Hall sensor and magnetic polarity? #87
Replies: 6 comments 7 replies
-
|
Hi @Gordon-Shumway2 , I have a quick question, the Hall sensor (a3144) in the ky-003 is 4.5v rated (i.e. needs 5v for proper operations). The pi generally uses 3.3v gpio-s. How do you connect the sensor to the pi? Do you use a voltage divider or anything? I am just wondering whether the module has enough supply voltage to operate properly. |
Beta Was this translation helpful? Give feedback.
-
|
Hi Abasz, the website where I bought the sensor (https://www.az-delivery.de) states "Betriebsspannung 3.3 V - 5 V" but I agree that there are datasheets for this sensor stating 4.5 V - 5 V. Cheers, |
Beta Was this translation helpful? Give feedback.
-
|
They are probably using a latching hall sensor. This type of sensor require to be "switched" back with an opposite pole magnet. Do you get 3 signals per rotation or 6? (I would assume that you get 3, similarily the C2 getting 6 with 12 magnets) EDIT: The advantage of this is that the high state can be controlled (as it will remain high until the opposite pole magnet arrives) and this could ensure that there is sufficient length of high state is present for a stable signal and hence read. |
Beta Was this translation helpful? Give feedback.
-
The C2 Model D/RowErg uses a 12 pole magnet to drive a generator circuit: it actually produces power to feed the PM5. My perception is that the data pin provides a rectified generator signal (sinoid-like, fluctuating between 0V and 15V), where the power pin gets a stabilized feed. |
Beta Was this translation helpful? Give feedback.
-
|
Hi Micheal, that is a very interesting find! I found a similar issue in a Concept2 (also 6 magnets), where I found a curve-like deviation with a period of 6: I have the impression it is deliberate, they seem to create some sort of frequency modulation. What helped stabalise my data was by using a flanklength of 12 (i.e. two full rotations, forcing a similar mix of high and low measurements, with a second period to stabalise). |
Beta Was this translation helpful? Give feedback.
-
|
Hello Jaap, yes, I saw that graph in the documentation of the backend_redesign branch - it was actually the reason I gathered my data to compare with :) You might want to check that sixth magnet. On my setup moving another magnet close to any of the six magnets instantly showed either Modifying flanklength is one of my options but I am using a Pi Zero 2 and your line about "What the practical limit on a Rapberry Pi Zero 2 W is, is still a matter of investigation." made me explore other avenues first. I might try to tap the original sensor's data first and feed it into the Pi. Why are there so few datapoints per stroke in your C2 graph compared to mine? Yours seems to show six measurements from minimum to minimum while mine has has about 40. Was your data condensed somehow? Keep up the great work, Michael |
Beta Was this translation helpful? Give feedback.
-
As the PM5 works well, I assume it works as intended. I do have a question outstanding with someone who measured the PM5 signal in the first place, to see if it is a one-off issue or systematic. After that, I guess I should contact C2 support with a more detailed question. The magnets in the C2 are pretty tough to reach, they are placed at the back between the flywheel and a fixed plate, near the axle. I am not that curious yet to find out what they did.
The thing is I have no clue how the Zero is using its CPU. I had my 4B on 24 datapoints (in the experimental branch Abasz and I are working on, we can trigger twice on a single magnet), and it handled it easily. That was a pleasant surprise to me as well, as some calculations are growing fast when flanklength is increased. Easiest way to find out is to replay a session and set the flanklength, but it could remove a lot of issues you might want to investigate first.
I typically have 80 datapoints in the drive section, and around 200 in the recovery. However, I tried to illustrate this specific behaviour, so I grabbed 3 full rotations at the bottom of the drive (i.e. almost no acceleration/deceleration). I was quite surprised by it as well, and I have the strong suspicion it is intentional/acceptable as C2's quality control seems to be pretty tight.
I will, I love this kind of puzzles. |
Beta Was this translation helpful? Give feedback.
-
|
Hi Abasz
I am getting a signal every time the magnet approaches the sensor. One
rotation yields six signals.
Cheers,
Michael
Abasz ***@***.***> schrieb am Do., 6. Okt. 2022, 13:16:
… They are probably using a latching hall sensor. This type of sensor
require to be "switched" back with an opposite pole magnet. Do you get 3
signals per rotation or 6? (I would assume that you get 3, similarily the
C2 getting 6 with 12 magnets)
—
Reply to this email directly, view it on GitHub
<#87 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ASXHKG2HMLFIHFZ6DR6BAO3WB2YHTANCNFSM6AAAAAAQ5MABC4>
.
You are receiving this because you were mentioned.Message ID:
***@***.***
com>
|
Beta Was this translation helpful? Give feedback.
-
|
Tada! Jaap mentioned that the C2's magnets were hard to access so I had a closer look at mine. A little push and they came off. Just a tight fit, not glued to the holes in the disc. I turned three magnets upside down and the issue is gone. Lovely graphs, no outliers. The rower's console reports the same (slight pleasing to my fitness) data as before - it is not affected by the polarity switch.
Thank you all! |
Beta Was this translation helpful? Give feedback.
-
|
Wow, that looks cool! Jaap |
Beta Was this translation helpful? Give feedback.
-
|
Just a quick wrap-up of this issue: I got in touch with the manufacturer (First Degree Fitness). The order/sequence of magnets' polarity is not crucial to their console's inner working, so switching the magnets does not influence it. And they are using a reed switch while I used a Hall sensor - and I believe a reed switch does not suffer from being biased to the polarity and hence it gets correct data from every magnet. Michael |
Beta Was this translation helpful? Give feedback.
-
|
Indeed, reed does not care about polarity. It's basically a mechanical switch triggered by magnetic traction (from either poles). |
Beta Was this translation helpful? Give feedback.
-
|
Michael, Are you willing to share the configfile and some raw stroke data, allowing us to make this a supported machine? Jaap |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi,
an interesting issue regarding sensors (rather than ORM):
I have a FDF Neon Pro V rower, which uses six magnets evenly spaced on a disc attached to the main shaft. ORM is running on a PI Zero 2 W, magnetic sensor is a KY-003 Hall sensor. Setup works fine, and has been used with this rower for some months.
When testing the backend_redesign branch I gathered raw csv data and created a graph.
The graph (one showing start of the row, one showing data mid-row) has regular outliers every three data points, alternating between slower and faster measurements.
So the pattern is
(... repeat ...)
It turned out that
If you move the sensor from the left to one of the south magnets it will react about 5mm before you reach the magnet.
If you move the sensor from the left to one of the north magnets it will react about 5mm after you have reached the magnet.
Same applies if the magnets are moved and the sensor is in a fixed position as is the case with our rowers.
This causes either a delay or a speed-up with the data reading every time that the polarity of magnets changes (which is every three magnets).
ORM still delivers sensible numbers even though they sometimes seem to jump unexpectedly (such as a 2:14 pace suddenly turns into a 2:10 at the end of a stroke before going back to 2:14).
Keep up the great work on ORM!
kind regards,
Michael
Beta Was this translation helpful? Give feedback.
All reactions