Obsidian surge 500, Stroke Detection Issues

[LGallis]

I was able to get everything working besides the stroke detection. I have an obsidian surge 500 water rower, and I was able to get quite accurate distance calculations and have tried just about everything to get the strokes to detect but nothing seems to work. At best, I can get 2 detections across 20 strokes and typically it detects none at all. I think my data is relatively clean, but I have attached the raw csv, an image of the log which shows that the recovery stroke conditions are never met, and an image of my current config settings.I have attached a single magnet to the axle of the rower, as well as repurposing one of the hall effect sensors as the original magnet was on the handle spool which moved back and forth. Any and all help is appreciated.


2026-02-25_06-54-35_raw.csv

[JaapvanEkris]

Your settings are far from optimal. With the following I get better results:

  rowerSettings: {
    numOfImpulsesPerRevolution: 1,
    sprocketRadius: 2,
    maximumStrokeTimeBeforePause: 6.0,
    dragFactor: 31000,
    autoAdjustDragFactor: false,
    dragFactorSmoothing: 3,
    minimumTimeBetweenImpulses: 0.18,
    maximumTimeBetweenImpulses: 0.50,
    flankLength: 3,
    systematicErrorAgressiveness: 0,
    systematicErrorNumberOfDatapoints: 1,
    minimumStrokeQuality: 0.20,
    minimumForceBeforeStroke: 25,
    minimumRecoverySlope: 0.08,
    autoAdjustRecoverySlope: false,
    autoAdjustRecoverySlopeMargin: 0,
    minimumDriveTime: 0.40,
    minimumRecoveryTime: 0.90,
    flywheelInertia: 1.0,
    magicConstant: 2.8
  }

It isn't perfect yet.

Please check if the magnet is connected to the impellor, not to the handle. The exploded view in the manual isn't conclusive about that.

I also see a ratio of 1:0.6 (recovery is way too short), which also might indicate that not rushing the recovery might provide better data. But as I have no insight in how the stroke is composed, it is difficult to optimize.

[Coulombo]

Obsidian Surge 500.pdf
G

Greetings Friends,
Jumping in here because I'm about to start this project on the Obsidian Surge 500. I have taken pictures of my machine, and it does, in fact, have a magnetic pickup off the "flywheel" or main pulley that is attached to the hydro blades. It is a 1:1 ratio, and I have timed it myself.
I do have a question. Why does this have two separate magnetic pickup sensors? Seems redundant, especially as they are located right next to each other. I could see the benefit of having them 180 degrees apart for accuracy, but I don't see why you would need two.

Could @JaapvanEkris help me with the configuration profile for this with this information? I will be building it to use all the the manufacture original hardware and tapping into the pickup to drive the GIPO input.

[LGallis]

That magnet is on the handle wheel, mine also had the same magnet/sensor placements. As that magnet moves in both directions since it is attached to the handle, it needs two sensors to discern which direction it is moving in. In my machine, I bought some small 4mm diameter magnets and placed a few of them on the shaft in the small space between the original wheel with the handle and the box below it. I don't know the software well enough to help with implementation of your current setup, but if you'd like help with placing the magnets I can help. In mine I simply placed the magnets using a long pair of pliers and then removed the sensors, and wired one of them into the raspberry pi. The other one I used to do testing and verify that it worked the way I expected.

[Coulombo]

I greatly appreciate your sharing that information with me. Would you mind sending some pictures over of how you have it set up?

[JaapvanEkris]

Could @JaapvanEkris help me with the configuration profile for this with this information? I will be building it to use all the the manufacture original hardware and tapping into the pickup to drive the GIPO input.

I'm always willing to help here. Please make sure you measure on something that moves with the impellor/blades amd not with the handle. For some pulleys, this is difficult to assess.

In the config you can turn on recording of raw files. If you can provide such a file from 50-ish strokes, I'll see if I can make it work better.

[Coulombo]

@JaapvanEkris , I am going to put a magnet on the shaft like @LGallis has. Ill send photos of placement and send the raw files once i get the system up and running. I am waiting for a few components to ship.

[Abasz]

the hall effect sensors

Just for the sake of accuracy and for future reference, on the pictures those are not hall sensors but reed switches. Those are passive component (unlike hall).

From a practical perspective this might not be too relevant but the noise handling of those differ (reed can suffer from bounce, which with so few magnets probably not going to be an issue).

[LGallis]

Sorry I totally forgot to send pictures of my setup. While its certainly far from optimal, I was able to get consistent responses from the reed switch like this

The magnet stack is on the left, and the reed sensor is on the right. Afterwards, I ran the reed wire using the same path it was originally wired through for the monitor, and replaced the monitor box with a 3d printed enclosure containing the LCD and raspberry pi

[Coulombo]

Sorry I totally forgot to send pictures of my setup. While its certainly far from optimal, I was able to get consistent responses from the reed switch like this The magnet stack is on the left, and the reed sensor is on the right. Afterwards, I ran the reed wire using the same path it was originally wired through for the monitor, and replaced the monitor box with a 3d printed enclosure containing the LCD and raspberry pi

Do you have the STL for the pi enclosure? I'm sure I can design one but.....if you've already done the hard work I'd really appreciate if you'd share it. I am going with the HiLetgo 5pcs Hall Effect Magnetic Sensor Module 3144E A3144. I am going to 3d print sometype of holder and figure out a mounting solution so you can adjust the clearance.

[LGallis]

My internet is currently down, but Ill send over the STL when it comes back up! I am using some 7 inch touchscreen lcd from amazon and a zero2w. I do also have a cutout on the right which I use to switch on/off the monitor that way I dont get any screen issues over time

[Abasz]

I am going with the HiLetgo 5pcs Hall Effect Magnetic Sensor Module 3144E A3144.

Please be aware that it might be rated for 5v while the rpi vcc is only 3.3v so you might need separate power supply. But I must say that I found some datasheets for the E variant that says 3.8v which might work with 3.3...

Test it and if you encounter issues try changing the voltage. The potential issue with these pre-made boards in general is that they use all sort of components randomly so its difficult to know what is on the board :D

Since these halls are open drain it should not be a problem if you connect 5v to the supply pin but the output pin is on the 3.3v rpi gpio. But please bear in mind that under certain worst case scenarios if you apply 5v to the rpi there is a chance that you damage the rpi (worst case is generally if the hall "breaks" down and do not function or if you short the 5v to the output pin and so on)

[LGallis]

Do you have the STL for the pi enclosure? I'm sure I can design one but.....if you've already done the hard work I'd really appreciate if you'd share it. I am going with the HiLetgo 5pcs Hall Effect Magnetic Sensor Module 3144E A3144. I am going to 3d print sometype of holder and figure out a mounting solution so you can adjust the clearance.

Here is the pi enclosure STLs, enclosure being the majority of the box and faceplate being the plate on top for aesthetic purposes
Enclosure.zip