Replies: 5 comments 5 replies
But it is a nice looking machine. We'll help you figure it out. There are multiple solutions available, and we just have to figure out what works best for you.
OpenRowingMonitor isn't really designed to handle this structure. We like to see what the impellor does, as it tells us how fast it decellerates (a key parameter in rowing). We typically assume a single direction movement. So there are four basic approaches:
In the end indeed. Jaap |
|
Hi Jaap Thanks for your reply. And thanks very much for your help !
Yes I have been reading the excellent docs - particularly the 'physics' doc and the rower-settings doc. I do understand the advantages of knowing the impeller velocity. The older Merach R23R1 water rower seems to be a close clone of the Topiom V2 water rower ... and I think if I had one of those I could get access to instrument the impeller drive pulley, which would be a great solution. I now feel that the R23R1 would have been a better choice for me to buy ... but that's hindsight :( Alas the R28R1 has changed the belt drive considerably. On the R28R1 the belt runs over a top pulley (the one I have instrumented up), a bottom pulley, then drives the impeller via a pulley with a clutch. This is directly under the water tank. But the clutch is internal to this pulley, and so the clutch pulley flank mirrors the movement of the top pulley. However underneath the machine frame the impeller shaft is just visible. I may be able to glue a N52 type magnet to the shaft and I may just have enough clearance to mount a hall effect sensor to read it. It's difficult, and I'm not certain of this yet. Work in progress ... As for adding a magnet to the impeller via the water fill hole - I think that is a none-starter. The bits of the impeller that are accessible via the fill hole are 5 - 8 cm away from the edge of the tank, too far for a magnetic signal I fear. The only option would be to split the tank, and attach the magnet to part of the impeller closer to the edge of the tank. But I think my chances of regaining water tightness after that would be zero ! I had also thought about optical sensing of the impeller blades, but because the tank is a darkend plastic, and because of all the water churn I suspect that this would not be reliable. I may investigate further as time and frustration play out. So I will be reading the rower settings doc again and trying to see if I can filter out enough of the bad recovery data to make the monitor useful. The only things I would like to get would be No of strokes in the session The current Merach monitor only gives me No of strokes ! the rest of the numbers it provides are basically made up ! |
There is an option there: the SmartRow pulley. It is a replacement of the stock pulley of the WaterRower (the brand, not the type) that is instrumented correctly. Don't know if it will fit, and it is expensive. But might be an alternative here.
Perhaps some 3D printed part to bring it more outside?
We'll be better, even just as a basic fixed setup. But being able to measure impellor speed is a major improvement.
Yeah, that is a much more general issue, hence ORM was developed. |
|
Hi Jaap Thanks again for your thoughtful reply.
Yeah I've heard of this. RowAlong John (UTube rower) has one, and he was going to try to fit one in a Merach R28R1 :) However the project seems to have gone strangely silent. Having seen photo's of the SmartRow pulley I suspect that the modifications required are in the 'angle grinder and welding set' league, but I may be mistaken, I'm only going on photos. Underneath the machine I can see the end of the impeller shaft. The shaft end is flush with the machine bottom frame plate, so my only option would be to a) drill and tap the centre of the shaft for an M3 or M4 bolt. That would allow me to fasten a collar to the shaft to extend it away from the frame plate, and then arrange a magnet and sensor. Or b) drill 1 or 2 mm hole near the edge of the impeller shaft but parallel to it's axis, sink a magnet inside that, then arrange a sensor below the shaft. Both are difficult. And there is only about 15mm clearance between the bottom frame plate and the carpet floor of my gym (bedroom ...). So perhaps calling it 'Work in progress' is an exaggeration ... maybe I'm still at the 'Feasibility study' stage :) -- Chris |
It is a company, but I know they have limited shipping for some time (Netherlands or EU only?)
Do they go through and through? Surprising. Perhaps a white sticker and optical might be easier for an initial try?
Yeah. All the Pi needs is a dry contact. So even approaches which complete a circuit in another way might work |
Now that's a clever thought ! Thanks. It will be pretty dark down there, but with a bit of work I could blacken the end of the impeller shaft, then add a white line (tipex ?), and a biased led for gentle illumination. I'll see what I can engineer. Thanks again for your help !!! -- Chris |
|
So a quick update ... I don't have a working monitor yet, but feel I'm pretty close. As I described the impeller shaft is visible on the underside of the machine. I have made a plastic cap which will fit over the shaft and hold a TCRT500 reflective optical sensor. The shaft has been painted matt-black and a tipex white spot applied to the edge. I have tested the cap + sensor with a test rig and the signal looks good. I get a duty cycle of about 10%, so a ~14 ms signal pulse with a 140 ms repeat length which I estimate is what I'll see on the rower. Yet to be worked out ... As currently designed the sensor will only give me 1 reading per revolution of the impeller. I reckon that a typical stroke is 7 - 8 impeller revolutions, which does not seem many data points. I could try to add a second tipex spot to the shaft, but suspect that doing that would introduce considerable jitter to the signal (as the spot position and reflectance would probably be incorrect). Will the 7 - 8 readings per stroke be enough ? I'm currently assuming that the Topiom_V2 water rower should be a good place to start my settings, but does anyone know how to estimate the drag-factor and Moment of Inertia for water based rowers ? I will pass along my findings in due course. -- Chris |
|
As expected this has taken longer than I thought ... A couple of photos show the underside of the rower with the marker disc glued to the end of the impeller shaft, and a sensor cap attached which carries the optical sensor and illuminator (secured with gaffer tape !). The box which holds the Pi, and the interface module for the optical sensor can be seen at the lower left. I have a couple of questions ... Q1) I'm trying to use the cyclic error corrector as the marker disc is not aligned perfectly, but notice the following warnings in the syslog journal at startup Could someone explain what the preconditions are, and why when I specify systematicErrorAgressiveness and systematicErrorNumberOfDatapoints I'm not meeting them ? Q2) The pace / force / and probably a lot of other outputs are critically controlled by the flywheelInertia parameter, but I really have no idea what this is. I started out using the same value provided by the Topiom water rower (0.8), but that seemed WAY to high and showed me exerting ~ 2000N on the handle :). I've currently got it set to 0.4, but have now real basis for that figure being correct. Could someone suggest how to tune this parameter. I have included the my new section from rowerProfiles.js, a copy of the raw rowing times, and the rowing data showing 20 strokes at about 20 str/min. The raw data shows the timing irregularites clearly. Notwithstanding the points above, this is now doing 95% of what I need. :) -- Chris PS. One thing that I did notice was that the Topiom 'sprocketRadius' parameter is misspelled as SprocketRadius. enc: ORM-marker-disc.jpg |
Looks cool!
autoAdjustDragFactor has to be true as well. The reason is that the CECFilter uses data from the dragslope calculation to identify errors in the pulse timing. If you can't trust that calculation, it would be foolish to use it to correct data.
It usually involves a lot of rowing :) and seeing if you find the pace/power/distance credible for the effort you just did. It helps if you can sometimes test what your effort brings you on a Concept2 (most gyms have one). Did that for my NordicTrack back in the days, and you get pretty good results.
So involving the CECFilter will probably help.
Nice to hear. Thank you.
That doesn't help :) Will fix that one in the next release :) |
|
The Last Post (cue trumpets ...) This is now working well, and I'm very pleased with the results. The only aspect I cannot get working is the autoAdjustRecoverySlope. I have set this false. The problem is that if I enable it, the syslog is filled with a lot of 'Delta Time trend is downwards' warnings ... I assume that this is because the calculations here are made before the delta-t times are filtered by the Cyclic Error Corrector, and due to the noisy nature of my raw data the recovery slope calculation picks up bad data. No matter, the 'goodness of fit' numbers show about 0.997 so I guess the recovery slope calculation is pretty solid. Stroke detection is 100%, and drive duration / recovery duration both seem to be about right. I'm including a snippet from the rowerProfiles.js file to show my final parameters. I didn't want to pollute this discussion with lots of 'I tried this and it failed, but I tried this and it worked ...' details, but I have written up a reasonably detailed guide to what-I-did, how-it-works, and how-to-make-one at instructables.com. This includes .stl files for the parts I designed and made. https://www.instructables.com/Using-a-Merach-R28R1-Water-Rowing-Machine-With-Ope/ I can supply other data if you wish this model to be added to the 'supported rowers'. Thanks again for all your help ! -- Chris |
Uh oh!
There was an error while loading. Please reload this page.
Hi
First off I would just like to say what an awesome job you guys are doing here ... a very big well done.
As the title ... I'm a 60-odd retiree with an interest in electro-mechanical stuff, so when my doctor 'suggested' that I get more exercise I immediately looked for something I could tinker with as well as exercise on :)
I bought the Merach R281 as space is limited and I liked the option of being able to use my guest bedroom as a bedroom once in a while. I was aware of the limitations of the monitor before I bought it (thanks RowAlong John) but I had not realised just how important the monitor is until I started using the rower. The Merach monitor really sucks !
The R28R1 has a upper pulley which is where the Merach monitor takes it's feed (via a magnet and two sensors) - handily the pulley has space for two magnets on the opposite side, so with a bit of imagination I was able to mount two hall effect sensors, and two magnets to this pulley. Now I can measure the pulley movement accurately.
Now the bad news. Because the sensors are on the upper pulley, and sense the oar travel in both the forward and backward directions, rather than being on the impeller, I'm having trouble picking out the power stroke vs the return.
Because I have two sensors I can actually discriminate between then by looking at the pulse timing ... if sensor A leads sensor B I'm in the power stroke, but if sensor B leads sensor A I'm in the return, but I don't know what I can do with this information.
If anyone else is interested in what I've done / how I'm doing it / or better still has any ideas how I can take this further please comment on here, and I'll gladly share photos / 3d cad files / purchase info / etc.
Even if I can't actually get this to work with my rower it's still a fun project, and likely to make me more interested in actually doing workouts !
-- Chris
All reactions