I notice the newest Quarqs are advertising “active temperature compensation”. Can someone please explain to me how temperature affects power measurement? Thanks!
strain gauges are little pieces of metal that reside somewhere in your bike (crank, rear hub, etc) and the degree to which they are ‘bent’ causes electrical signals that are eventually converted to power.
Materials change slightly with temperate, so the signal you get from the gauges can change with temperature too. The degree to which this is a problem depends on the materials used, and the orientation of the strain gauges.
Different ways to approach this problem:
-
Careful selection of materials and placement of strain gauges to minimize the effect of temperature on data (info crank is doing very interesting things here)
-
easy, seamless auto zero functions - the powertap is good at this, coasting or a few seconds can trigger a reliable auto-zero
-
active compensation - an on board temperature meter is used to correct data from the strain gauges
#3 is pretty cool if you can’t accomplish #1, and if you can do it reliably. #2 works very well in 99.9% of cases, though you could imagine scenarios where it wouldn’t (like a long hill climb TT where you won’t be coasting)
Where I live, if I use a SRM, I would have to manually zero it several times during a ride due to swings in accuracy because of temperature swings of 30 degrees or more during a ride.
Excellent, makes sense! Thanks!
Along these lines, if I calibrate my current (non compensating) quarq inside the house, then go outside and ride, I’m defeating my calibration, right?
Along these lines, if I calibrate my current (non compensating) quarq inside the house, then go outside and ride, I’m defeating my calibration, right?
Yeah, best procedure would be, set the bike outside ahead of time, calibrate before you ride. AND/OR - calibrate it after your warm up
I’ve been doing it wrong a long time!
I’ll have to look at my Garmin 810 to see how to calibrate mid-ride. I probably should have been doing that too. I’m going to upgrade my Quaq to a temp compensated one. That will help.
Where I live, if I use a SRM, I would have to manually zero it several times during a ride due to swings in accuracy because of temperature swings of 30 degrees or more during a ride.
How much does your ZO vary?
Along these lines, if I calibrate my current (non compensating) quarq inside the house, then go outside and ride, I’m defeating my calibration, right?
Depends on the temperature differential between your house and the outside. The easiest thing to do if you don’t have active temperature compensation is to be sure to backpedal four full rotations after you’ve been outside for a little while (“a little while” depends on whether the temperature differential is small or huge). I generally have to ride about 15 minutes through town, stopping at several stop lights, before I get to the “start” of my serious training route, so I have several opportunities to trigger a manual torque zero. In the grand scheme of things active temperature compensation is handy to have but as long as you remember to manually zero, not having it isn’t a deal killer.
I’ve been doing it wrong a long time!
I’ll have to look at my Garmin 810 to see how to calibrate mid-ride. I probably should have been doing that too. I’m going to upgrade my Quaq to a temp compensated one. That will help.
All you need to do is pedal backwards at least 4 times consecutively to reset the zero while riding with a Quarq. No need to manually zero through the head unit. Easy Peasy.
Isn’t it slightly more accurate to stop and do it through the head unit though? Especially if you have a gummy freehub?
Well, learn something new ever day! 4 full revolutions, one foot? I can do that not far from my house on a descent. Thanks!
Isn’t it slightly more accurate to stop and do it through the head unit though? Especially if you have a gummy freehub?
Well yeah…but don’t do that (the gummy freehub part) 
Well, learn something new ever day! 4 full revolutions, one foot? I can do that not far from my house on a descent. Thanks!
At least 4 full revolutions. (edit: the reason is, as I’ve been told, the zero offset value is stored in 4 separate registers in the spider electronics and the offset applied is the average of those 4 values. Upon a backwards pedal “event” one of those registers is replaced with a new value, so in order to make sure all 4 have been replaced, you need to backwards pedal at least 4 times).
No need to do it one footed…you can keep both feet in the pedals.
What I’ll usually do is first “soft-pedal” forward (i.e. coasting, but with the pedals moving forward but not fast enough to engage the freehub) at ~50-60 rpms while coasting and see if the power reading on the head unit (Garmin 500 in this case) goes to zero. If it shows a “residual” couple of watts while soft-pedaling like that then I’ll do the backwards pedaling zero procedure.
That said, with the latest firmware that compensates out the electronics “warmup” slight drift (another thing that affects zero offset in power meters) I don’t see the zero offset shifting much at all after zeroing it out through the head unit before starting out, unless the temp changes VERY significantly.
What I’ll usually do is first “soft-pedal” forward (i.e. coasting, but with the pedals moving forward but not fast enough to engage the freehub) at ~50-60 rpms while coasting and see if the power reading on the head unit (Garmin 500 in this case) goes to zero. If it shows a “residual” couple of watts while soft-pedaling like that then I’ll do the backwards pedaling zero procedure.
What would it show if the delta zero value was negative and not positive?
What I’ll usually do is first “soft-pedal” forward (i.e. coasting, but with the pedals moving forward but not fast enough to engage the freehub) at ~50-60 rpms while coasting and see if the power reading on the head unit (Garmin 500 in this case) goes to zero. If it shows a “residual” couple of watts while soft-pedaling like that then I’ll do the backwards pedaling zero procedure.
What would it show if the delta zero value was negative and not positive?
It would still go to zero. BUT…the typical case I’m looking at is where the temperature is getting warmer as I go along, and what I’ve seen is that this ends up with the zero offset drifting such that it indicates a higher power reading than what I started at. In any case, it’s usually only a watt or 2.
-
Careful selection of materials and placement of strain gauges to minimize the effect of temperature on data (info crank is doing very interesting things here)
-
easy, seamless auto zero functions - the powertap is good at this, coasting or a few seconds can trigger a reliable auto-zero
-
active compensation - an on board temperature meter is used to correct data from the strain gauges
#3 is pretty cool if you can’t accomplish #1, and if you can do it reliably. #2 works very well in 99.9% of cases, though you could imagine scenarios where it wouldn’t (like a long hill climb TT where you won’t be coasting)
#1 would be reasonably tricky. You really need a zero CTE material (which exist) and it has to be a structural material (not sure, probably expensive if they exist). This could be approximated by using composites since they have near-zero CTEs at room temperature. Unfortunately, composites have a temperature-dependent CTE; metals, the CTE is roughly constant through >>“normal” temperatures. Composites can and do switch from a positive CTE to a negative CTE over a handful of degrees (say 30ish °F) just north of room temperature, especially the materials used in bicycle production. Luckily, there are much easier ways to accomplish this - #3.
#2 I guess works but could be automatically performed without burdening the rider. How does the manufacturer determine an appropriate zeroing interval, and prompt the rider? I live in socal - one of the most benign environments in the US - and this time of year I could head out early morning at 40F and get back home a couple hours later at 80+F
#3, as far as I’m aware, is how everyone (in Aerospace) handles this problem. Remember, airplanes may go from +160F on the ramp to -60F at altitude and telling the pilot to constantly zero or otherwise account for strain drift in a safety of flight instrument is unacceptable. So, we make the crank arm or spindle (depending on where the strain gages are) out of some material - doesn’t matter, but metal would significantly simplify things. Our material, Metal X, has a known CTE through a known temperature range - therefore we can consider the thermal strain of the crank and subtract that from out measurement. The only other outstanding issue is CTE of the actual strain gage (and instrument drift due to temperature), so we run two gages: one bonded to the part measuring part strain, one next to the other strain gage but not really attached to the part. By doing this, we measure the following strains - Gage 1 - thermal strain, mechanical strain due to power input, instrument drift; Gage 2 - instrument drift. Then it’s a simple subtraction problem: rider power = something converting strain to power * (Gage 1 measurement - theoretical thermal strain - Gage 2 measurement).
I’d like to think “active compensation” refers to #3. Anything else, IMO, is (repetitive correction of) a passive system. Strain gages aren’t that expensive and coding a slightly different equation into the firmware seems like a minor effort compared to developing the firmware.
#2 does happen automatically. Any time you happen to coast to a stop sign, or before a turn, or down a hill, if certain conditions are met (speed high enough, zero out of whack enough, etc) it resets it.
Obviously the downside here is if you are in a race where you aren’t ever going to coast for a few seconds, no autozero. Works perfect for training though, you can pretty much not even know anything about zeroing and almost always have good data anyway. Only thing I do is make it a point to coast a few minutes into a trainer sessions.
#1 - info cranks is doing some pretty clever things to try to pull it off, I’m not enough of an engineer to know how good it will be. If you dig they hae some technical documents somewhere. Apparently they do have a temp sensor on board, but they haven’t seen any need to use it for compensation yet.
Can someone please explain to me how temperature affects power measurement?
ZO drift, but perhaps the more important question is, how do non defensible, sloppy/slippery marketing slogans incite questions that a resourceful person would simply use Google to self educate, join the wattage forum community and become instantly bored…or is wasting Jack Mott’s time a community pastime?
Quarq is a tiny mom and pop shop, but surely they could have shelled out a few grand for some message testing research. “Precision personified” might hunt a bit better than “active temperature compensation”, and would obviate threads like this altogether, because you’d already own one
What I’ll usually do is first “soft-pedal” forward (i.e. coasting, but with the pedals moving forward but not fast enough to engage the freehub) at ~50-60 rpms while coasting and see if the power reading on the head unit (Garmin 500 in this case) goes to zero. If it shows a “residual” couple of watts while soft-pedaling like that then I’ll do the backwards pedaling zero procedure.
What would it show if the delta zero value was negative and not positive?
It would still go to zero. BUT…the typical case I’m looking at is where the temperature is getting warmer as I go along, and what I’ve seen is that this ends up with the zero offset drifting such that it indicates a higher power reading than what I started at. In any case, it’s usually only a watt or 2.
OK, but if it did go the other way (e.g. warm bike to begin with, then it’s cooler outside), you wouldn’t see anything in the power numbers with your coasting slow forward pedal procedure that prompted you to perform a manual zero as a negative TZ presumably still reports as zero power in that case.
Be much better if the screen easily displayed both current and set torque values, like you can easily do instantly on an SRM powercontrol, or with a little more difficulty on an old Powertap cervo in test mode. Funny how bike computers that are 10+ years old are superior at this fundamental process than many current day whiz bang models.
Personally I still think a torque zero should only ever be done when no torque is being applied. How one can guarantee that on a spider/crank/pedal based power meter while clipped into the pedals I don’t know.