PM5 obsolete?

[Carl Watts]

I almost get the feeling that some would be happy if the PM5 were a 'black box' hidden away, and you then plop in your phone/tablet and connect ergdata via bluetooth, then everything is magically ok as the main screen is now a device lol.

For me it's more a case of "wouldn't it be better to do less on the PM5 and more on a user's device " ? (and yes, it's a question, answer may be "no, it's not"). The power you have on today's user devices is insane compared to not long ago so maybe it's worth revisiting the decision.

But what JaapvanE said (about the very specific hw you need to process the high-freq sensor data with low latency) may end the whole discussion.
Or the fact (already pointed out by others) that c2 focus may be less about home individual users and more about boathouses and gyms. But yeah, in that case, it's normal to have home users "complaining".

The data transfer from a smart black box to the phone is not rocket science, just put a small micro in there to do the job these days. Hell just use the current PM5 micro, its a tiny ARM based one and low power.

The PM1 back in 1986 could do the math, it was like a 4 Bit NEC micro. Any small micro can take that single channel data stream that is incredibly slow by micro standards and sent the data via Bluetooth to your phone.

The black box just sits where the current black tach resides, it needs a couple of LED's for power green and connected blue and a tactile button to turn it on from sleep or hold it down for a reset.

You don't want multiple devices sitting on the monitor arm when your phone or better still tablet can do it all now.

[Sakly]

But what JaapvanE said (about the very specific hw you need to process the high-freq sensor data with low latency) may end the whole discussion.

I don't think hardware and their resources are the main problem. Surely it would be possible to buffer and transfer the sensor data over USB to any device. The interface is absolutely fast enough to do that (wireless obviously not, at least for real-time capabilities)

In embedded software development we typically distinguish between "General application OS" (i.e. Android/Linux, iOS/BSD, Windows) and Real Time Operating Systems (RTOS). A general application OS is good at picking up a big task and completing it, then picking up the next big task. If it is doing its big task, a next task must wait in line before it is even considered. In normal life, this switching is so fast, as a user you will not notice it. An anology: a cashier in your supermarket. Typically you will get there, but response time is in no way guaranteed as it is highly dependent on the traffic in front of you.

In embedded systems we work with Real Time Operating Systems, or Near Real Time Operating Systems (i.e. Linux with a highly modified/PREEMPT kernel). These are not designed to handle big heavy tasks. In fact, as soon as a higher priority task comes in, it will drop everything it is doing and pick up that higher priority task. An anology: a medic in an emergency room. The response time for high priority tasks (latency in processing) is almost guaranteed, but lower priority taks might suffer some starvation. The latter is the reason why it isn't generally useable for consumer products yet.

Both have their place: for general applications a normal OS with sufficient CPU and memory suffices. However, for processing high frequency signals, this isn't acceptable as you need to process the data devices) omes (as you need to tell apps like EXR a stroke has finished for example). And let's not start about finish times being only close to the real time.

I don't know how C2 made the current hardware, but as I read their schematics on the FCC website, they have parts of dedicated chips reserved for processing the tach-signal. Looking at a software based design, it is really tough to guarantee a decent response time overall. That is why ORM has a three-stage design combined with some OS-level tricks:

• we have an extreme high frequency part, which measures the timing with a specific flank in nanosecond precission. You need this precission as it is needed for getting detailed info about the angular velocity and acceleration (between successive measurements you don't get as much difference in times, so these small measurements matter). This part is extremely small, runs extremely close to kernel level, has a dedicated high performance CPU core all by itself and has a huge priority (in practice, only the kernel clock is more important).
• On a C2, this flank passes by every 5 to 10 miliseconds, which is reported to the second stage. This second stage thus needs to process this data as soon as possible and recalculate all dependent metrics. Please note, as you need stroke detection for drag calculation, all metrics currently shown by a PM5/ErgData are direct byproducts of these calculations. Typically this also runs at its own dedicated CPU core which is set to high performance at a near-kernel level priority (just below the stage 1).
• The third stage are the consumers. ORM has a webserver and client on board, which usually gets updated every 80 ms, but the client can be offloaded to other hardware (as a browser is a pretty CPU intensive thing to run, especially on older RPi's). ANT+ will broadcast a position report every 400ms, which isn't that time-critical. However, the PM5 BLE interface will instantly broadcast as soon a stroke ends. This is essential in the user experience as several clients sync the visualised stroke on screen.

As said, offloading the webclient can be done easily. Offloading the bluetooth part is tough.

Initially we experimented with an Arduino for stage 1 (as it is interrupt-driven, you need to do less tricks to get it measuring with enough accuracy). It can't handle the calculations. So we offloaded stage 2 and 3 on a Raspberry Pi, and put a USB cable in between. It wasn't succesful as USB is designed to process large data volumes in bulk, not small messages with extreme low latency. So, we ended up killing that avenue as it wasn't stable. Currently there are integrated industrial boards combining the two, but these aren't generally available to a wide audience.

The main problem would be the compatibility of the external device, which is not under control and to ensure proper function with thousands of different devices.

There is another part here indeed. In development of ORM we see a lot of off-brand machines. To get it working on a random machine, you need quite some data about the mechanics of the machine, like number of pulses per rotation, flywheel inertia, sprocket size, minimum and maximum flywheel speed, etc.. You need this engineering data to get stroke detection and drag calculation working. This typically is contained in the PM (we have been working for over 2,5 years to get the optimal data for a C2...). Although not extremely secret, it opens doors that IMHO should remained closed.

What I mean is that it opens the door to "weightdoping" like scenario's, where people are adjusting parameters (selecting a different device than they actually use) to suit there ego/needs. In cycling these have been already seen. This is why ORM simulates certain PM5 messages, but deliberatly lacks others (aside not being able to generate a verification code), as it could be used to cheat as all parameters can be adjusted by the user.

Another factor is: to go for big fleets or competitions, were your need machines without external devices, this would cause the need to have a device like PM5 with the same capabilities (at least for sensor processing and main metrics) anyway. So development would increase without any benefit but probably loss in money due to lesser HW equipped machines. Or the money must be compensated by license costs of the external app running on the external device.

This is where I think rowing has a huge advantage over cycling. In cycling, you have small devices (power meters, etc.) who spit out simple measurements. But there is no way to verify what the overall result of the effort is, as it lacks context. Indoor cycle races now require two independent power meters to be in constant agreement to be able to compete. Still people were able to cheat, as nothing you can see can be trusted, as it all runs on hardware that is not closed off to the user.

The PM5 is designed to be stand-alone, and it can be told to control the entire race for that specific machine. It controls every movement, and it can sign the result with a verification code. Aside physical tricks on the PM5, that is a huge step beyond the simple metrics these simple cycling sensors provide, as the results can be trusted. Network issues and all other things might fail on you, but as long as a PM5 is given the right race parameters and comes back with a signed result, it is almost foolproof.

Thx for your explanations I'm mostly aware of coming from a background of embedded programming of ECUs
I was talking about a cable connected solution rowErg sensor/blackbock to any device and not talking about connection of different rower devices to the "new thing, that does all the math and display stuff". So speed wouldn't be an issue, but - as you say - speed for data transmission is not the only factor, for some data the point of time, when calculations are done is important and that has hard impact on the complete system design. If you cannot control the system design (=> thousands of different external devices), you won't get a stable solution.

[jamesg]

The C2 model sees what the flywheel does: the results of our effort, not the effort itself.

If we wish to see the results, C2 PM is satisfactory, since it mimics rowing, where we see when the bowball crosses the line.

Any other analytical system will be of no practical use if it only shows us what we already know.

So what do we want to know that we don't know now?

[JaapvanE]

Thx for your explanations I'm mostly aware of coming from a background of embedded programming of ECUs

I am in good company then

If you cannot control the system design (=> thousands of different external devices), you won't get a stable solution.

True. And you see all kind of weird stuff, where you essentially are fighting battery management of every specific phone/tablet to keep a process running well enough to function. Any sane company would avoid that swamp.

And in all honesty, I really like the abstraction level of the PM5 BLE interface. It will tell you strokes, and all there associated data like pace, etc.. It provides a well considered abstraction level were all machine-like issues (like having a Model B with 3 magnets, or a RowErg with 6 magnets) are left to the machine manufacturer and the monitor, while providing more than enough data for running other apps to run on top of that without the fuss of recognizing hardware and their underlying particularities.

If we wish to see the results, C2 PM is satisfactory

I'd agree that a PM5 is brilliant as a stand alone device in a gym/boathouse where people simply hop on an erg to warm up. But for serious training, you already need tools like ErgData as you need more data and want to record stuff to your logbook.

Any other analytical system will be of no practical use if it only shows us what we already know.

The PM5 screen is limited in what it shows, and lacks certain flexibility. There is a lot to say for a a better presentation of data. For example, I want my force curve and all related data on the PM5. I can show the force curve on the PM5, but I need ErgData next to it to show me drivetime, drivelength and peak force. In all honesty, I absolutely love RP3's force curve: https://www.youtube.com/watch?v=0HJzHe5zji4
It is brilliant in its simple setup, and their shadow mode helps improve stroke to stroke consistency, and their reference curve is simply brilliant (yea, it is on my wishlist for ORM).

So what do we want to know that we don't know now?

Do you got a couple of hours? Key data like drivetime and recoverytime (or even better Ratio...) are standard on a coxbox, but the Pm5 can't display it. Nor can ErgData. Distance per stroke is still missing in action both on the PM5 and ErgData, despite it being a pretty good indicator of fatigue setting in. As said, the current forcecurve is useable, but as RP3 has shown, it can be done in a much better way. ErgData shows what the peak force is, but doesn't tell where it is in the stroke (thus is person's stroke frontloaded or backloaded). That kind of data is quite important for boat selection and crew selection (you don't want to mix frontloaded rowers with backloaded).

Beyond that, being able to record all these metrics is essential. So you might see a metric, but is gone for all eternity with the next stroke.

[edward.jamer]

Great discussion.

As a home user looking to buy an indoor rowing machine, my top criteria was:
* strong and accurate analytics reporting
* rugged design with low/easy maintenance; bonus points if the company has a strong reputation for customer support
* no running subscription fees
* no requirement to be plugged in; bonus points if the rower is easy to transport

There have been a few amazing new pieces of erging hardware released in the last decade, but (after doing some light online research) C2 won my money and I couldn't be happier with the result. I'm a light rower compared to many here, but the past few years have seen hundreds of hours and millions of meters spent on my RowErg - including work outs during power outages and in different locations.

PM5 may not be beautiful, but it's extremely practical and functional. Calling it obsolete seems like a reach... although I can understand why people want better visualisations and additional software features. Some other rowers are offering large screens with gorgeous interfaces and lots of custom features; they target a different user segment though, and don't fit my personal targets as well as a simple C2 RowErg.

For me, the ErgData app improvements have gone a long way to providing a smoother user interface. I'd love to see the C2 logbook get a refresh eventuall, allowing a deeper dive into data from an individual session, stronger session planning capabilities, and dynamic aggregate reporting tools. Improvements can be made. My take is that those improvement belong mostly on the software side, however, and not with the PM5 hardware itself. The question, of course, if whether the development cost is actually worthwhile for C2... and that's not so easy for me to know.


Others will disagree. That's fine! It's not like there is a single correct answer here.

However, it might be worth noting that a search for "best rowing machine" online once again shows a near consensus opinion that C2 RowErg remains at the very top of the list, despite the market being flooded with knock-off designs and feature-rich competitors over the past few years. There is some pressure to evolve and compete, but for my money the RowErg is still first-in-class without doubt.

[stevegaspars]

Key data like drivetime and recoverytime (or even better Ratio...) are standard on a coxbox, but the Pm5 can't display it. Nor can ErgData. Distance per stroke is still missing in action both on the PM5 and ErgData, despite it being a pretty good indicator of fatigue setting in.

ErgData shows drive time, drive ratio and distance per stroke. You just need to enable the metrics in the customisable displays.

[JaapvanE]

ErgData shows drive time, drive ratio and distance per stroke. You just need to enable the metrics in the customisable displays.

Cool, I must have missed that one! Did it come with a recent update (when c2 moved to the new app, I've looked averywhere, as these are quite relevant indicators for stroke quuality.

Edit: are you sure about Distance per stroke, as I still can't find that one..

[gvcormac]

I resisted for some time but can't hold back the following curmudgeon comment.

I like the Concept 2 just fine the way it is. I don't want to buy a television with my rower, and I certainly don't want to pay for a subscription. I don't want a touchscreen anything.

I will keep my 2016 car with tactile controls as long as I can hoping that the world will some day reject touchpad iThings in favor of actual gauges and knobs. I like my Garmin Forerunner 45 way better than my smartwatch. And I really wish I could buy a drink at the airport without having to navigate an iThing.

And I really wish I had my Blackberry Bold back. Or rather, that it still worked and played with modern networks and email providers.

Maybe C2 will go the way of Blackberry if they don't jump on the immersive bandwagon. But that strategy didn't do much for Blackberry, did it? Fortunately, there are millions of existing C2s out there, and they're indestructible ...

[Ombrax]

I like the Concept 2 just fine the way it is. I don't want to buy a television with my rower, and I certainly don't want to pay for a subscription. I don't want a touchscreen anything.

+1. I agree 100 percent

But, the next question is what's the best strategy for C2 long term?

We all want what's best for them too.

[JaapvanE]

I like the Concept 2 just fine the way it is. I don't want to buy a television with my rower, and I certainly don't want to pay for a subscription. I don't want a touchscreen anything.

Completely agree with the sentiment that you don't want to be forced to buy a subscription. IMHO, that is where Peleton's machines are useless: you pay a small fortune to buy a machine, that requires a monthly subscription to be functional. That doesn't go down well.

I like my Garmin Forerunner 45 way better than my smartwatch.

Funny you mention that. I owned a Garmin Venue, which was mainly touchscreen driven. I moved to the Garmin Epix a year ago, and started to love the physical buttons so much I nearly turned off the touchscreen all together. I sooke to some friends considering the same. You certainly are not alone on that one.

Maybe C2 will go the way of Blackberry if they don't jump on the immersive bandwagon. But that strategy didn't do much for Blackberry, did it?

BlackBerry had the issue that its main focus was business customers, and these customers liked the flexibility of the touchscreens more, and they were run over by Apple's marketing. Microsoft had the same issue, despite having touchscreens years before Apple did. It wasn't the lack of a touchscreen per se that killed BlackBerry, it was the brilliant marketing of Apple and an overreliance on a pure technical more advanced product by its competitors.

But, the next question is what's the best strategy for C2 long term?

As a company they shouldn't ignore the competition and respond. Companies like RP3 are making waves, also in the boathouses. Product looks more appealing, and their approach is quite interesting. You can slap on a simple Kindle Fire and it's working. Won't work in a clubhouse or gym, but for national teams were people do respect expensive equipment it will. Having a touchscreen in some way can open interaction scenarios where buttons simply don't work. As an example, the setting of a goal force curve by RP3. This doesn't imply the PM5 should have a touch screen, but it also could mean that phone/tablet integration should improve. Improving screen resolution and colors might introduce other interaction scenarios as well. You see on the ErgData app that it helps me stay in my training zone. This is helpful, but why can't the PM5 alert me as well when I leave my zone?

RP3's analytics portal is way better than C2's. As rowing is a metrics driven sport, they have some work to do there. Issue is that RP3 even outperforms RowsAndAll at the moment, so you can't even get by by using painsled for example.

Some integration with tools like trainingpeaks is expected. You want a schedule to be ready (Pete's plan?) and just import it into the PM5 directly or via the ErgData app. ErgZone has similar features, so why hasn't C2 accepted it needs to do something beyond being a machine for a single session.

Gamification of training is truly a thing for home rowers (see this video from team GB were some use EXR to train from home: https://youtu.be/rt_xLjyH8qM ), and C2 should embrace it. Cycling has launched quite a succesful competition in Zwift. World of Rowing is determined to make this happen as well and it is only the question wether it will happen on EXR or RowCave (as EXR is more accessible to consumers, my money is on them). C2 is in a pretty good position here, but looking at scenarios were people do not want a simple monitor but do want an immersive experience, you need some improvements. It could be as simple as making sure the monitor can be tucked away behind a tablet without much hassle. Improving connectivity scenarios (with ORM we are always ready to connect, so EXR will always find us. Why can't the OM5 do that for a home scenario?).

[stevegaspars]

Edit: are you sure about Distance per stroke, as I still can't find that one..

Sorry, my bad. I read your wishlist as stroke length.

[JaapvanE]

Sorry, my bad. I read your wishlist as stroke length.

That is stroke length. You probably are thinking about drive length (i.e. the distance travelled by the handle during the drive). There are some websites that confuse the two unfortunatly, so you are not alone .

[hikeplusrow]

I resisted for some time but can't hold back the following curmudgeon comment.

I like the Concept 2 just fine the way it is. I don't want to buy a television with my rower, and I certainly don't want to pay for a subscription. I don't want a touchscreen anything.

I will keep my 2016 car with tactile controls as long as I can hoping that the world will some day reject touchpad iThings in favor of actual gauges and knobs. I like my Garmin Forerunner 45 way better than my smartwatch. And I really wish I could buy a drink at the airport without having to navigate an iThing.

And I really wish I had my Blackberry Bold back. Or rather, that it still worked and played with modern networks and email providers.

Maybe C2 will go the way of Blackberry if they don't jump on the immersive bandwagon. But that strategy didn't do much for Blackberry, did it? Fortunately, there are millions of existing C2s out there, and they're indestructible ...

Great post. Completely agree. I drive a fourteen year old car without central locking and electric windows at the front only

[Carl Watts]

Electric windows and central locking are a bit of a must have these days but the size of the user manual that came with my new car is something else and I seriously do not need half of the stuff but you must realise that the stuff is there because someone wanted it and lots go put in for safety reasons.

The whole Concept 2 rower could come with two options, those that want the existing PM5 setup and one for those that now want to move to a tablet that not only has the same monitor as an App but you can have a totally customisable setup like the widgits on my car display where you just integrate ErgData into it.

You could even just sideswipe between the PM5 and ErgData and something like EXR on the same device depending what you wanted to look at. Basically the PM5 on my rower became a waste of space, all it was doing was providing a bridge between my PC to RowPro and the my tablet to run ErgData.

What people are not thinking about is that the PM5 also has a pretty limited life, its just something you don't really want to pay for when most people are already updating their smartphone every few years anyway so your new monitor is essentially "Free".

[Ombrax]

The whole Concept 2 rower could come with two options, those that want the existing PM5 setup and one for those that now want to move to a tablet that not only has the same monitor as an App but you can have a totally customisable setup like the widgits on my car display where you just integrate ErgData into it.

I would be truly surprised if C2 hasn't had IT + Rowing experts thinking hard about doing this very thing. Certainly, as they see their competitors come out with fancier and fancier displays, if they're still stuck in a PM5 and nothing else mindset, that would be equivalent to sticking your head in sand and hoping the danger passes you by.

I'm sure they've been watching their market share erode over the years (I assume it has been) and while there's no need to defend their historically high percentage of it at all costs (which would also be a mistake) they do need to think strategically about what they offer vs their competitors, and how those competitors are successfully differentiating themselves. Anything less than that would be foolish.