Group Actions
Engagement
  • Author Author: verrverr
  • Date Created: Date Created
  • Views 1886 views
  • Likes 0 likes
  • Comments 8 comments
Related
Recommended

Pas Home / 1-wire vs Tektronix oscilloscope

verrverr

Pas Home - Index

 

Hello, this week I did a lot of testing and debugging with the oscilloscope. I try to see and solve the problem of my network of temperature 1-wire sensor.

 

 

I try first to change my connection to the 5v and 3.3v use, I also change the cable connection together. Using the 5v is really not advise for raspberry pi but its still work reasonably well for 6 months.

Result : failure no captor are detected

 

imageimageimageimage

 

Then I tried to change the way power sensors using the parasitic mode. As some suggest me in another post

Result : no difference, no captor detected

 

Here are some screenshots of what I normally should have as working signal

With 1 sensor on 3.3v

imageimage

With 1 sensor on 5v and level switcher

imageimageimageimage

According to this guide Guidelines for Reliable Long Line 1-Wire Networks - Tutorial - Maxim my curve have slow rise time and maybe some reflected problem


Finally I retest on 5v but with a level converter https://www.adafruit.com/products/757 to send only 3v on raspberrypi port

Result : Success all captor are detected and all communicated data are correct

imageimageimageimage

Interesting curve. Yellow is the 5v side data and Blue the converted 3.3v result

image

 

I'll do other tests removing resistors supposed to improve communications and see if it really helps and I will also make a soldering version on a perma-proto board.

Do you see anything else that could improve my circuit?

 

Old schematic

image

 

New improve schematic

image

  • Have you thought about using a DS2482 as a controller? It's an easy way to connect Onewire to I2C, and avoids the problems caused by bitbanging the Raspi GPIO.

    You can get the chip as sample from Maxim and it's quite easy to use. I'm currently designing a board to connect to Raspi which uses this chip and I've tested it on breadboard without any problems.

    Using the chip would also reduce the number of external components as well as give you active pull-ups, so the communication should stay a lot cleaner.

     

    There's some notes on using the DS2482 in the Guidelines for Reliable Long Line 1-Wire Networks which are worth checking out.

     

    I've been using DS9490R for Onewire-networks with cable lengths up to 40 or 50 meters without any problems. The sensors (DS18x20) have been parasite powered with no external components (no pull-ups, series resistors etc), just connected to the DS9490R directly. No problems what so ever.

     

    I could try and post a brief blog post about using the DS2482 next week, I've been busy with EPE'14 for the last couple of weeks and really haven't got anything done regarding this project. However, I've had some ideas and now it's time just to get them done image

  • I don't use 1-wire enough to know, but it definitely looks like the pull-up is insufficient, and needs to be a far lower resistance, maybe halve it, as a test? (Assuming you can do that according to the spec).It definitely doesn't look healthy : ( could perhaps need a bypass capacitor in case these devices are resetting since bit spacing looks odd. In general that should be 100nF ceramic, and possibly some small electrolytic too. At least a 100nF ceramic by each device.

  • Do you know the capacitance per foot of the wire you are using?