With two monitors installed in the exact same way (CTs oriented in the same way, installed on the same breaker) and at the same time, on the same firmware, detection should be pretty much the same across both. Nonlinearities might enter from slight component tolerance differences, but these would be slight.
Was so glad to see you respond @RyanAtSense
I would be interested if this is a subject you can further elaborate on. Especially about the above quote and just how far off on a tangent one box could go from the other and the details about it.
I don’t know if if you have had or know of actual experience with this or it’s speculation.
There aren’t many of us that seem interested in this subject but I think there are at least a few that are very interested.
Call it “educated speculation.” We’ve never tested exactly what you’re suggesting. We have many homes with multiple monitors, but they were installed at different dates (like @samheidie’s case). We also have a testing setup where a whole lot of monitors are monitoring a single circuit and those tend to find the major appliances on that circuit at around the same time.
FYI: I had a couple of monitors on the same panel for a while but what I noticed was the different voltage calibrations (which Sense corrected at my request) on an early vs late model. I have a couple of Flukes so could also see things were off. Precise placement of the CTs aside (hard), it would seem that monitor component variation would be enough to throw expectations of perfection. These are 200Amp CTs with a +/- of something other than zero. As far as the realworld “informative” aspects of having multiple monitors: having one on, say, a water heater (as I do) or other dedicated circuit is a good way of keeping the monitors “honest”. You can also unambiguously set alerts on a dedicated-circuit monitor. Having read the “Oops, I left the kiln on to burn down the house” blog post recently, I can see that there might be crucial setups where you want some redundancy.
Frustrations & Questions
My prediction differs from a couple responses to this thread. @RyanAtSense and @kevin1 think they will be close and Ryan actually has access to a setup that is somewhat similar so I may be way off. While I do believe that detections will be similar after a long time, it’s the first detections or order of those detections I expect to differ.
I’m not quite understanding precise CT placement. I have a foot and a half separation in the CT’s for this forest unit now and will have exactly the same on a second unit except I will swap. Meaning I have a CT at the top of the panel on the left lead and bottom of panel for right lead. I’ll take the second unit and put on top for right main and bottom for left. This should ensure enough distance to prevent any crossover or bleed.
I think you were monitoring separate circuits unless I misunderstood. I plan to monitor the entire panel on both.
Do you have a prediction?
A high quality current probe like the Fluke i310S is rated at +/- 1%. You might get lucky with the Sense CTs and end up with measurements within 1% of each other but my prediction, if you want to really quantify things, is, say, a 1% difference with the caveat that compared to the real current both the monitors may be off by a % or 2. My “dual” setup (both on the entire panel) was more like 2% difference because the voltages weren’t calibrated until Sense stepped in and re-calibrated the older monitor. Now, after factory resetting both monitors, I have one on the entire panel (#1) and one on a hot water circuit, #2 (reactivated at the same time). The “inferred” water heater usage on #1 is very close to the water-heater-dedicated monitor #2. Right now for example, #1 shows a stable 4,149W and #2 hovers close to that. What’s interesting though is that #2 varies by about +20/-10W around 4,149W. What that translates to is total water usage for the water heater last week according to the whole-panel meter #1 = 65.0kWh; according to #2 = 66.7kWh. Around 2% difference.
That is very interesting and 2% isn’t that bad in my opinion. It would be nice to get closer but probably not possible.
I did find it curious though at the low wattage for your water heater. I also have 4500 watt elements and mine runs 4600-4700 watts. My voltage is pretty steady at 123.4 on each leg.
So do you think they ran pretty close at detection speeds and the devices detected overall?
Despite recalibration the difference in voltage readings for Sense #1 vs #2 is around 1V. The gen1 (?) Sense (#1) seems to read higher than the actual voltage as measured with a Fluke 117 RMS; the gen2 Sense #2 seems closer to the Fluke readings. Meaning: if you have 2 gen2 units it looks to me like they calibrate better and you’ll get closer results. @RyanAtSense could perhaps speak some truth to this. I’m just speculating.
As far as tank element wattage reading goes, my tank seems to be running at close to the rating: it’s a “Rheem Performance Platinum 40 Gal 5500/5500 Watt Elements”. That’s the 240V rating. My tank is wired for 208V ==> 4130W. You’re probably running 240V to yours? Rheem has good white papers on this stuff – https://www.rheem.com/whtechbulletins/
I didn’t pay very close attention after the reset of the monitors as to how long they took to recognize the water heater. I’ll have to look back over the data and try and parse it. From memory the detection speed was quite different and in the opposite way to what you might expect: Sense#1 (whole panel) was faster than Sense#2 (dedicated to tank)! I do remember that the first Device detection on Sense#1 was my floor heating element (also 208V)
Three phase explains it.
Hopefully in the next few weeks I can get another monitor. I’ll have to ask questions about generation of model. They way it’s made to sound all monitors are created equal and can be updated to the same firmware. It’s not something I felt could be true with the differences the early adopters have had compared to those of us that came later. It seems their monitors are stuck and won’t catch up. I’m not sure if that’s from hardware alone or the way sense hangs on to last data.