Interference between channels and spurious readings

Next on my list was going to be the ADCs. All of the components that you are looking at should be the same for all inputs, but you are showing different noise on the inputs associated with each ADC.

Unless you have some MCP3208 ADCs hanging around, I would need to send you some. If I’m going to do that, I may as well just send a new board. But if you don’t mind going the distance, I think Ill send both if you will change the ADCs on this one just to try to resolve this.

Yes I agree and are willing to change the ADC’s on my board. After all we want to find out what caused my board to not behave as intended. I’d be glad if you could send some over and good idea to include a replacement board just in case it’s still not playing along nicely.

My last order was #128573 is that sufficient to find my address or is there a way to send a PM with the address?

I checked I could source the ADC’s locally. Shall I do that as next step or do you want to send them over with a new board?

Only if you want to potentially resolve this sooner. I’ve got a few thousand of them so I will send along with a replacement PCB. Standard economy shipping was about two weeks with your prior orders. It should go out tomorrow.

Perfect will do it like that then :+1:
Thank you very much and I will replace the ADC on my board as soon as the shipment arrives. I will be in touch.

1 Like

The replacement board and ADCs arrived. I have done a minimal config on the new board and powered it with the USB power brick. The results are the same as with my original board. There are still many phantom readings. The original board with replaced ADCs does perform better than before but is also not quite and reads power where there is none.


This is the replacement board, it has been powered for over 12h


This is the original board with replaced ADCs, also powered for over 12h

Somehow I think that we have reached the limits of this hardware and that all the units most probably show this behaviour of phantom readings if no current at all is flowing. I assume that in most cases this is not an issue because the measured circuits always carry some current and the issue therefore is not noticeable. Just for my case it isn‘t because I mostly measure or supervise circuits that are switched off for longer periods of times and then the observed effect is most noticeable.
I think it might be solvable in firmware by lowering the resolution and for example instead of going down to 1W resolution use 10W instead, that would be still sufficient accuracy for my case but most probably get rid of the phantom readings. What do you think? Possible?

How shall we proceed?

I did some tests on the replacement board before shipping. On the bench, with AC connected and all channels configured, the Status display indicated solid zero for all channels. I plugged a 100W load into each channel in turn and measured 100W on the active channel and zero on all others.

The IoTaWatt is measuring current at each channel. I did these tests with 120V reference voltage, so at 240V a low current reading would be doubled and consequently the power doubled. But the original complaint was that the Status display showed non-zero when inputs were inactive, and changed in sympathy with other active channels. In fact, they were showing 3-4 Watts. Is that still the case?

Some possible solutions to your issue may be to upload the data to something like influxDB or Emoncms and filter the low readings to zero there, or even simply to download the data in CSV form into a spreadsheet to filter low readings.

Yes, it is still the same and I am puzzeld because I was sure you would test the unit before sendig it out. But the replacement unit does behave very similar to my original board. All input on ADC 1 show phantom readings arount 3 Watt. ADC 2 is much better and shows around 700mW with 240V reference. With a 120V reference would probably read near zero. But ADC 1 channels readings should also be visible with 120V reference? Below you find a screenshot.

I want to use the unit with Home Assistant, so in theory I could define some template sensors and filter the low readings. This would work for all non cumulative readings like power or current but energy would not work that way, that would have to be done in IotaWatt as far as I know. That is why I was asking about the firmware options.

Would you be willing to try to optimise this in the firmware?

Just to try this as well. I configured a US 120V VT. Now the phantom power for ADC1 is about half, which is to be expected. For ADC 2 there is not much difference, the readings are very similar to the 240V readings. The screenshot shows the transition and I have also added the voltage for reference.

Configuration:

All I can say is that I tried all of those inputs on 120V before sending it out. They showed zero on the status display. I don’t remember the exact date, but it is probably in your datalog at the beginning and you can plot the various inputs to see what they were reading. Under 500mW would be zero on the Status display.

I wouldn’t know where to begin without risking unintended consequences. Making changes to the fundamental electronics, sampling or logging would not only be risky, but would take many months to roll out into a release. Respecting the apparent satisfaction of thousands of users worldwide, with working voltages from 100V to 480V, it could have repercussions that would be difficult to resolve.

The firmware is open source and available on Github. You are free to make changes to your unit.

Unfortunately I couldn’t find any data in the logs prior to last Friday. There is only a section in den message log stating that the unit was started once on May the 30th, but no data in the data logs. It´s ok it doesn´t matter anyway.

I would like to thank you for your help with my rather strange issues with the IoTaWatt, it is very much apprechiated and I wish you all the best for the future.

One last question, than I´m done and this thread can be closed: Do you want me to send you the replacement unit back?

It was interesting. Please keep the extra IoTaWatt.

Thank you very much.