Phantom usage being reported on unused channels

[wci68]

I finally was able to complete the installation of a pair of GEMs along with all of the CTs in my panel recently.

I do have some unused channels in each GEM and noticed that they will report minimal usage occasionally. Both GEMs exhibit the same behavior, and all of the unused channels seem to be consistent over time. Watching the live data page in the GEM setup utility I would notice that one or more of the unused channels will show the minimal 0.02A for one refresh cycle (8 seconds).

After about 12 hours, all of the unused channels on both GEMs show a cumulative watt-hour value of between 2.5 and 3.

This seems similar to the topic discussed in http://www.brultech.com/community/viewt ... =29&t=1210 but in my case this is happening on channels with nothing connected.

I have not examined or tuned any of the used channels yet so I don't know if they are similarly affected.

The PTs for the two GEMs are connected to opposite legs of my service. I'll be setting up my Fluke 189 to record for a while to get a better view, but I do already know my voltages can vary quite a bit.

It would seem noise may be a factor, but it does impact both GEMs similarly. The GEMs are mounted side by side (about four inches between them), about two feet from my electric panel. There are no sources of noise (RFI or EMI) close by.

Would anyone have any suggestions on how to diagnose this?

[ben]

It sounds like a noise issue. Have you tried jumpering ports 1,2 on the unused channels? That should add some resistance.

[wci68]

I plan to try that but have not yet done so.

Would the jumper be across terminals 1,2 or 2,3?

[ben]

I plan to try that but have not yet done so.

Would the jumper be across terminals 1,2 or 2,3?

My mistake, jumper 1 and 4 for empty channels. 1 and 2 is for when you have CTs connected to 2 and 3.

[wci68]

Unfortunately placing a jumper as noted (1,4) didn't seem to make any difference.

I have the channels set to a Type-B (Micro-50) CT, but also tried a Type-A (Split-200) with no difference.

I wonder if the source of the noise may be via the PTs but that would be pure guesswork.

If only the unused channels were affected it would be nice to be able to select a channel as 'unused' such that the GEM would ignore that channel completely. However I have not determined whether the other channels that actually have CTs connected are also similarly impacted. Hence the reason I would prefer to generically nullify or eliminate the noise altogether.

Since both GEMs are exhibiting the same behavior I do believe the issue is with my environment. . .the (not-so) "fun" part is identifying that actual issue.

PS: To confirm, both have COM version 4.0 and ENG version 1.49. Both are serial-only.

[Teken]

Unfortunately placing a jumper as noted (1,4) didn't seem to make any difference.

I have the channels set to a Type-B (Micro-50) CT, but also tried a Type-A (Split-200) with no difference.

I wonder if the source of the noise may be via the PTs but that would be pure guesswork.

If only the unused channels were affected it would be nice to be able to select a channel as 'unused' such that the GEM would ignore that channel completely. However I have not determined whether the other channels that actually have CTs connected are also similarly impacted. Hence the reason I would prefer to generically nullify or eliminate the noise altogether.

Since both GEMs are exhibiting the same behavior I do believe the issue is with my environment. . .the (not-so) "fun" part is identifying that actual issue.

PS: To confirm, both have COM version 4.0 and ENG version 1.49. Both are serial-only.

If adding the jumpers do not help to reduce the noise the next step is to complete the other fine tuning procedures outlined in the other thread.

[wci68]

That would be my thought. However both GEMs being affected similarly, with all unused channels being affected similarly (consistently) would tend to indicate something in common across them all.

The PTs would seem to be the most likely source. If only some channels were affected (IE: just those from circuits on one side of the platform, or those on just one leg of the service) it might indicate otherwise, but all unused channels are indeed impacted equally on both GEMs.

Otherwise, I wouldn't see how to "tune" a channel that has nothing connected. These aren't channels with simply no load - they actually have no CTs connected at all. Suggestions on how to "tune" these would be appreciated. It seems rather odd having to "tune" a channel that actually has nothing connected.

[Teken]

That would be my thought. However both GEMs being affected similarly, with all unused channels being affected similarly (consistently) would tend to indicate something in common across them all.

The PTs would seem to be the most likely source. If only some channels were affected (IE: just those from circuits on one side of the platform, or those on just one leg of the service) it might indicate otherwise, but all unused channels are indeed impacted equally on both GEMs.

Otherwise, I wouldn't see how to "tune" a channel that has nothing connected. These aren't channels with simply no load - they actually have no CTs connected at all. Suggestions on how to "tune" these would be appreciated. It seems rather odd having to "tune" a channel that actually has nothing connected.

My reply was more in a general sense and not specific to the channels. Leaving the jumpers in place I would adjust the PT voltage and see what the results are. If doing both doesn't change the results I would try the ferrite choke.

Another solution I used not documented in the other reply was the use of another power supply feeding the GEM. I also used a isolating transformer but that was used more to decouple various hardware devices from potential electrical surge events.

[wci68]

I'm not sure how much impact the absolute voltage would have, but I do plan to check and calibrate this if necessary.

The challenge with ferrite beads is identifying the frequency range of the noise to inhibit. This is really where having an oscilloscope would be very useful, both to confirm if there is noise as well as to identify its characteristics to know how to filter it. Sadly this is one piece of test equipment that I do not own.

Then there is the fact that the PTs are AC output. Ferrite beads would only help for higher frequencies. If the noise is a harmonic of the 60Hz supply these would not be of much help.

It does appear that all channels are affected. I have two channels for two circuits with no load at all and they are behaving similarly, which is not surprising. Having a phantom usage of roughly 6 watt-hour per day per channel adds up quickly: 384 watt-hours per day (2 GEMs), or roughly 11.5KWh per month.

[ben]

Sounds like it could just be an offset issue. You can try adjusting the offset to eliminate these values.

- In the GEM Network Utility, under Advanced, start by reading the offset.
- Click the Read Amps button, it'll show the amp rating for each channel.
- Change the offset value, adjust it by 3, click Save Offset, wait a couple seconds and click Read Amps.
- If its 0, you're good, if it's gotten worse, adjust in the opposite direction (add/subtract 3 from the original value).

Make sure you wait a few seconds in between each adjustment. Once calibrated to 0, reboot the GEM to make sure the value sticks.