My dual-GEM installation

Working on a home project? Post it here.
Post Reply
wci68
Posts: 43
Joined: Thu Dec 17, 2015 4:31 pm

My dual-GEM installation

Post by wci68 » Mon Feb 08, 2016 6:02 pm

I recently was able to get a new pair of GEMs installed on my panel and have been working through some tuning here in this forum. Now that the troubleshooting and tuning are mostly complete I got all the wires run properly and closed everything up.

I went with two GEMs:
  1. I want to monitor quite a few more than 32 branch circuits, plus both legs of my service entrance.
  2. The voltage difference between both legs of my service can vary quite a bit. It may be as low as one or two volts (not a huge difference) or as high as eight volts. I can speculate why this may be but that tis an entirely different topic. As a result I wanted to monitor both legs independently.
I have a pair of Split-200 CTs, one on each leg of my service. One CT is channel 1 of one GEM, while the other CT is channel 1 of the other GEM. Similarly, I have the PT for each GEM connected to a circuit on the same service leg as that respective CT. IE: PT1 and CT1 monitor the same service leg and are connected to the same GEM1. Similarly PT2 and CT2 monitor the same service leg, opposite to PT1/CT1, and are connected to the same GEM2. I have Micro-50s on all of my branch circuits.

After a kitchen renovation project many years ago, I took the opportunity to completely redo my panel. I didn't change my service, but along with the large number of circuits I added for my kitchen I corrected many of the existing issues and cleaned up the mess left from previous owners. I spent a bit of time arranging the circuits to keep both legs roughly balanced based on expected estimated usage.

After installing CTs on all the circuits, getting the GEMs set up, and starting to monitor my panel I found that the legs were pretty well balanced, with the exception of one circuit. Thanks to my investment I was able to immediately identify this and make the appropriate change in my panel.

Both GEMs are located right next to my panel so I thought about how to keep the install neat. I ended up with a 3/4" flexible conduit from the panel to a 2.5" trough, and mounted both GEMs on top of that trough. I punched another 1/2" hole on the bottom of the trough and inserted a grommet to feed the wires from both power supplies and both PTs through. At the opposite end of the trough is a 1/2" flexible conduit used as a sleeve to pass a pair of Cat-6 cables through, which run back to a 19" rack elsewhere where all of my equipment resides:
IMG_0762.JPG
IMG_0762.JPG (350.44 KiB) Viewed 6175 times
Below the panel seen at the left edge of the picture is a 1900 (aka 4") box with a pair of duplex receptacles where both PTs and both power supplies are plugged in. In the picture you can see a temporary wire tie around the blue NMT holding these wires where they loop under and into the trough.

All of the CT wires run through the trough into the respective GEMs via the offset nipples on the left side. The serial, power, and PT wires run through the trough into the respective GEMs via the offset nipples on the right side. The extra length of wires are bundled and run down the trough in a single loop.
wci68
Posts: 43
Joined: Thu Dec 17, 2015 4:31 pm

Re: My dual-GEM installation

Post by wci68 » Mon Feb 08, 2016 6:26 pm

During troubleshooting noise came into question. It turned out not to be an issue, however I was able to gain access to an oscilloscope to see how "clean" my power really is.

The outputs of the power supplies were a nice clean flat DC output, though I didn't look too hard to find distortion or noise. The outputs of the PTs were also pretty clean:
Output waveform from PT on leg 1
Output waveform from PT on leg 1
PT1-Leg1.jpeg (66.46 KiB) Viewed 6169 times
Output waveform from PT on leg 2
Output waveform from PT on leg 2
PT2-Leg2.jpeg (66.51 KiB) Viewed 6169 times
There is some front-end noise from the scope itself, but otherwise the output was pretty clean and consistent over time with these captures. This is actually better than expected as I have seen quite a bit worse elsewhere.
Post Reply