I've gotten a few rides in on Quarq's 2nd generation power meter, the CinQo Saturn, and I could tell already it was going to be a boring review. Glue a magnet to your chainstay, and the CinQo goes on just like any other pair of cranks. Pair it up with a Garmin or an iBike (I used an iBike, and pairing was dead simple) and you're good to go. Power and cadence come from the CinQo, speed is measured by the iBike's ANT+ sensor. Readings matched my Powertap so well I didn't see much point in doing a detailed analysis like I did for the Polar. So I had nothing left to say but that it costs $1800, $2300 with a Garmin 705, $2500 with an iBike, and at 820 grams it weighs about the same as an SRM at a fraction of the cost. So I decided to have a chat with Quarq founder Jim Meyer to find out more. A lot more.
Andy Shen: We have the Powertap and the SRM, they're pretty well established. What was your thinking when you decided to make another power meter? What did you think you could do better, or differently?
Jim Meyer: The main thing was the SRM was really expensive. Powertap, the pricing was more reasonable. And then of course at the time that I started, both of those were wired devices, so the idea was to come out and do a wireless crank based power meter that has the functionality of an SRM with pricing more in line with the Powertap. That was the initial concept. Then, of course, things evolved after that, with ANT+ coming out, everything going wireless, it took away that as a differentiating feature, but it also confirmed that we were on the right track to start with.
AS: So when you started ANT+ hadn't been established yet?
JM: That's correct. It didn't exist.
AS: So in the process of developing the CinQo you had to change over and adapt to ANT+?
JM: Well, we'd actually decided to use the ANT originally for our wireless, and we had made that decision on our own without knowing that ANT+ was going to happen, and that Powertap and SRM were both looking at the same technology. We had just found ANT on our own, figured that it looked like a good system for us.
AS: So ANT is the frequency and ANT+ lets everything work together?
JM: That's basically it, correct. ANT on its own is a frequency and actual hardware and communication protocol put together. ANT+ is the last bit you need where the manufacturers agree on the standard data tables and the messages. And that's what lets things inter-operate.
AS: And I noticed Powertap just went to ANT+ too.
JM: Correct. They were on ANT previously for their radio, so for them it was a matter of changing their messaging to conform to the documented standards.
AS: Will all the systems be able to work with each other in the future, will the Quarq be able to work with the SRM head or the Powertap head?
JM: It's a matter of implementation, but that's basically the intent. But there's details in the actual implementation of everything. I think some point in the future there'll be a compatibility chart so you'll see what's actually implemented. The hardware, like the SRM Power Control VI, is ANT, and reads the SRM unit, and I think officially they're supposed to make that compatible with the PowerTap and the CinQo, but as far as I understand they haven't done that yet.
Powertap, on the other hand, they're playing by the rules a little more strictly with their computers. They've got a compatibility challenge, 'cause they've got a lot of equipment in the field already. So what they've set up is, their hubs actually transmit on two separate channels, so their new ANT+ hubs they send out both the current ANT+ message that's compatible with the Garmin and other things, and their old message that they used to have, so that maximizes their compatibility with everything.
AS: Right now, the CinQo just works with the Garmin and iBike?
JM: Correct, that's the only ones that have been implemented to date.
AS: And how's the Qranium (Quarq's computer head) coming along?
JM: We started developing that originally, when we started the project, ANT+ didn't exist, the Garmin didn't exist, so when we started we had to make our own computer. So that's why we started building the Qranium. But as time got on, we've ended up, now that we have the 705 (Garmin) and there's other computers that are under development in different places, we've been mainly focusing just on the CinQo. Right now we don't have a projected release date on the Qranium, and when we do, it'll be mainly be a hacker toy to start with, and not a really viable option for most people to switch to.
AS: So it's taking advantage of the open source nature of it?
JM: Right. I think over time the Qranium could develop into a really good alternative to the Garmin, but it's not going to be there soon.
AS: There would be a need for something like the SRM, which displays everything at the same time, and not cost something ridiculous.
JM: Their computer itself is, I think, 8 or 9 hundred dollars.
AS: The new CinQo Saturn has a ring connecting the bolt holes, is that to reinforce it, make the flex more...predictable, accurate?
JM: The new ring on the Saturn, you see the ring is separate from the little chainring, it's its own part, so you can use whatever little chainring you want, that ring helps the calibration remain more stable through a broader range of conditions, and it actually turns out that it also makes the production process a little more predictable, and therefore more repeatable and faster for us. It does nice things for the stability and the calibration, makes it easier to produce.
AS: When someone wants to just buy the spider, you recommend sending in the rings and letting you install the rings. Does installing or changing the rings affect the calibration of the unit?
JM: Not really. The claimed accuracy is +/- 2%, when you swap the rings, typically you're going to stay within 1% of the factory calibration. For the most part, when people send it in, it's a good way to do it to begin with. It's not so much the chainrings as it is the installation on the crankarm. We feel better when we can do that here. That said, it's kinda changing a little bit, as we get more experienced it's getting more predictable for us to ship them out unassembled. There'll be some models that will almost always be sent out that way, like the Cannondale and the Specialized models, it's natural to send those out as a separate spider that can be installed by a shop.
AS: You mention the Specialized and the Cannondale models, those are BB 30? And what about compact versions?
JM: Specialized and Cannondale are both BB30 and will be available in 6-8 weeks. We are working on a compact version for all the cranks now, also expected to ship in 6-8 weeks.
AS: As far as resetting the offset by backpedaling as opposed to the Powertap and the SRM, where you do that at the computer head, is the thinking there that by having the process within the crank, it's one less operation at the computer head since you don't produce the computer head? (To zero the offset on the CinQo, you simply backpedal a few revolutions.)
JM: It's multiple reasons. We support the calibrate command that comes with the Garmin. That's helpful, because when you do that the CinQo will return a number, and that number is really helpful for troubleshooting. We do like that, but there's some people who don't necessarily know that they need to calibrate every once in a while, they don't read instructions, and it's not immediately obvious. So we built in the auto zero function so that if a user is out in the field, they're just riding around, they don't anything more than just pedaling a bike, chances are, sooner or later, they're going to pedal backwards sometime, and that'll reset the zero point.
The idea was to take one piece of training and knowledge away so that it's just not required. If you don't know anything it'll still work. If you do know things, then you pedal backwards before each ride, and you'll zero it that way. It's also nice because it doesn't rely on communication with the head unit in order to make it happen, it's just automatic on its own.
AS: Does it really need eight revolutions?
JM: It's four revolutions now. Also, if you do two one day and two the next, it's using all of them. If you want to completely reset it you do four revolutions. We found that that's better, we had it originally on a single revolution, but you're putting all your eggs in one basket. If you pedal backwards once, all the next thousands of revolutions are referenced off of that one measurement, and so it doesn't provide as consistent a zero.
AS: Is that equivalent to the y-intercept?
JM: Correct. y=mx+b, it's basically setting the y intercept.
AS: Any concerns that by taking the zero offset with your feet attached to the pedals, it won't be as accurate?
JM: No, your feet are isolated from the torque measurement. When you backpedal, the CinQo does feel the reverse torque of the drivetrain friction, but we compensate for this internally.
AS: People make a big deal out of being able to calibrate the SRM, putting weights on the pedals, inputting numbers, recalculating the slope. Is that something that can be done or needs to be done with the CinQo?
JM: We basically tried to get away from that as much as possible. We're opening up our dealers now, and we're building a calibration system that will consist of a weight that will go on the crank, and it's actually an Eee PC computer with our own software loaded on. That'll be a step by step guide for doing the calibration with the weight, much the same way the SRM does now, except with the SRM you have to apply the weight, write numbers down, calculate the new slope, and then enter that slope.
The system we're building will do all that automatically, put on the weight, press a button, no weight, press a button, flip sides, do it here, do it there, and it just steps you through. And at the end, it tells you if you're in range or out of range, and if you're out of range you can set the new slope. We don't have a system where you can put the slope in willy nilly to whatever you want, and that's intentional, because we feel like if we're going to allow someone in the field to set the slope, the hardest part isn't putting in the numbers, it's knowing which number to put in. So we feel we need to provide a complete solution to know what that number needs to be. If we let people adjust that on the fly we're going to have more problems with that than we are solving things. Our results have been very very stable, so it's not a common thing to have to do.
AS: Any word on whether the iPhone will ever be opened up to this?
JM: Not sure, that really depends on Apple, if they decide to allow arbitrary communication over the serial port, on the dock connector. Right now, in the SDK they don't allow communication over the serial port, so until they totally open that up there's not much anyone can do for making devices that talk over that dongle. The devices that you do see for car systems, that's actually talking on a predefined communications system that outlines commands like forward, back, pause, volume up, volume down, and it's restricted to those commands.
AS: It's the kind of thing that will probably have Garmin quaking, right?
JM: The flip side is that the screen isn't daylight readable. The Garmin display is transflective, which means it works in direct sunlight, and that's one of the catches with the iPhone. Just set it on your bike and turn around 360 degrees on a sunny day, you'll see it sometimes, but not all the time.
AS: Battery life, too.
JM: And battery life. But having said that, the iPhone does have Garmin worried, and they're working on a cell phone, so it'll be interesting. Maybe we'll see something from them this year, and who knows what kind of impact that'll have on bicycle technology.
AS: Yeah, there's apps for the iPhone that do 90% of what the Garmin will do, for 3 bucks.
You sell the CinQo with Rotor Rings. Do they affect the slope?
JM: Very slightly, typically we calibrate them with the rings on, but in the field they're typically within 1%.
AS: I've used them with an SRM while also running a PowerTap, and I could've sworn it was more than that. I don't know off the top of my head.
JM: I haven't looked at it on an SRM real close, I've heard about people having problems, seeing a pretty big change in slope when you swap rings.
AS: It makes sense, your rpm is lower during the power phase, it's going to be lower than the average rpm, so...
JM: You've got two errors that start to take place. One is the actual slope change due to swapping the chainrings and the stiffness of the chainrings. And that's one of the things that's nice about the new Saturn design is that it helps minimize the changes in slope due to the different ring designs.
AS: I hadn't realized that. Are you saying since Rotor Rings, which basically have a ring, unlike normal rings which have a gap between the bolt holes, are you saying that difference in rigidity changes the slope?
JM: Yeah that changes the stiffness of the whole assembly.
AS: So since the Saturn has a ring it's less of a change.
JM: Right, it helps. The SRM has a ring around the outside as well, so both of us have gravitated towards the same thing.
AS: That solid pentagon...
JM: Yeah, and the other error that you get that's introduced by Rotor Rings is that the pedal velocity within the stroke is not constant. It accelerates and decelerates a bit, depending on the ring diameter. Even without that, on a round ring, you still have some acceleration and deceleration within the pedal stroke due to inertia and the change in velocity of the whole bike. You do speed up a little bit on the power stroke, and decelerate a little bit in between.
AS: So even though the geometry of the Rotor Ring dictates that you should decelerate this much, it's not that much of a difference.
JM: It's not that much. Part of it is, with the Rotor Ring, that you can almost make the pedal velocity more constant during the stroke. I've done a little bit of modeling on it. You have to look at the power output of the rider on a given hill, headwind tailwind situation, and you can start to see the inertia effect, the inertial load on the bike, versus the aerodynamic and rolling resistance, pure drag. And then if you assume some sort of torque profile produced by the rider, you'll see each downstroke the whole bike accelerates, and then every time you're at the dead spot, with the cranks vertical, the bike is coasting and decelerating. So the bike velocity and the pedal velocity are not constant anyway. It varies sinusoidally with every stroke.
AS: So by that logic, these power meters should be most accurate in high inertia situations because there's less velocity change, and while climbing they'll be less accurate?
JM: Right. The more the inertia you have and the less velocity change, that means the crank revolutions will have a lower change in velocity with a revolution, and that helps everything. Even on the outside, the effect, assuming all your worst case stuff, it's hard to push it much over 1, 1.5%. It's pretty bounded.
AS: I guess high cadence would be more accurate than low cadence?
JM: Typically, yes.
AS: But in all these cases it's not enough to be significant.
JM: It comes out in the wash by the time you're done. All the power meters, us, SRM, Powertap, we're basically measuring the angular velocity once per revolution. We have multiple reed switches but it's more to turn the unit on and off, to sense when the cranks are moving backwards. The SRM has one reed switch, the Powertap, I think there's one reed switch in there. So all of them are making the assumption of constant angular velocity within the revolution. The PowerTap, of course, looking at wheel speed as opposed to crank speed.
AS: In other words, you're only measuring once per revolution, so you have to assume that rpm is constant through that whole rev.
JM: You have a time for each revolution, 600 milliseconds or whatever.
AS: Your website says you have a track version that'll analyze with more detail, for standing starts?
JM: To start with we're going to have it use the same broadcast messages as the road version. But then we'd also like to implement a high speed data mode where you can get all the messages from the unit. We'll be working on that over the next year.
AS: Is that having more reed switches and reading more than once per revolution?
JM: It'll mainly be giving torque information throughout the stroke. We can pass along the reed switch state which will give us some indication of where the crank is within the revolution. But overall, reed switches are not rotary encoders. They work ok because from one rev to another, the error can't really accumulate, so if the switch trips a little early on one rev, it's going to be a little late on the next. You can't have an accumulating error. But when you use multiple reed switches, say you have five reed switches spaced out evenly on the crank, they're all a little different, so they'll all trip at different times, so you get a basic constant of where it is in each rev, but you don't get an exact - this is exactly 72 degrees, and here's another exact 72 degrees, they don't have that sort of resolution.
AS: But you can still get torque through the entire revolution with one switch?
JM: Correct. What we can do is broadcast all the torque that we see as measured at 60 hertz, and then we can broadcast all the torque messages and then whenever a revolution completes, we can send that, piggyback that information on, so you can see a full torque profile measured at 60 hertz, and then the markers where each rev completes.
AS: When you say 60 hertz, does that mean that at, say, 120rpm you'll get 30 readings per pedal rev, or every 12 degrees?
JM: Yes, at 120 rpm you get 30 measurements per rev. However, each of the 30 measurements are not single instantaneous samples, they are composed of many of sub-samples that span a portion of the time between each main measurement.
AS: Is the track version out?
JM: The prototype actually just got in today, and we're just hooking it up on the fixie right now. We'll give it a try, and then it'll be sent out to the Hawk Relay team with Sarah Hammer.
AS: What issues arises from not being able to coast?
JM: Not too much. We'll find out if there's anything we've forgotten about, but certainly you do have a reverse torque on the whole system. The challenge with that is to try to get the zero offset as stable as possible. So we have to make sure the reverse torque doesn't mess anything up. That goes in the category of field validation. And then of course the reverse pedaling doesn't happen, so zeroing is going to have to happen as a manual command.
AS: I heard with an SRM, as soon as you apply reverse pressure, the offset will be off.
JM: That would not surprise me. I haven't tested one directly, so I don't know for sure, but that wouldn't surprise me one bit.
AS: Who are you sponsoring?
JM: The track side, Hawk Relay, which has Travis Smith, Adam Duvendeck, Sarah Hammer, and on the road side it's the Cervelo Test Team, on the men's and women's side.
AS: How's it going with Cervelo?
JM: Really well. We just got back from Portugal on Sunday night. We spent a couple weeks there getting everyone trained, they'll be training and racing with it on all their bikes this year. It's a pretty big commitment on our part.
AS: Are they using Garmin heads?
JM: Yeah, for this year.
AS: Is your lightest unit comparable to SRM or heavier?
JM: All of the units on the same cranks are all lighter than the SRM units.
AS: I weighed mine, I have it at 820 grams, at least compared to claimed weight on SRM's it's heavier. I was thinking, 'ok, it's a tad heavier but it's a lot cheaper', but you're saying it's not even heavier.
JM: No, it's not. We've seen their website, we've subtracted, unless there's something we're missing... We have them here, and the Dura Ace SRM weighs in at around 195 grams, 200 grams, versus 150 for most of our models. Typically unit to unit, it's between 30-50 grams lighter.
AS: So there's really no...no advantage...why would someone buy SRM?
JM: As near as we can tell, just from watching people discussing it on forums, the whole reason why anyone's buying an SRM now, is that they've been around a long time, and everyone trusts them. That's the reason why people are buying them now, they've got a track record. I think that that's ok, we'll just keep chipping away at that, 'cause we plan on being around for a long time too.
So have you had good luck using the unit and everything?
AS: Compared to the Powertap it's 1 or 2% difference at the most. One particular ride, I haven't done any max efforts per se, but it seemed like the max was pretty different, but I just haven't ridden it enough to make a judgement on that.
JM: Right. Typically what we've seen is that the Powertap will throw larger maximums than what you'll see on the CinQo. I don't know enough of the details in the torque reporting of the Powertap to know exactly why that is. One of the things is that the CinQo will always measure a complete revolution, so both power phases and both rest phases. Whereas a Powertap is on a fixed time updating, it can't know where your legs are, so it can happen to get a sample that includes three power phases and two rest phases, and when that happens it's going to look like the average power for that interval is higher than what you've got for a complete crank cycle. So that's typical, I haven't worked with an SRM with a Powertap together too much, but I think people have found similar results using an SRM and a Powertap.
AS: I have, but I haven't written it down enough, but anecdotally, almost every time it's higher on the Powertap.
Have you done much with live CdA with the iBike?
JM: Personally I've just done a little bit, but not much. Typically, you start a company related to bike stuff, you end up riding a little less. So most of the work I've done with the iBike is strictly from the data standpoint, making sure the CinQo data is interpreted properly and that everything's working.
AS: I wanted to review the CinQo, but I had nothing to say. You install it like a normal crank, you put on a magnet, and it just works. There's nothing to say about it.
JM: Honestly, that's what we're going for. If your story ends up being 'this thing's kinda boring', well, we like that story! We want it to just go on the bike and just disappear, and you never worry about it. You get numbers on your head unit that are always right, that's our goal. We want the the whole thing to be as unobtrusive as possible.