Merach R50?

[YAMLcase]

Has anyone had success replacing the useless monitor on a Merach R50? I'm looking to build an ORM for this but wanted to check with others first on their experiences.

[JaapvanEkris]

I haven't encountered the machine in person, but given the strong similarities with a Concept2, it will probably work, In essence ORM will work on any machine, but the amount of DIY varies to make it work. I like to be as little invasive as possible, so I like to re-use existing sensors, but several others also added their own sensors.

When you want to keep the existing sensor in tact: the main question is if they copied Concept2's latest approach where the flywheel also powers the monitor (and thus the wire caries 15 Volt AC) or if it is just a small pulse like the older C2's. I use ORM on a new RowErg and it works well. What kind of connector is on the end of the pick-up wire, and can you attach a volt-meter or something to it?

Once the signals are picked up, it is pretty easy to configure the software. There are a few parameters, but out of the box it should already produce metrics. People typically feel if pace presented is realistic, and change parameters based on that.

We are here to help you if you run into any issues.

[Abasz]

This is just a rebranded version of the "Generic Air Rower". Mine called Zocco Body Fit (https://zocobodyfit.ro/produs/aparat-de-vaslit-zoco-body-fit-air-rower-pliabil-ecran-lcd-eficient-si-util-negru/#). In New Zealand they call it XEBEX V2 Air Rower (https://www.megafit.hu/xebex-ar-2-air-rower-legellenallasu-evezogep-cross-trening-edzeshez-v2-consoleH). We had a discussion on it: #28).

on aliexpress they call it indoor rower :) here is the link

Some "manufacturers" change small things like the performance monitor and they use different sensor. but the Machine is the same. Even the flywheel should have very similar inertia. They are good machines but to have them a very similar experience like on C2 there are 3 things (in addition to the clunky things on them they call "monitors") that needs changing (in the order of importance): 1) handle, original C2 is compatible, 2) shock cord (again original C2 is compatible), seat rollers (but this is debatable on mine it was rubbery that was irritatingly unsmooth compared to C2). The one you posted has the same handle I had on mine before change.

You rollers seems to be a better one and seeming the foot strap is better. 500$ is a good price actually for these.

One thing you need to check whether the reed sensors are checking the rotation of the handle sprocket with two reads or the flywheel rotation (I am 90% its the handle). The good new is that even if its the handle there is a high chance that you can just move one of the reeds facing downwards to face the flywheel and ad a magnet there:

The cheap AliExpress copies (and mine) use the same injection mold for all kind of version so they have the sitting for reeds on the flywheel where their provided reeds fit exactly only they have chosen to measure the nothingness on the handle... :D::D (on the picture I added a hall sensor that is why there you can see 3 pair of wires). The black and white is the reed. they fit into the whole on the front.

[Abasz]

oh man, I could've save another couple hundred bucks AND gotten a better monitor? Well, live and learn I guess.

This monitor is junk :) this is what mine has :):) it measures handle pulls only... it calculates the metrics solely from how many times you pull the handle :):)

So, once stuff is connected, the Generic_Air_Rower profile should work, or at least provide a good starting point.

More or less yes. assuming the flywheel is the same as well :D

[DXCanas]

Has anyone ever cracked either of these open to see if there's an ESP32 inside? 👀

[Abasz]

There is no esp inside. There are two SOCs and one IC:

1. ARM: https://www.nuvoton.com.cn/products/microcontrollers/arm-cortex-m0-mcus/mini51-base-series/mini58zde/
2. Telink TLSR8251 for bluetooth https://www.telink-semi.com/products/bluetooth-le/tlsr825x
3. An LCD driver IC on the other side for driving the screen

You can see that bluetooth is actually added by a ready made dev board :):) The two SOC communicate to each other via UART.

I opened it quite some time ago when I was evaluating whether I could rig the LCD to be used with ESP Rowing Monitor but it became clear quite quickly that this is not going to work :)

[DXCanas]

Finally got around to disassembling the head unit on the R50. Different, but not in any impactful way :)

I am not very well versed in electronic hardware 😅 But this is what the AI overlords say:

---

Here is exactly what is driving your Performance Monitor based on the FCC ID (2AMDBBLE006) stamped on that metal shield:

The Brains: FREQCHIP FR8018HA
That red chip is a BLE006 module manufactured by Chang Yow Technologies. Underneath that EMI shield sits a FREQCHIP FR8018HA System-on-Chip (SoC).

The Specs: It is a bare-bones, ultra-low-power Bluetooth 5.1 SoC. It operates on a simple 1.8V to 3.3V logic level (powered by your AAA batteries) and handles basic GPIO, UART, and I2C inputs.

The Ecosystem: Chang Yow is a massive OEM supplier for fitness displays. This exact same BLE006 module is the hidden brain inside dozens of different Adidas, Reebok, and Cosco treadmills and exercise bikes.

Reading the Main Board

Because the SoC is so simple, the main green PCB (CHANG YOW SI57058-A02) is incredibly straightforward. Looking at the hot-glue potted wire terminals, we can actually map out exactly how the hardware we found in the manual talks to the chip:

• S1 - (Bottom Left): This is the input for the Reed Switch (Sensor 1). Every time the magnet on the flywheel passes the sensor, it shorts this connection, sending a pulse to the SoC to count a revolution.

RS2 / RS3 / RS4 (Bottom Right): A potentiometer uses three wires (VCC, Ground, and Signal). Those route right into this terminal block to pass the analog voltage of your damper setting to the chip.

BUZZER (Center): The basic piezo speaker to give you auditory feedback when you press the plastic console buttons.

---

The leftmost goes to the battery compartment. No mention of the middle 3, but I'm pretty sure that's where HR gets connected.

Don't think this display will be helpful on the Pi 😞

[Abasz]

The 4 pin with the JST connector is probably serial UART connection. The only thing I found puzzling is that they might drive the LCD directly from the SoC :D Normally people use a dedicated IC for this as it takes a lot of IO pins :D

[YAMLcase]

The R50 has a 2-wire sensor with a magnet that looks hot-glued onto the flywheel (I'll probably be fixing that sooner rather than later). From my comparisons I gather the R50 is closest to Concept2's Model C. the D, E and RowErg all have the 3-wire transformer/sensor.

My plan is to get a hold of a PM5 and a Model C replacement pickup to measure resistances and see what a known reading looks like. Hopefully it is exactly like the R50 sensor so I can just build a pigtail from JST SM to 2.5mm mono jack and plug that sucker into the PM5.

R50 tachometer

Funny thing, the monitor has 3 connections going into the support arm but only 2 exiting. One is of course the tachometer and another measures the resistance setting (I'm presuming it's a potentiometer). I'm going to practice patience and resist the urge to open up that arm to see what the heck that 3rd connection is doing.

[klamp169]

The +0.07% refers to the time given a specific distance, as this essentially a function of the dragfactor and the number of pulses. Once we can stabilise that on its optimal value, then we can tweak sprocketsize to adjust drivelength and handle forces.

Ah! That explains it. Since this also does not depend on stroke detection (as long as you stay away from the drive). ORM drag factor corresponds to my rwnp-evaluation to within ~0.05% stdev (exluding some obvious ouliers, like the first draw).

But when you say: "tweak sprocket size" It gives me the shivers. The quarter inch chain will cause havoc on your tweaked sprocket!

[JaapvanEkris]

But when you say: "tweak sprocket size" It gives me the shivers. The quarter inch chain will cause havoc on your tweaked sprocket!

Nope, it is just a parameter in the config to be optimized so that results will match reality best.

[klamp169]

Apologies for being unclear!

This video explains a lot, I think/hope. It depicts a "standard" way of measuring inertia. It is fundamentally more accurate than the "pendulum method". It is a "standard" undergrad physics experiment. Note that in the video, the rotation axis points vertical, wheras the rower flywheel axis points horizontal. This makes it even easier, since no idler pulley is required.

This thread shows the 12 encoder lines on the flywheel of my concept2 model C, causing 12 pulses per revolution. Normally used for determining the angular speed during rowing, but now I use it for determining $\frac{d\omega}{dt}$ to estimate the moment of inertia.

From de the delta-times I calculate $\omega = \frac{2\pi}{12\Delta t}$. Plotting $\omega$ versus time gives slope $\frac{d\omega}{dt}$ as a constant, which can be plugged in in $J=mr^2(\frac{g}{r\dot\omega}-1)$, were $m$ represents the mass attached to the string, $r$ is the radius of the "fan-drum" and g is acceleration due to gravity.

In my case it was particularly/surprisingly easy to carry out. The only thing I had to do is take off the screen from the flywheel housing (instead of just the front cover of my model C) and wind a string with weight around the "fan-drum". And I lifted the flywheel assembly 1 meter up such that I had more "drop-height". I did the experiment with two different weights: 172 [g] and 35 [g]. Results for $J$ were very similar, but I preferred the data obtained with the higher weight. I also had to make one or two tweaks in my analysis software, since during rowing, very low speed (<200 [rpm]) data is discarded.

[Abasz]

Ok I had this totally wrong. I was thinking you are dropping the flywheel... But I understand that you are actually rotating the flywheel with the weight. I get it now thanks :D

Now thinking about it actually this will also be quite easy to replicate for me too as I can just stand the rower basically vertically and use the same hardware like for rowing.

[Abasz]

@klamp169 I moved the inertia measuring alternative to a separate discussion. I did some initial measurments and I have some general questions on the method.

[cagnulein]

@YAMLcase your best shot is bluetooth implementaton as we discussed, i just need the bluetooth name of the device :)

[JaapvanEkris]

Implementing a bluetooth bridge will not fix that his metrics still suck. The monitors of these cheap manufacturers essentially are glorified bike computers. These monitors just produce garbage data as they don't take in account the mechanics of a rowing machine. They deliver some kind of data, but they are too stupid to translate the dynamic flywheel behaviour into meaningful rowing metrics (see this explanation). On EXR we see people rowing 10K's in sprint world records on these machines.

This is actually why OpenRowingMonitor is created: to show it can be done and help people to get decent metrics out of decent hardware.

Aside the quality of metrics, OpenRowingMonitor provides a fully functional PM5 BLE emulator that is tested against ErgZone, KinoMap, EXR (the Zwift for Rowing) etc.. This is much better suited for communicating all rowing metrics than FTMS is, as it communicates state-based on the rowing stroke (i.e. start of drive and recovery), which makes data much more responsive and allows EXR to sync the stroke movement to the app.

[cagnulein]

ok I wasn't aware of the "garbage" issue :)

[JaapvanEkris]

ok I wasn't aware of the "garbage" issue :)

Yeah, where on bikes the calculation of power is relatively simple, it is pretty tough for rowing machines (the flywheel varies a lot in speed throughout the stroke, and you need to calculate the air resistance at the exact right time). Even companies like Peleton and NordicTrack don't get it right, despite the extremely expensive hardware.

[Abasz]

The R50 has a 2-wire sensor with a magnet that looks hot-glued onto the flywheel (I'll probably be fixing that sooner rather than later). From my comparisons I gather the R50 is closest to Concept2's Model C. the D, E and RowErg all have the 3-wire transformer/sensor.

As Jaap said, do not try to fit a pm5 here. It will not work. The inertia is not the same as pm5 (inertia on mine is around 0.07848 while C2 is around 0.1).

Funny thing, the monitor has 3 connections going into the support arm but only 2 exiting. One is of course the tachometer and another measures the resistance setting (I'm presuming it's a potentiometer). I'm going to practice patience and resist the urge to open up that arm to see what the heck that 3rd connection is doing.

Are you sure one of the 3 measures the resistance on the air resistance leaver? Its possible that the 2nd and 3rd wire is for the handle rotation detection, so they can detect whether machine is drive or recovery. This helps with their thing to calculate metrics better. Though looking at the picture, there does not seem to be sensor on the handle sprocket so it is not likely.

Any way, converting this machine to ORM is as simple as it gets:

1. need to connect the tachometer to the right pin on the Rpi
2. load the Generic Air rower profile :)

Generic Air Rower profile is calibrated for one magnet exactly so it should work out of the box.

[YAMLcase]

Ok, curiosity got the best of me and I'm glad it did... what the heck is this thing? It was just kinda jammed in the middle of the arm. I see no model/brand markings, only component labels. the back is blank. grok identified it as a microphone, but this doesn't seem likely unless the lizard people are spying on me lol.

[JaapvanEkris]

The wires are labeled: 3V, GND, and OUT. This suggests it is a powered component, and that it will generate some signal. Weird place to insert it, unless you want it out of the monitor for some reason.

[Abasz]

I vote for lizard people :):)

Honestly they made a good job so no-one can see what this is. the flat is probably some kind of IC. I would even go as far as an MCU considering the 3.3v that is standard for these. The big black thing I though its a electrolyte capacitor but slikscreen A1 and A2 is unconventional marking and the shape is a bit strange. Not to mention that for a microcontroller you dont need such a huge capacitor.

What you can try is what happens if you dont plug this in and run the monitor (I wonder if this is the BLE module and may be they have an NRF chip there. This does not expalin the big component (may be an antenna??? unlikely. also the one output wire is a bit shy for data for such device and even if that it is in would be the right name so the monitor can provide the data. ).

But to be fair this is not necessary for ORM to work.

(honestly I am really intrigued by this design and I hope you decide to find out what it is because it is very interesting :))

Just a quick question. are you sure that one wire goes to the resistance leaver? Because that means that their drag factor is fixed numbers are in a lookup table, but at least they measure the flywheel rotation.

[Abasz]

it says: "can also take my HR from my Garmin watch."

My best bet is that this is infact bluetooth and for connecting to a HR. big chunky thing can be antenna (explains A marking). Also one output singal to communicate the HR value to the monitor is sufficent. Again I would try what happens if it is not connected or, while connected to an HR monitor like a watch or strap and disconnected what happens.

[YAMLcase]

Just to wrap up some of the findings in this thread:

that mystery device in the arm is most likely the HRM sensor. I spent a bit of time researching, crowdsourcing on reddit and asking grok with no conclusive answer but HRM sensor is most likely. definitely not an off-the-shelf part . I don't have a Polar or Garmin to test it out so... for me this quells my curiosity.

the reed sensor pickup on the R50 won't work with the PM5: different kind of sensor is needed. I got the Model C pickup, tore it apart and found that it was a coil. it appears the PM5 measures the magnetic field, not resistance.

The flywheel won't work with the Model C pickup. I got no measurements when I just swapped the pickups. looking more into it I found that the concept2 flywheel has a completely different configuration for the magnets. there are 3 magnets evenly spaced and this just seemed too much trouble to try and figure out how to get working. I ran across a reddit thread where someone had successfully made a retrofit by purchasing a Model D/E pickup, axel, flywheel... well of course why not just get the Concept 2 at that point.

[Abasz]

Thanks for the HR thing, I think that is the most logical answer. Just to complete your research on the way the different C2 pickups:

PM3: it uses resistance to measure the magnet. Basically it has a coil that when the magnet passes changes its resistance, which changes the voltage and this change is measured (via an amplifyer of course) by the PM (https://c2forum.com/viewtopic.php?f=10&t=207567). So here the magnet size is also important as the resistance change is dependent on the amount of magnetic force (this could mean that if you have too big magnet it will not work though this would depend on the amplifier and the comparator inside the circuit of the monitor).

PM5: it has a power generator cicuit (hence it has 12 magnet with alternating direction south/north which when rotating generates power in a coil) and has 3 sleeves. one is ground, one is the power "charger" that generates power for the PM so the battery can last longer when you row, and 3rd is the signal which is relatively high voltages (I measured 18-20v) in a sinoid wave. (https://c2forum.com/viewtopic.php?f=10&t=207567)

Also as a side note, my copy rower flywheel inertia is around 0.078 while the C2 is around 0.1 ish. Since the PMs bake their values in they will not work relyably with other flywheel. Actually PM5 can change the erg type (https://www.youtube.com/watch?v=hcqCwNniArI) because it is retrofitable to almost any model (even Model B and based on their web site Model A too?). In such case I suspect it loads the constants (number of impulses and all that) and also "switchis" to the right signal detection mode.

[DXCanas]

For the record: rowed with an HRM strap. Disconnected that harness. HR no longer displayed.

So the 3 harnesses at the factory PM are: HR, resistance lever, (reed?) sensor.

[DXCanas]

Bringing this conversation up as I am a new r50 owner and am very interested in getting ORM working, particularly the ANT+ -> Garmin component. Exciting!

This comment inspired hope:

Any way, converting this machine to ORM is as simple as it gets:
need to connect the tachometer to the right pin on the Rpi
load the Generic Air rower profile :)

#140 (comment)

"out of the box" specifically was really exciting.

I've got a pi3 laying around, I think. And an ANT dongle. I've never been much of a hardware guy apart from building basic desktops 15yrs ago; once I identify the tachometer pin, then the rPi pin… well. At risk of looking really stupid: how do I make them touch?

Update: Read the docs! Ordered more Dupont Cables than I will ever need in this lifetime!

---

Also @Abasz :

Just a quick question. are you sure that one wire goes to the resistance leaver? Because that means that their drag factor is fixed numbers are in a lookup table, but at least they measure the flywheel rotation.

I haven't popped the thing open myself, but all evidence points to "yes" — the manual has a section dedicated to calibrating the resistance lever arm, and some light teardowns on youtube show the cable running to the lever.

Is… is that a helpful sensor to have for ORM? or just like "hey, at least they tried!"

also, btw @JaapvanEkris IRT:

On EXR we see people rowing 10K's in sprint world records on these machines.

I think these guys are blatantly cheating (putting a towel over the fan). In my and I think most reviewers' experience, this thing undercuts you a fair amount.

Some sources:

• reviewer demonstrating the cheat/miscalibrated lever
• reviewer quickly popping the thing open, showing cables running to lever
• UK manual, VR Calibration Mode on page 25

[JaapvanEkris]

Bringing this conversation up as I am a new r50 owner and am very interested in getting ORM working, particularly the ANT+ -> Garmin component. Exciting!

Welcome. We'll get it to work, no worries.

This comment inspired hope:

Any way, converting this machine to ORM is as simple as it gets:
need to connect the tachometer to the right pin on the Rpi
load the Generic Air rower profile :)

#140 (comment)

"out of the box" specifically was really exciting.

I've got a pi3 laying around, I think. And an ANT dongle. I've never been much of a hardware guy apart from building basic desktops 15yrs ago; once I identify the tachometer pin, then the rPi pin… well. At risk of looking really stupid: how do I make them touch?

Update: Read the docs! Ordered more Dupont Cables than I will ever need in this lifetime!

Know that feeling. I got a whole bunch of them as well :)

Also @Abasz :

Just a quick question. are you sure that one wire goes to the resistance leaver? Because that means that their drag factor is fixed numbers are in a lookup table, but at least they measure the flywheel rotation.

I haven't popped the thing open myself, but all evidence points to "yes" — the manual has a section dedicated to calibrating the resistance lever arm, and some light teardowns on youtube show the cable running to the lever.

Is… is that a helpful sensor to have for ORM? or just like "hey, at least they tried!"

It is somewhere between "hey, at least they tried!", "how stupid can you get" and "Why not complete what you are doing?". Drag (the actual resistance, not the lever) can change due to dust collection, air pressure, etc.. So knowing where the lever is might be helpfull, but to real accurate stuff, you need to measure how the flywheel behaves in a session. The key math is actually pretty straightforward. It becomes complex if you want it to be precise and robust. The added complexity of a sensor and calibration procedure is probably more work than get the math right.

also, btw @JaapvanEkris IRT:

On EXR we see people rowing 10K's in sprint world records on these machines.

I think these guys are blatantly cheating (putting a towel over the fan). In my and I think most reviewers' experience, this thing undercuts you a fair amount.

There are some machines who just report insane power production. Even the WaterRower is 10% too optimistic, but other machines are much much worse. I know the brands, I've tested them. It is just that the engineers got the job to "produce numbers", instead of "produce credible metrics".

Some sources:

• reviewer demonstrating the cheat/miscalibrated lever
• reviewer quickly popping the thing open, showing cables running to lever
• UK manual, VR Calibration Mode on page 25

Thanks.

[Skodavrs007]

Hello everyone first of all I want to thanks for such great project.

Merach R50 it´s not same like other chinese copies of concept row machine you follow.
This is high quality clone with better seat, better foot straps, better chain tensioner and more stable than Concept row erg.

I read lot of tests and deffietively decide to buy this Merach r50 than Concept row erg.

Only originál display is not good readable and not show correct values.
This was sloved with OMR now its great machine.
Configuration of ORM is same like for Concept D only sprocket rádius is 1,3 and only 1 magnet all other could be same.

[JaapvanEkris]

Merach R50 it´s not same like other chinese copies of concept row machine you follow. This is high quality clone with better seat, better foot straps, better chain tensioner and more stable than Concept row erg.

I read lot of tests and deffietively decide to buy this Merach r50 than Concept row erg.

Only originál display is not good readable and not show correct values. This was sloved with OMR now its great machine. Configuration of ORM is same like for Concept D only sprocket rádius is 1,3 and only 1 magnet all other could be same.

Good to hear! Would you be willing to share your entire config and the raw recording of 20 or more srrokes? This way I can include it in the standard configurations, and keep it consistent when there are changes to the engine.

[Skodavrs007]

Yes i will send it to you.

[DXCanas]

@Skodavrs007 did you hardwire a USB battery into the 5V inputs on the board? Looks neat! I have been thinking about how I might be able to keep my machine free from wall outlets

[Skodavrs007]

@Skodavrs007 did you hardwire a USB battery into the 5V inputs on the board? Looks neat! I have been thinking about how I might be able to keep my machine free from wall outlets

I use this hat with regulator its safe solution.
direct conecction on pins its not safe.

https://shop.pimoroni.com/products/wide-input-shim?variant=2168104321034

Its still not finished it means final back cover will have switch connected to this hat to cut off battery. I will share photos when will be finished.

[aschmid]

@Skodavrs007 your setup looks awesome. i recently got an R50 myself and im about to set it up with ORM.
Are you saying you used the Generic_Air_Rower profile and set sprocketRadius: 1.3 as the only profile customization?
numOfImpulsesPerRevolution is already set to 1 for that profile. unless you are talking about the Concept2_Model_C or Concept2_RowErg profile which would need both params to be updated.

Also i hope you didnt have to make any other hardware updates on the rower itself as i do not feel very comfortable with that part :)

thanks guys for this awesome project @JaapvanEkris and friends!

[Skodavrs007]

When you guys agree that the set config is final, please let me know, and provide a sample (at least 20 strokes, preferably more) so that I can keep the settings working when engine updates change its behaviour. Please note, the new 0.9.7 version of ORM will change the engine, and will allow for systematic errors (magnet placement issues) to be corrected automatically.

Yes its final config for original apartment but I am not able to provide sample because I upgrade sensor for Hall and add 3 magnets it means I have now different config.

I have one guestion. I have now problém that some time my ORM is reset after 15 minutes of rowing. Stop communication between Cozweed and OMR. On the screen its not possible to start again HR and Concept connection but in diplay of OMR count correctly but from zero. Is it possible to find any log to see what happens?

[JaapvanEkris]

I have one guestion. I have now problém that some time my ORM is reset after 15 minutes of rowing. Stop communication between Cozweed and OMR. On the screen its not possible to start again HR and Concept connection but in diplay of OMR count correctly but from zero. Is it possible to find any log to see what happens?

That is a new one. I use ORM on EXR daily with quite a similar configuration and this doesn't happen. And it shouldn't happen. OpenRowingMonitor logs to syslog (accessible via journalctl | grep npm), that should contain the logging of the crash. Can you grab the data around the crash?

[aschmid]

ill do my best to provide a profile and a sample as soon as ill be fully set up. but that may take me a bit unfortunately

[klamp169]

You dont need to disassemble anything. Set 1.3 and check value on screen drive lenght during the Rowingdata. Than check real lengh on chain. Easiest way to check it is to mark on chain by tape starting point of your row phase and with second person measure chain lengh if you are in the end of rowing phase.

For chain-driven flywheels like those used in Concept2 and Merach R50 rowers the equation for the sprocket radius $r_s$ is:

$$r_s = \frac{L_c N_s}{2\pi}$$

where $L_c$ equals the chain pitch (1/4 inch for Concept2 rowers) and $N_s$ equals the number of gear teeth (usually 14) on the sprocket. This leads to $r_s$ = 1.415 [cm] for Concept2 rowers.

Now someone has to determine both quantities $L_c$ and $N_s$ for a definitive value for $r_s$ for the Merach rower.

[DXCanas]

For posterity, the sprocket radius is pretty much confirmed. Through some manual checking but also with manufacturer schematics:

#140 (reply in thread)

[aschmid]

ok guys i think i have something to work with. The numbers seem plausible but i wanted to see what you think of the settings and the data it produced.

  rowerSettings: {
    numOfImpulsesPerRevolution: 1,
    sprocketRadius: 1.4,
    minimumTimeBetweenImpulses: 0.007,
    // maximumTimeBetweenImpulses: 0.0145,
    // smoothing: 1,
    flankLength: 3,
    minimumForceBeforeStroke: 1.5,
    minimumRecoverySlope: 0,
    minimumStrokeQuality: 0.4,
    // autoAdjustRecoverySlope: false,
    autoAdjustRecoverySlopeMargin: 0.05,
    minimumDriveTime: 0.2,
    minimumRecoveryTime: 0.6,
    dragFactor: 110,
    autoAdjustDragFactor: true,
    dragFactorSmoothing: 1,
    minimumDragQuality: 0.8,
    flywheelInertia: 0.10148,
    maximumStrokeTimeBeforePause: 6.0,
    magicConstant: 2.8,
  }

uncommenting maximumTimeBetweenImpulses, smoothing and autoAdjustRecoverySlope will cause issues with the data. i guess the value for maximumTimeBetweenImpulses may be too restrictive but not sure about what the issue is with the other 2.

2026-01-17_13-45-15_rowingData.csv
2026-01-17_13-45-15_raw.csv

i also tried to get it to work with a sprocketRadius at 1.3 but i feel like the data seems to make more sense with the settings above.
just for good measure ill give you what it produced with that config:

rowerSettings: {
    numOfImpulsesPerRevolution: 1,
    sprocketRadius: 1.3,  
    minimumTimeBetweenImpulses: 0.007,
    flankLength: 3,
    minimumForceBeforeStroke: 1.3, 
    minimumRecoverySlope: 0,
    minimumStrokeQuality: 0.35, 
    autoAdjustRecoverySlopeMargin: 0.05,
    minimumDriveTime: 0.2,
    minimumRecoveryTime: 0.6,
    dragFactor: 110,
    autoAdjustDragFactor: true,
    dragFactorSmoothing: 1,
    minimumDragQuality: 0.75, 
    flywheelInertia: 0.10148,
    maximumStrokeTimeBeforePause: 6.0,
    magicConstant: 2.8,
  }

and the data here:

2026-01-17_13-48-26_raw.csv
2026-01-17_13-48-26_rowingData.csv

Thanks!

[aschmid]

@JaapvanEkris i tested adding smoothing and autoAdjustRecoverySlope and replay the raw data and it does work now.

[JaapvanEkris]

I saw this:

    minimumTimeBetweenImpulses: 0.007,
    // maximumTimeBetweenImpulses: 0.0145,

And looking at the data you provided, I would expect this:

    minimumTimeBetweenImpulses: 0.04,
    maximumTimeBetweenImpulses: 0.07,

This works well with the given raw files.

[aschmid]

ok makes sense, ill try that out.
im still learning what every setting means right now and i could have saved a lot of calories and time if i would have known that i could have just replayed the raw data instead of rowing for a while every time i changed settings on ORM and test it out again. :)

[JaapvanEkris]

ok makes sense, ill try that out. im still learning what every setting means right now

You could take a look at https://github.com/JaapvanEkris/openrowingmonitor/blob/main/docs/rower_settings.md , which explains it all

and i could have saved a lot of calories and time if i would have known that i could have just replayed the raw data instead of rowing for a while every time i changed settings on ORM and test it out again. :)

Yeah, but you bought the machine to get fit and healthy, and this helps :)

[aschmid]

i actually did look at the rower settings docs but its a lot of info and i didnt have the patience to dig into every single one of them :)
and yes i probably rowed a good 3-5k to get that config done. haha

[aschmid]

ok my best and final.
this feels the best and shows a more or less correct drive length based on what i measured.

rowerSettings: {
    numOfImpulsesPerRevolution: 1,
    sprocketRadius: 1.365,
    minimumTimeBetweenImpulses: 0.007,
    flankLength: 3,
    minimumForceBeforeStroke: 1.3,
    minimumRecoverySlope: 0,
    minimumStrokeQuality: 0.35,
    autoAdjustRecoverySlopeMargin: 0.05,
    minimumDriveTime: 0.2,
    minimumRecoveryTime: 0.6,
    dragFactor: 105,
    autoAdjustDragFactor: true,
    dragFactorSmoothing: 1,
    minimumDragQuality: 0.78,
    flywheelInertia: 0.10148,
    maximumStrokeTimeBeforePause: 6.0,
    magicConstant: 2.8,
}

ill use this one unless one of you guys gives me inputs to improve it.

[cgiacofei]

So, I had no luck with this profile. ORM didn't detect any strokes for me. I ended up using the generic profile which had worked for me as far as detecting strokes and subbed in the sprocketRadius, dragFactor, and flywheelInertia
from your custom profile. (just realized I forgot about the magicConstant)

I'm a total noob who's been rowing for all of 4 days, but the results at least seem within the bounds of reason to me.

[JaapvanEkris]

So, I had no luck with this profile. ORM didn't detect any strokes for me. I ended up using the generic profile which had worked for me as far as detecting strokes and subbed in the sprocketRadius, dragFactor, and flywheelInertia from your custom profile.

Interesting. Could you provide a sample raw file for me, so I can have a look?

(just realized I forgot about the magicConstant)

Unsless you use OpenRowingMonitor with a kayak simulator, 2.8 is the correct setting. Some apps explicitly check this parameter (for example EXR) and reject data if it is set differently.

I'm a total noob who's been rowing for all of 4 days, but the results at least seem within the bounds of reason to me.

Yeah, it takes quite some effort to get it close to perfection. In essence you need someone who knows his pace in a boat or a C2 very well, or someone who is willing to row on a C2 in the gym and compare his feel and pace to his own machine. Not the best process, and as long as results are reasonable, nobody will really care.

[aschmid]

interesting for sure.
if u copy/paste the settings above it doesn't work at all?
would be curious to see the raw data to see if maybe its a different build?!

[JaapvanEkris]

would be curious to see the raw data to see if maybe its a different build?!

The physics engine hasn't changed between 0.9.5 and 0.9.6. With 0.9.7 we do change the engine, but as the Merarch only has a single magnet, it shouldn't be affected by it (we address structural noise due to magnet placement errors).

[DXCanas]

Throwing my hat in! Figured some input from just the generic rower settings would be helpful.

I wouldn't call myself experienced, but I've put some time into a few C2 RowErg/Model D's, as well as some Hydrow machines. The power and m/s reading didn't seem feel very far from what I'd expect 🤷

My settings, specifically:

   rowerSettings: Object.assign({}, rowerProfiles.Generic_Air_Rower, {
    // Uncomment these if you want to force the other physics values from the forum too:
    // sprocketRadius: 1.3,
    // If ever add magnets
    // numOfImpulsesPerRevolution: 1,
  }),

So. "Stock".

2026-01-26_04-13-23_raw.csv.gz
2026-01-26_04-13-23_rowingData.csv

[DXCanas]

Please look at the profiles for 0.9.7, where for the model C this has been increased to 50, and on the RowErg to 22.

Will take a look! Although

Could be that your machine behaves better and thus requires less minimumForceBeforeStroke to kill noise.

Yeah, there isn't much/any noise as far as I can tell. But there's also only like 15 datapoints per recovery at damper 10 😅 as noted by the red bits in table on my screencap.

---

On another note:

// somewhere between 0.097and .104 for SPI/HR efficiency; going with Model C's

I was extremely iffy about going as high as .104 or even .105 as suggested by avgPwr/avgHR or NP/avgHR until it occured to me to shine a light into the damper. Concept2's tolerances/engineering are pretty clear; the Merach lets in a lot more air at damper 1 and across the whole range:

I don't have a C2 on hand, but going off of a youtuber, the gaps look even wider than damper 3ish on a C2. Checks out.

Been weird balancing out "wow I suck today" with "wow those numbers look amazing!"

[Abasz]

Please also keep in mind that its not just the intake but the exhaust is also relevant. Actually on my chineese clone the tiny holes on the metal cover were smaler (or not so "dense") than on C2 and I was not able to get as high drag factor (and feeling) as on C2. Then I swapped out one from our boat house for testing and it was almost instantly fixed. So I after swapping it back I accepted that I will need different leaver for home :D

[DXCanas]

Yeah!! Funny you mention I was thinking I’ll rig something to direct the exhaust and lower the drag a bit at the same time.

[DXCanas]

Alright, I have done a handful of workouts, steady state and sprinting, with this config. The data feels accurate, and the HR efficiency numbers align basically perfect across (admittedly more limited) C2 sessions.

If there's no glaring issues, this is what I would submit for PR.

  rowerSettings: {
    numOfImpulsesPerRevolution: 1,
    // Worried about low-damper sprints
    // Lowest in steady state damper1 was >.04
    minimumTimeBetweenImpulses: 0.03,
    // Highest in steady state damper10 was <.075
    maximumTimeBetweenImpulses: 0.08,

    sprocketRadius: 1.43,
    // Tricky. 1 datapoint per rotation.
    flankLength: 4,
    minimumStrokeQuality: 0.3,
    
    // Not a ton of data points (16) on decel for high damper/slower wheel
    minimumRecoverySlope: 0.0,
    // New C2 Standard
    minimumForceBeforeStroke: 25,
    autoAdjustRecoverySlope: true,

    // Perf holes seems similar to C2 but leaky damper?
    flywheelInertia: 0.105,

    // Default, lowest value seen on Damper 1
    dragFactor: 107,

    // Safe defaults
    minimumDriveTime: 0.50,
    minimumRecoveryTime: 0.90,

    // Allow down to 15 SPM
    maximumStrokeTimeBeforePause: 6.0,
    autoAdjustDragFactor: true,
    minimumDragQuality: 0.85,
    dragFactorSmoothing: 3,

    systematicErrorAgressiveness: 0,
    systematicErrorNumberOfDatapoints: 0,

    autoAdjustRecoverySlopeMargin: 0.05,
    magicConstant: 2.8
  },

Steady state 10k raw data
Steady state 10k rowing data
Steady state 6k raw data
Steady state 6k RowingData

Average cadence of 21 for both, with some faster bits. 2ish min/500m on the 10k and 10ish s slower per 500m on the 6k

Drive length should be in 1.3ish-1.4ishm (I'm just under 183cm tall), as measured by a friend (55ish in).

DF is a little lower on the 6k: I think it was probably a slightly lower damper, but I also positioned the Pi to be on top of the perf cage, and that might have effected it somewhat.

Goodness of fit was consistently pretty high, went with a flank length of 4 just to give the curve more shape. Data was pretty pristine — didn't see much noise apart from the expected wind up/down to/from cool. Pause behaviour seems snappy but forgiving enough for a mid-row sneeze or quick screentap.

Still a bit worried about this, from earlier:

Tempted to get more range in TimeBetweenImpulses for Super-Slow-Cadence + High-Damper or Sprints + Low-Damper. Maybe like .025-.12 or so? Though the pause feature rn is rather luxurious compared to the GenericRower.

Is this worth thinking about? @JaapvanEkris

Apart from that — if there's no glaring issues — I would call this my final draft.

[aschmid]

i tested this config and it feels very solid! voting for it

[aschmid]

Took me a bit but wanted to share my R50 setup :)

[JaapvanEkris]

Cool! Looks extremely nice

[DXCanas]

Nice!! Is that head unit actually thin enough to fold back without scraping against the cage? Unlike the stock one? 😅

[aschmid]

@DXCanas haha. yes it folds back, there is maybe 5mm of space between the cage and the monitor. my stock one was able to fold back also but it wasnt totally straight when up unless i tilted it a bit by force :)

[aschmid]

not 100% cleared by the cage when folded down but good enough