuate past Mfg spec's. But then again, none of the units I have logged locked up when I was on that unit.

I have said to the GC and the owner that I think it takes a perfect storm of issues to come together at the same moment. It is there...finding it is the challenge.

I am shifting my attention to the ground and neutral, gather details on the power supplies and revisit the MFG with the 485 wiring.

 

[gar]

130929-2350 EDT

RS232, 422, and 485 are not going to be reliable without dielectric isolation in the communication paths when you can not guarantee that all logic circuits are referenced to one reference point and there are no ground currents in what can be described as the EGC connections between the end points of the communication path.

See my discussion at http://beta-a2.com/noise_grnd.html, http://beta-a2.com/i232.html, http://beta-a2.com/I232B2_photo_files/I232-BLKDIAG.html, and I have a printed discussion on ground and noise with more detail that is not on the internet.

Also I should point out that equipment that corrupts data from transient conditions is badly designed. There is much of this bad design by people that have not thought about the problem. Equipment that I have designed going back to 1975 that had CMOS battery backed up data storage did not have this problem unless there was hardware failure.

.

 

[Sparty D]

Branch Circuit Voltage

Branch Circuit Voltage

I read and article Mike Holts wrote on bonding and grounding for 3 phase systems. I went to our outlet that is a dedicated run to the subpanel and measured the voltage between ground and neutral. .2 to .4vac. Based on the article I was expecting to see 4.5vac. 3vac for the panelboard and 1.5vac for the branch. Can you let me know if I am missing something?

Also the owner let me know that nearly all the electronic thermostats in the complex for the residents HVAC unit have been replaced due to failure and some of the replaced ones have gone out again.

 

[iwire]

Excuses me but a 1600a service and Romex I find it hard to believe are used in the same paragraph.

1000s of examples of that all over New England

 

[SceneryDriver]

1000s of examples of that all over New England

The Romex into subpanels got me thinking. I read another thread here recently ( http://forums.mikeholt.com/showthread.php?t=143804 ) about the prohibition of using SE cable to supply ranges and dryers from subpanels, as the bare neutral could cause an inadvertent ground loop through contact to grounded parts in the subpanel. Could something similar be going on here? Did the original, dodgy EC mix up / cross connect the neutrals and grounds in the subpanels? That could explain your ground-isn't-ground issues.

Aluminum or copper wire on the feeders to the subpanels? Was anti-ox compound used on the terminations? Perhaps a corrosion-induced bad neutral connection that comes and goes?



SceneryDriver

 

[Sparty D]

The Romex into subpanels got me thinking. I read another thread here recently ( http://forums.mikeholt.com/showthread.php?t=143804 ) about the prohibition of using SE cable to supply ranges and dryers from subpanels, as the bare neutral could cause an inadvertent ground loop through contact to grounded parts in the subpanel. Could something similar be going on here? Did the original, dodgy EC mix up / cross connect the neutrals and grounds in the subpanels? That could explain your ground-isn't-ground issues.

Aluminum or copper wire on the feeders to the subpanels? Was anti-ox compound used on the terminations? Perhaps a corrosion-induced bad neutral connection that comes and goes?



SceneryDriver

Several EC's including the one I want to inspect the site, believes there is a strong possibility of a mix up. A plain reversal or a inadvertent ground. The first EC did several short cuts we know of. There were at least six breakers labeled and when the cover was removed.....no wires on them. So the site was probably phase imbalanced for at least a year. Add the harmonics and it is not a hospitable environment for us low voltage folks. Thanks.

 

[Sparty D]

UPDATE

UPDATE

Several EC's including the one I want to inspect the site, believes there is a strong possibility of a mix up. A plain reversal or a inadvertent ground. The first EC did several short cuts we know of. There were at least six breakers labeled and when the cover was removed.....no wires on them. So the site was probably phase imbalanced for at least a year. Add the harmonics and it is not a hospitable environment for us low voltage folks. Thanks.

It has been a while and I thought I would share some info. I was able to bring in a 30 year EC vet and look over the breaker boxes. We only pulled 3 or 4 covers but quickly found the bankrupt EC didn't pull a neutral for every hot. One panel was missing several and leaves one to guess if AA, BB, CC was done the field. The low voltage system is on the backup gen panel along with the emergency lights. Those emergency lights come on and off several times a week still. I think we found the problem. Shared neutrals and probably a neutral used as ground.

 

[GoldDigger]

Except for project specific rules, there is nothing intrinsically wrong with a shared neutral in an MWBC configuration. But it must not double up on a single phase and it requires handle ties and grouping of the phase wires used and their neutral.


Tapatalk...

 

[ActionDave]

I agree with Gold, if done correctly there is nothing wrong with sharing a neutral.

Neutral and egc cross pollination somewhere in the system can cause a lot of problems.

 

[kwired]

I have to agree with others so far, you need to do more digging before determining your neutrals have any problems. As far as handle ties and grouping circuit conductors together, it would depend on how old the installation is before determining it was incorrect, as some of those rules were more recent changes.

add: I just noticed this is a thread that has older posts, my reply was based on thinking it was a new thread, so forgive me if I missed something mentioned earlier, did not read all the posts, and did not read any except the very recent ones before my post.

 

[Sparty D]

I have to agree with others so far, you need to do more digging before determining your neutrals have any problems. As far as handle ties and grouping circuit conductors together, it would depend on how old the installation is before determining it was incorrect, as some of those rules were more recent changes.

add: I just noticed this is a thread that has older posts, my reply was based on thinking it was a new thread, so forgive me if I missed something mentioned earlier, did not read all the posts, and did not read any except the very recent ones before my post.

The engineer who did the plans was contacted and we opened a huge can of worms. The plans state that all branch circuits must have their own neutral. It also states no wire larger than #6 can be spliced. Two things that are clearly visible by removing the sub panel cover. It should have never been approved by the engineer or the city who stamped the plans IMO. The EC that was hired after the bankrupt EC went under "verified neutrals" and one would have thought the fact it was not to plan would have been mentioned two years ago.

The next step is to go through and see/find the cross of neutral and ground and try to make it to plan by adding neutrals.

 

[GoldDigger]

The engineer who did the plans was contacted and we opened a huge can of worms. The plans state that all branch circuits must have their own neutral. It also states no wire larger than #6 can be spliced. Two things that are clearly visible by removing the sub panel cover. It should have never been approved by the engineer or the city who stamped the plans IMO. The EC that was hired after the bankrupt EC went under "verified neutrals" and one would have thought the fact it was not to plan would have been mentioned two years ago.

The next step is to go through and see/find the cross of neutral and ground and try to make it to plan by adding neutrals.

In most cases, the inspector is tasked with verifying the code compliance of the installation, including such things as location of required lighting and receptacles. But not with checking that the installation matches the plans, particularly in terms of plan requirements that go beyond Code requirements.
Details vary from one AHJ to another.

 

[petersonra]

This just does not sound like a ground or neutral switched issue. Normally the only thing that happens then if GFCIs trip.

But it is typically pretty consistent.

 

[petersonra]

You did mention you have rs232 and rs485 circuits. they are usually not isolated. you might want to install isolaters in those lines. not a solution, but maybe a bit of a bandaid.

I can't imagine that a suggestion that they shut off the power as part of diagnosis would go over real well.

Linear power supplies usually have a high degree of isolation and ride through so whatever is downstream of them should be fairly clean. did you ground one leg of the DC for some reason? if you did, and there is some weird voltage coming through the ground wires, it could get past the DC power supplies since the isolation would be defeated.

I would be checking the ground wires coming into all of the panel boards and seeing if there is current flow in any of them. record what you find, but don't worry all that much about it for the time being.

get a long wire, maybe a couple hundred feet and an analog meter (do not use a high impedance digital meter). measure the voltage between an EGC at one end of the building to an EGC at the other end of the building.