ATO issue

epicfatigue · May 24, 2021

Hey everyone,

Recently setup reefpi and i having some issues.

I have setup an ATO using a Non contact water sensor, and I have it connected to a level shifter.

It works fine majority of the time but sometimes I am noticing Reefpi isn't detecting the 3V3 from the level shifter.

Anyone had this issue?

attiland · May 24, 2021

Not sure if this helps but;
"A voltage between 1.8V and 3.3V will be read by the Raspberry Pi as high; anything lower than 1.8V will be read as low." My guess is you have voltage close 1.8V sometimes or you don't check often enough so you miss some of it.
This is where I would start anyway
Hope this helps.

epicfatigue · May 24, 2021

attiland said:
Not sure if this helps but;
"A voltage between 1.8V and 3.3V will be read by the Raspberry Pi as high; anything lower than 1.8V will be read as low." My guess is you have voltage close 1.8V sometimes or you don't check often enough so you miss some of it.
This is where I would start anyway
Hope this helps.

I'll setup the scope and monitor it I check every 1 second

robsworld78 · May 24, 2021

epicfatigue said:
Hey everyone,

Recently setup reefpi and i having some issues.

I have setup an ATO using a Non contact water sensor, and I have it connected to a level shifter.

It works fine majority of the time but sometimes I am noticing Reefpi isn't detecting the 3V3 from the level shifter.

Anyone had this issue?

It could be caused by a floating pin, try adding a pullup resistor to 3.3v on the data line coming out of level shifter going into Pi.

epicfatigue · May 25, 2021

robsworld78 said:
It could be caused by a floating pin, try adding a pullup resistor to 3.3v on the data line coming out of level shifter going into Pi.

Was thinking this Rob, just a 10k ?

robsworld78 · May 25, 2021

epicfatigue said:
Was thinking this Rob, just a 10k ?

Yeah anything between 1k - 10k should be ok.

epicfatigue · May 27, 2021

Well not sure what was going on here, im almost suspecting a crap logic shifter, it appeared it would get stuck on, pull ups pull downs no avail.

I have ordered another and bypassed it for the time being.

Kinda a silly way of doing it but i had a spare channel in the 16 relay PCB so i did the following

Tied the GPIO pin 20 to ground with a 1k resistor, and connected to a common of a spare relay.
Tied the NC to 3.3v in series with a 1k resistor.
Tied the non contact sensor output wire to the relay control pcb.

So this way when the sensor (NCWS) is active nothing is going to the GPIO and it is pulled low with the 1k resistor to ground.
When NCWS toggles it will turn off and connect the 3V3 in series with the 1k resistor to the GPIO pulling it HI.

i did 100 random samples and it didnt once fail, where as the others using the shifter would fail after 3-5 attempts mostly when the duration is short.

Any ideas?

Tempted to do away with the lot and use a voltage divider with a pulldown resistor, to emulate what i am doing with the relay.
I just couldn't be bothered soldering and it was a quick dirty check.

epicfatigue · May 28, 2021

madness this failed aswell....
Only thing i can now put this down to is either a reef-pi issue or a raspberry pi issue.

Currently re-installing, after that ill swap out the pi....
pretty lost here.

@robsworld78 any ideas?

Things tried to date with XKC-Y25-NPN V
current config is
5v to VCC
GND + mode tied
Output Device

Tried using level shifter without pullups
Tried using level shifter with pullups
Tried using level shifter with pull downs (it hated that hahaha worth a try)

Controlling a relay to shift hi and low manually

Reef-pi set to read every 1 second

robsworld78 · May 28, 2021

Hmm, that is strange. What voltage do you get on the signal wire going to Pi when you touch sensor vs released? As you touch and release it should go from 0v to 3.3v.

epicfatigue · May 31, 2021

Okay so i cracked it and finally got out the crow and did some testing, silly me should have done this ages ago.
I was confident with my logic that this was not needed and have now tracked down the issue to Reefpi and will raise a defect.

I have my Disable on Alert set very short as i have a very fast ATO pump, set to like 3-5 seconds.
I can have the sensor tripping and going on and off all day long with alerts disabled and never hit an issue.

When i have disable on alert, if i trigger it multiple times, Reefpi disables the ATO function but leave the pump running, after awhile it eventually comes back and disables the pump, I'm suspecting its getting stuck on a call disabling the ATO

See the below videos for a real world demo, this explains why i never had any issues when testing as i never had ALERTS ON!.

@robsworld78

attiland · May 31, 2021

epicfatigue said:
Okay so i cracked it and finally got out the crow and did some testing, silly me should have done this ages ago.
I was confident with my logic that this was not needed and have now tracked down the issue to Reefpi and will raise a defect.

I have my Disable on Alert set very short as i have a very fast ATO pump, set to like 3-5 seconds.
I can have the sensor tripping and going on and off all day long with alerts disabled and never hit an issue.

When i have disable on alert, if i trigger it multiple times, Reefpi disables the ATO function but leave the pump running, after awhile it eventually comes back and disables the pump, I'm suspecting its getting stuck on a call disabling the ATO

See the below videos for a real world demo, this explains why i never had any issues when testing as i never had ALERTS ON!.
https://photos.google.com/photo/AF1QipNxUcM_OrxajviZIhMgpf6gV4triONQ69xNYEp9
https://photos.google.com/photo/AF1QipNxUcM_OrxajviZIhMgpf6gV4triONQ69xNYEp9

I think you have set your video private

epicfatigue · May 31, 2021

attiland said:
I think you have set your video private

haha i was fixing that as you were typing had to upload them to youtube check now ?

attiland · May 31, 2021

epicfatigue said:
haha i was fixing that as you were typing had to upload them to youtube check now ?

Good now

robsworld78 · Jun 1, 2021

epicfatigue said:
Okay so i cracked it and finally got out the crow and did some testing, silly me should have done this ages ago.
I was confident with my logic that this was not needed and have now tracked down the issue to Reefpi and will raise a defect.

I have my Disable on Alert set very short as i have a very fast ATO pump, set to like 3-5 seconds.
I can have the sensor tripping and going on and off all day long with alerts disabled and never hit an issue.

When i have disable on alert, if i trigger it multiple times, Reefpi disables the ATO function but leave the pump running, after awhile it eventually comes back and disables the pump, I'm suspecting its getting stuck on a call disabling the ATO

See the below videos for a real world demo, this explains why i never had any issues when testing as i never had ALERTS ON!.

@robsworld78

Nice find, maybe create a new issue report on Github so @Ranjib and others know to look into it.

Issues · reef-pi/reef-pi

An opensource reef tank controller based on Raspberry Pi - Issues · reef-pi/reef-pi

github.com

epicfatigue · Jun 1, 2021

robsworld78 said:
Nice find, maybe create a new issue report on Github so @Ranjib and others know to look into it.

Issues · reef-pi/reef-pi

An opensource reef tank controller based on Raspberry Pi - Issues · reef-pi/reef-pi

github.com

@Ranjib @robsworld78
Yeah the more I think about it I question if that disable on alert is actually not what we think it is, it's almost like is a maximum amount of time per X time.

I will do some more testing before a raise a report and I might look at the code because that polling period I wonder if it's check frequency changes once it's triggered. Because I wonder if if it's set to per 120 seconds it might poll that often but once it's changed state it might be making a thread to poll it more often, at least that's how I would have wrote it, in which case me setting it super low might cause an issue.

robsworld78 · Jun 1, 2021

epicfatigue said:
@Ranjib @robsworld78
Yeah the more I think about it I question if that disable on alert is actually not what we think it is, it's almost like is a maximum amount of time per X time.

I will do some more testing before a raise a report and I might look at the code because that polling period I wonder if it's check frequency changes once it's triggered. Because I wonder if if it's set to per 120 seconds it might poll that often but once it's changed state it might be making a thread to poll it more often, at least that's how I would have wrote it, in which case me setting it super low might cause an issue.

Interesting, maybe that is the case. Let us know what you find.

Ranjib · Jun 5, 2021

epicfatigue said:
Okay so i cracked it and finally got out the crow and did some testing, silly me should have done this ages ago.
I was confident with my logic that this was not needed and have now tracked down the issue to Reefpi and will raise a defect.

I have my Disable on Alert set very short as i have a very fast ATO pump, set to like 3-5 seconds.
I can have the sensor tripping and going on and off all day long with alerts disabled and never hit an issue.

When i have disable on alert, if i trigger it multiple times, Reefpi disables the ATO function but leave the pump running, after awhile it eventually comes back and disables the pump, I'm suspecting its getting stuck on a call disabling the ATO

See the below videos for a real world demo, this explains why i never had any issues when testing as i never had ALERTS ON!.

@robsworld78

Which version of reef-pi you are using, I remember there was a bug like this that we have fixed in past