Flow Sensor Counts Not Accurate in Firmware 2.1.9 (3)

[bena]

It sounds like some work is being done on a new firmware version so I wanted to see if the following two potential bugs could be addressed:

I have an OpenSprinkler 3.2 DC and I recently upgraded firmware from 2.1.9 to 2.1.9(3).

After I upgraded I noticed that the flow sensor was no longer accurate.

I tested this further by disconnecting my reed switch flow sensor and manually shorted the GND and SN1 pins multiple times manually while a station ran. OpenSprinkler was very inconstant on the number of clicks it counted and never matched my manual count.
I made sure my L/Pulse setting is 1 and I factory reset the unit with the same inaccurate results.

When I downgraded to Firmware 2.1.9 (0), it worked correctly again and matched my manual counts every time. I re-upgraded and the bug returned.

Second, I noticed the API jl function has stopped reporting flow counts for each station run.
The data returned looks similar to this where the flow count field is always 0.00:
[3,2,180,1586908985, 0.00]

I know the OpenSprinkler is still capturing the sensor data, as the jl type=fl option still returns data correctly.

Thanks

[Ray]

I believe the code for handling sensors is exactly the same between 2.1.9(0) and 2.1.9(3). I double checked the two versions source code and didn’t see any difference. So I can’t explain why it makes a difference on your side. I will do some additional test of 2.1.9(3) on flow sensor, but I’m pretty sure I’ve tested it before and it works fine.

Regarding flow count at the end of each station run vs. the type=fl: they are not the same: the former is recorded at the end of a single station run, the latter is recorded when the program queue is empty (i.e. all stations finish running). It’s possible that the former is 0 (because it didn’t register any click during the station run) but the latter is not 0.

[bena]

After further testing with my flow sensor connected, I am seeing values from the API jl call as well as with the type=fl parameter. It appears to be working correctly as you stated.

However, even after further testing by manually shorting the sensor pin to ground or by attaching my flow sensor and trying different firmware versions, the counts are still not accurate with 2.1.9(3) as compared to 2.1.9(0).

Let me know if you are not able to re-create this even by manually shorting the sensor pin 20-25 times for a 1 minute station run. If not, then maybe there is a problem with my specific unit (OpenSprinkler 3.2 DC).

[Ray]

I just tried 2.1.9(3) and cannot reproduce the issue you reported: it correctly registered all the button clicks I did.

Since you said it’s inaccurate: how off is it?

Also, I assume you are using the stock firmware 2.1.9(3). If you compiled the firmware yourself, I can think of many reasons why it may not be accurate.

[bena]

Thanks for double checking this.

I just re-ran two tests with the following results:

Ran a station for 20 seconds and manually shorted the sensor 25 times for each firmware:
2.1.9(0) – recorded 25 pulses
2.1.9(3) – recorded 21 pulses

Ran a station for 1 minute and manually shorted the sensor 101 times for each firmware:
2.1.9(0) – recorded 101 pulses
2.1.9(3) – recorded 19 pulses

See attachment for OpenSprinkler log results for each test.

Also, I am using stock firmwares for both versions.

Attachments:

1. 

   Test_Results.jpg
   

[Author]

Posts