Activity – Marion

Marion replied to the topic Sleep code not working with Maxbotix 7389 in the forum Mayfly Data Logger 4 years, 10 months ago

We just finished the last of a series of tests of the Maxbotix. With the changes which Sara suggested (close the serial port after each log interval and restart it for each new consectutive interval), the sleep code does work if the sensor is continuously powered.

Setup: Maxbotix 7389 sensor with the target either stream water or a fixed/flat metal plate approx 1/2 meter on a side. Before all testing (with 2 sensors), the sensor faces were coated with a very thin coating of fluoropolymer grease (very hydrophobic) to help shed any condensation.

Results:
1. Sensor mounted over a standpipe in water. Ambient temperature near 20 C. Frequent bad data (300) with normal serial read using code originally supplied by Shannon Hicks. Sensor powered by Mayfly digital pin only when logging.

2. We suspected that there might be some interaction between the sensor and logger. Tried reading in the serial data with 50 milliseconds delay between each digit. This very slow read out of the data- have never seen this in the literature, gave a partial solution. This eliminated the bad data until fall when temperatures dropped down to about 10 C at night. Then bad data again occurred.

3. Sensor tested with a boom mount over the water or metal plate with intermittent power, but no standpipe. Results still showed some bad data points, even with traps to reread up to 20 times in the event of bad data. The bad data occurred at random times, when the temps were well above freezing and below freezing. The conclusion is that condensation on the sensor face is not the issue.

4. The final test was a 10 day test, 5 minute intervals, with sensor continuously powered independently by a separate battery power pack. During that time, the angle of the sensor from perpendicular to the target was varied from 0 degrees up to 5 degrees (to help shed potential condensation if any occurred). During that time, the temperatures ranged from as high as 15 C and as low as -5 C. Absolutely no bad data points were measured during the entire test.

Conclusions:
a. We saw no evidence that the sensor had any issue related to condensation during the tests given that there was a hydrophobic coating on the sensor face and that the bad data occurred at random temperatures as opposed to only when the temperatures were low.

b. Reading out the serial data with a pause between digits rather than at normal speed seemed to help eliminate bad data when the weather was warm. But, this did not solve the problem when cooler weather arrived in the fall.

c. There was no bad data during the testing of the sensor with continuously power independent of the Mayfly, including at temperatures well below freezing.

Recommendation: If the Maxbotix 7389 sensor is used and a criterion is to log absolutely no bad data (such as when the data is used to control a pump sampler as we intend), it is recommended that sensor power be supplied continuously by an independent power pack.

I will be happy to forward or post data files if anyone wants to see the raw data.
- Marion replied 4 years, 10 months ago
  
  Sara: <My responses re your comments:
  
  a. The fluoropolymer grease used was Krytox. It is very expensive, but a very small amount is required to coat a sensor face. The advantage vs normal hydrocarbon based greases (such as Vaseline) is that this polymer is very resistant to oxidation and/ or evaporation of volatile components which leads to hardening or degradation of the properties. So, in short it will last indefinitely and of course is not water wettable or water soluble.
  
  b. I stand by the result I report. I have the files of data collected all summer. There is nothing I have found in the literature, but it did solve the issue till the weather got much cooler in the fall. Inserting a pause/delay between reading in each character definitely worked for several months during the summer- absolutely no bad data. I believe that a possible explanation is that somehow the act of reading the serial data introduces some momentary instability in the sensor chip. I might add that using a more sophisticated function to read the sensor, such as "parse int", should have worked at least as well as the code written by Shannon. But on the contrary, the results were much worse for the two Mayflys and two sensors we have on hand.
  
  c. Re the idea of powering up for 5-10 sec before the data readout. We tried this and it did not solve the problem. Admittedly that power was supplied via a digital pin and not from one of the switched power Grove sockets.
  
  I am happy with the current solution of using an independent power supply. This has been extremely frustrating. Frankly I have been surprised that others have not complained about the issue. Up to this point I would not have recommended this type sensor to anyone.
  
  Thanks