SuperBad-SuperNight-EachNight

[ELA]

Posting this mainly for those interested in the details of diagnosing a severe Noise Generator.

Had my entire Insteon Network begin failing for several nights in a row.  After a power cycle in the morning everything was good again. Initially thought it was consistent with intermittent failure mode of the PLM. My PLM was from 2015 so seemed likely. 

After several hours of testing I came to understand it was a SuperNight CT-1250 switching power supply generating high levels of noise in the Insteon freq. range.  This supply powered several SuperNight LED strips that have been in service since 2017.   The power supply still powered its LED loads without fail outputting  ~12V @ 3-4amps.  However it was now also emitting a very large level of noise into the power lines.   The noise level was so bad (SuperBad) that it completely killed all Insteon comms involving the PLM. 

I took the time to investigate and document in some detail since I had never encountered such a severe noise issue before.   The vast majority of issues I had encountered were "signal sucker" issues that affected limited parts of the network but never the entire network.  I also encountered a few minor noise generators, which only affected one, or a few devices in the network.  Never would I have imagined a noise level this severe.

    Below are a couple of O’scope captures showing the old defective supply as compared to a new replacement supply (ALITOV ALT-1206) that was to be installed. Noise levels were captured using a CP000 coupler to isolate the 60Hz components.  The levels are “worst case” as measured on a test bench directly at the power supplies power plug.

 

It can be seen that the new supplies noise level is much, much better (old supply was failing).    I still had a slight concern that the new supplies noise level is not exactly stellar. I imagine the old supplies noise level was probably similar when new so it should be ok. If needed I would add a filter later.

After installing the new supply I tested the noise levels at an outlet near the PLM.  They were very much attenuated ( lost in background noise level). The outlet is located over about 25ft. away from the supply.  Comms testing to devices closest to the new supply show 3 hops, 3 remaining consistently.

After removing the old supply I opened it up and found that the main 100uf (400V) electrolytic cap. was bulged and no longer working.

[IndyMike]

@ELA, very nice job of tracking down the offending noisemaker. 

This is an EXCELLENT example of how an existing device can fail and bring an Insteon installation to it's knees (particularly if it's close to the PLM).  Unfortunately, it can be extremely difficult to isolate the failing device.  Many times the device has to fail completely before it can be located.

Would you mind going into a little detail on how you zeroed in on the CT-1250?

Nice plots and FFT by the way.  I had to zoom in to see the order of magnitude difference between the old and new supply.

IM

Edited February 6 by IndyMike

[Brian H]

I saw a troubleshooting tutorial for X10.

I cheap phone charger. Outputted a signal on the power lines. At the exact frequency of 120Khz the X10 signals frequency.

[ELA]

Thanks for your compliment IndyMike,

    It took a couple of days to understand that the issue was mostly only happening in the evening.  I was then able to concentrate on only those items that were on only at night or early morning.  The supply is switched on and off by an Insteon Switchlinc each evening and thus was isolated from the network most of the time.

     I am very fortunate to have good Insteon troubleshooting tools such as an Oscilloscope, CP000, 9Khz high pass filter and the ELAMontor.    Having such tools makes things a little easier to find.  I was able to monitor the line, observe the noise, and then turn off items one at a time while monitoring the line.  I was initially monitoring from a location roughly 60 ft away (electrically) from the offender and could still see 1 vp-p at that location.  This high level of noise was able to penetrate deep into the network.   The noise being such a large amplitude also made it easier to montitor.  I recall trying to track down minor noise makers, with levels close to the background noise floor, and that makes things more difficult.

Hi Brian ... thanks for your input.

In my case the first harmonic is likely the main offender and it is at ~120kHz.  The two captures in the original post were reduced in size to post them side by side.  Those were using the CP000 coupler which attenuates the fundamental.  The highest peak in those is the 120kHz harmonic.   Below is a full size capture for easier viewing taken using the 9Khz high pass filter.  That filter allows one to see the fundamental and upper harmonics without attenuation.

 

[IndyMike]

Hey @ELA,

You have some rather powerful tools that allow you to track down noise and signal absorbers.  However, it was your observational powers and process of elimination that lead you to identifying the culprit.  These tools are available to anyone. 

Insteon users just have to recognize that it's not always a "new" device or problem that is causing a communication issue.  Devices do go fail, and dual band does not cure everything. 

Thanks again,

IM   

[larryllix]

So this "condenses" down to a component that was afraid of the dark?
 

 

[ELA]

Yes Larryllix a real fraidy cap

I thought you might find this of some interest given I recall that you had installed a lot of LED strips like the ones this supply powered.

[larryllix]

6 hours ago, ELA said:

Yes Larryllix a real fraidy cap

I thought you might find this of some interest given I recall that you had installed a lot of LED strips like the ones this supply powered.

 

IIRC the automotive industry still calls them condensors.

Yes. I only use five RGBW strip controllers now after moving. I have had a few power supplies go dead on me, but never create noise, that I know of.

I still run them, along with 17 x RGBCW bulbs,  3 x RGBW bulbs, and 2 x WW/CW floor lamps via my MagicHome software. They all work well for the last 5-6? years. They keep bringing out new styles of bulbs and I have to hack the differences to tweak the code. The special effects are handy for notifications of people outside the door, and midnight (go to bed) reminders.  

Edited February 9 by larryllix