I’ve been testing a set of Favero Assioma pedals as well as Garmin Vector 3 pedals for several months, and while I’m impressed with their performance both on Zwift on IRL, there’s one Zwifty niggle that bugs me. I call it “sticky watts”, and it’s what this post is all about.
Ready to head down the rabbit hole with me? Let’s go!
Sticky Watts: an Introduction
Zwift processes the data for certain power meters in such a way that the wattage reading in-game gets “stuck” on a particular wattage for 3+ seconds whenever a rider stops pedaling quickly.
To illustrate, here are two quick test rides I did yesterday.
The first shows how power data should look. It compares Vector data from my Wahoo ELEMNT computer with Tacx NEO1 trainer data from Zwift’s .fit file:
Chart 1: Vector (via ELEMNT) vs NEO (via Zwift):
The second shows sticky watts. It compares Vector data from my Wahoo ELEMNT computer with data from the same Vector pedals from Zwift’s .fit file:
Chart 2: Vector (via ELEMNT) vs Vector (via Zwift):
Notice how, in chart 1, the power reading is changing every second, for both lines. This is how power data should look – when we’re pedaling, our power output is constantly changing. (True, there are two times I see where the NEO had the same reading for two seconds in a row, but those occurrences aren’t consistent, or very long.)
Now look at chart 2. Note how, whenever I stop pedaling, the blue line (remember this is the Vector power data from Zwift) “hangs” for three seconds (and sometimes as much as 6 seconds!)
Pointing Fingers
I’ve replicated the sticky watts issue using Assioma pedals as well as Vector 2 and Vector 3s. You may ask: are sticky watts created by the power meter pedals themselves? Or is it due to Zwift’s recording of their power?
Chart 2 above shows that it’s not the pedals. That chart shows data recorded from the same set of pedals, on the same ride, but recorded on two different sources. The purple line is my bike computer, and the light blue line is Zwift. You can see my bike computer recorded the data properly, while Zwift’s data (and only Zwift’s data) displays the sticky watts issue.
The astute observer may say, “But sticky watts didn’t happen in chart 1!” And you would be correct. Sticky watts on Zwift only happen with certain power meters. It doesn’t happen with the smart trainers I’ve tested. Let’s look at why this is…
Event-Based vs Time-Based Power Meters
Power meters come in all shapes and sizes, but if you look at the way they calculate and transmit data, there are two types: event-based and time-based. Here are the definitions straight from the excellent “Power Meter 101” article on Slowtwitch:
- Event-based. This type calculates and transmits power information based on ‘events’. For example, with a crank-based power meter, each cadence revolution is an event. That finite piece of time is the starting and ending point for our time in the power equation. Data will not transmit to the head unit if events aren’t happening (i.e. if you stop pedaling).
- Time-based. This type is identical to the above, but with a caveat. Events happen and power is calculated the same – but this information is stored first, and then transmitted at a specific time interval. The CycleOps Powertap hub uses this. Why? Generally speaking, your wheels spin faster than your cranks – and when you stop pedaling, your bike keeps moving. Events happen much faster and for more of your ride, so they essentially bottle them up and shoot data to the head unit at regular intervals of one second.
As I understand it, pedal and crank power meters are generally event-based. Smart trainers are generally time-based. And that, my friends, is why Zwift appears to process their data in two different ways.
Full transparency: I’m no power meter/electronics expert. And I’m certainly not privy to the details of how Zwift’s code processes power data. But my guess is, this is what happens when I stop pedaling:
- My pedals stop sending data. If they’re paired as the power source in Zwift, that means no power data is coming in. Zwift doesn’t know if this is a data dropout, or me just not pedaling! So it does what Zwift is programmed to do in the event of a data dropout – it holds onto the last power reading it received, for a few seconds. Then it drops to 0, because it is still paired with the pedals, even though it isn’t receiving power data from them.
- My smart trainer keeps transmitting a power reading at regular time intervals, even when I’m not pedaling. So if I’m reading power from my smart trainer, Zwift gets a zero watt reading when I stop pedaling.
Can Zwift Fix It?
Comparing bike computer data to Zwift data might lead a person to think that sticky watts is a Zwift bug – but I don’t think that’s the case. Rather, it’s simply how Zwift has to work given the realities of data dropouts and event-based power meters. Let me explain my theory (and again, I’ll remind you that I’m not an expert here).
My ELEMNT head unit is receiving the same data as Zwift. So when I stop pedaling, it stops receiving data from my pedals (because they aren’t sending anything). So why doesn’t my bike computer show/record sticky watts? Well, the truth is, it does. Sort of. My Wahoo ELEMNT does, at least. if I stop pedaling, it will get stuck on a particular wattage for a few seconds. But it doesn’t record those sticky watts. That’s the crucial difference.
We have to remember that immediate power data on Zwift is much more integral to the experience when compared with power data on a head unit. If Zwift wants to keep its users from seeing a zero watt number (and slowing down!) whenever even a short data dropout occurs, it has to allow for sticky watts. And those sticky watts directly affect the motion of our avatars in game.
In contrast, my ELEMNT can display those sticky watts, but it doesn’t write them to my .fit file. My guess is, the ELEMNT’s data processing code detects a lack of signal and keeps displaying the last wattage read for a few seconds – just like Zwift. But the ELEMNT has the luxury (and Zwift does not) of being able to buffer its data and effectively go back in time and write a different value to its .fit file than it was displaying on the screen. And that’s why we see data like chart 2 above.
Implications for Riding/Racing
Now we get to what I think is the important part of this article: how do sticky watts affect our Zwift experience, if at all?
My guess is, most Zwifters using a “sticky” power source haven’t even noticed the extra watts. This could just be me projecting my own lack of awareness, though! While I’ve ridden probably 20,000 miles on Zwift with power pedals as my primary power source, it wasn’t until I started testing these pedals around April 2020 that I noticed the stickiness. But a look back at my early dual recordings on ZwiftPower (such as Sprintapalooza back in September 2019) shows the issue existed back then.
The good news is, sticky watts don’t matter, most of the time. As long as you’re pedaling, sticky watts aren’t an issue – and we all know that it’s rare to stop pedaling in a Zwift race.
There are a few situations, though, where it can have a significant effect on your race.
Situation #1: Going Into the Supertuck
The supertuck is one of my favorite moves in a Zwift race, since it allows me to hang with the pack on a descent while resting. But timing is crucial when supertucking – you want your watts to hit zero only when the road gradient is -3% or more, and only when your rider is going fast enough to drop into the tuck.
Sticky watts can wreak havoc on a well-timed supertuck, or they can actually make your supertuck an even more effective move. They wreak havoc when you stop pedaling, but Zwift sticks your watts at some low number (say, 100W), which means your dropping into the supertuck is delayed and your power is low while you wait. This can cause you to drop from the group you’re trying to hang with.
But if you hit the wattage hard then stop pedaling immediately, your watts will typically get stuck at a really high number – perhaps 500-600 watts! This speeds you up, then drops you into the supertuck as soon as the stickiness ends.
Is it cheating to use sticky watts to your advantage when entering a supertuck? I’d say so, if you’re doing it on purpose. But here’s the sucky part – event-based power meter users are stuck between a rock and a hard place here. Because the only way to avoid sticky watts getting stuck at a low number and possibly messing up your supertuck is to 1) go hard then stop immediately (cheating) or 2) slow decrease your watts, so your pedals send power readings all the way down to zero. Neither of those approaches is a good one.
Situation #2: TTT Sink and Sprint
The “sink and sprint” move is used by TTT riders to move from the front of the pack to the back of the pack (learn more about it on TTT Secrets from Matt Gardiner). Sticky watts making timing your sink a real challenge, since your watts get stuck at a particular wattage for a few seconds before going to zero. That means you have to stop pedaling before you really need to.
Situation #3: Mini Microintervals
I know for a fact that racers have attempted to exploit the sticky watts bug in events by doing super short intervals – that is, pedaling hard to spike power, then stopping the pedals so the power gets stuck at that high wattage for a few seconds. Then do it again, and again!
Here’s a power file of a short microinterval test ride I did using my Assioma pedals (the purple line is the power data from Zwift (notice the sticky watts) and the blue line is the power data from the same pedals from my bike computer):
Look at the difference in average wattage between the two readings: Zwift’s power is 20% higher!
Would this approach work over an entire race? Not for me – I couldn’t continue doing microintervals for 40+ minutes. But perhaps some riders could. More likely, a cheater could use this technique for a short (1-2 minute) attack at a key point in the race, or in the final sprint.
As I understand it, ZwiftPower has some detection built in for this sort of repeated sticky watts behavior.
Situation #4: Shortened Sprint Finishes
If a rider’s setup consistently produces sticky watts at or near their peak power, they can use this to “coast” across the finish line at full power. Hit the sprint hard, put in a few final pedal strokes at maximum effort, then stop pedaling for the final 3 seconds over the line.
You may laugh – but riders do it already in Zwift races.
Conclusions
Armed with the knowledge of sticky watts, event-based power meter users find ourselves in a bit of a predicament. Do I keep using my pedals as my primary power source (because I want consistent readings between indoors and outdoors) even though it can affect my race data? Or do I switch to my smart trainer – which may be less accurate and will generally not match my outdoor power meter?
The best answer will vary from rider to rider due multiple factors. But I think it’s safe to say this is a concern that will continue to simmer in the Zwift racing community for the foreseeable future.
Your Thoughts
Have you noticed sticky watts with your Zwift setup? Do they affect the Zwift experience in ways that I missed? Share your thoughts below!
Thanks for sharing your thoughts.
I had sticky watts with my single sided favero Bepro, now i am in a 2 week old Kickr Core and it sometimes als registers sticky watts..my one week old Assiomas Duo dont, at least not in the Elmnt bolt. havent tested it to Zwift as power source. Iam in ant+ connection though.
Still trying to get my head around this
Your head unit will never record sticky watts. It’s a Zwift-specific thing… as explained above.
Sticky watts can also happen if your signal is poor – but that’s not a repeatable occurence.
Funny thing is that it now happens with my pedals, i see it on my headunit and when oauring to zwift its also there..i will dual record my TTT and see what happens, but i am almost certain its the Ant+ connection
I’ve tested ANT vs Bluetooth, but it doesn’t seem to make a difference.
Your head unit probably SHOWS sticky watts, but doesn’t actually record them. That’s why my ELEMNT does, at least.
Wahoo computers have sticky watts too. They don’t correctly record 0 watts after signal is lost, but smooth it out for a few seconds, from last received power to zero. Take a closer look and you’ll see. It’s not as bad as repeating same value for few seconds but it’s not correct power recording either.
Generally with data display units… the display is updated at a slower rate than the actually recorded data.. That is so that the observer doesn’t see numbers flying around all over the place.
I have to differ about head units not recording sticky watts. My stages gen3 using ANT+ have the sticky watts. Using bluetooth does not. My head unit is lezyne mega xl. I always thought the sticky was more an issue with the power meter.
Great article – super helpful. Zwift definitely feels a bit different when I ride with Vectors vs. KICKR – and this answers a lot of those questions why. Thank you!
Great write up! I have been struggling with this and pulling my hair out trying to figure out how to get supertuck to work faster (using Assioma’s as my Power source). At least this explains why!
This is a great article and also explains for those curious, why avatars continue to sprint for a few seconds past the finish line in a live race broadcast. I’ve investigated this myself a bit and have not only seen it with the pedals you mentioned in zwift, but it is also reproducible in rouvy. In testing and looking at rider data, I generally see only 1 second of sticky watts with these particular pedals, but observe it to be far worse with a DZero for example, often in the 3 second range. I filed a bug about this and… Read more »
It would be great if pedal PMs could get updated firmware that continued sending a signal of 0 watts when pedaling stopped. Hmm….
One easy solution would be to just remove the sticky watts for races and keep the current behavior for everything else.
The sticky watts are there for a reason, though – without them, wattage would drop to zero anytime there was a dropout. There would be a LOT of power drop complaints, I think, if sticky watts went away!
Probably it would be best to offer both: races with sticky watts and without them.
All the more reason to fix/address the dropouts, then. (Still waiting for the Kickr V5 ethernet dongle to be released…)
I just had another thought on this that I would imagine would be simple to implement. For those of us with a reliable connection, why not make it a client-side option? I would then just go into settings and toggle off dropout tolerance / sticky watts, after 1 second with no events, power drops to 0.
Therefore zwift can fix everything by making the Direct Connection Hub to work on zwift
so being an avid Zwift racer I have of course seen the post addressing this issue. My thinking is that since the only reason Zwift needs to know your power, is to it can plug that in along with all your other data and determine your estimated speed and simulate what it would be like riding IRL. That being said if you are riding IRL and you stop pedaling, yes your watts will immediately drop to 0, but does your speed? So my question is how do smart trainers and/or Zwift deal with that momentum? and then which is actually… Read more »
Riding in real life?
I don’t think we’re anywhere near that. Not at all.
This is another issue Zwift laid firmly with the hardware manufacturers and called them out for poor design.
More collateral damage in the march towards world domination.
I would have to agree. I have been using a kickr snap for a few months, and if I stop peddling (e.g. in a casual group ride) for a few seconds, my wattage drops to about 1/2, then trails with the actual speed of the wheel (remember, wheel on). Recently I thought I’d treat myself to a nicer trainer and went for a neo2t and absolutely hated it. If I stop peddling, then within 8 seconds I’ve come to a complete stop from about 20 mph. This might be useful to level the playing field for racing, but for recreational… Read more »
I have always preferred my double vector 2 pedals, rather than the power of my Hammer. although in zpower analysis my pedals score about 2% less, these sticky watts create favorable physics. Some events already require the trainer as the primary power for the event. okay i adapted to any of the physics. but I prefer my pedals.
It isn’t dropouts that zwift/other head units are compensating for by giving you sticky watts, but low cadences. if you want to allow cadences down to 30rpm you need at least 2s of sticky watts as that is the time between crank events, down to 20rpm requires 3 seconds.
That’s a good point. But Zwift is ALSO concerned with dropouts – I know that for a fact. So I suppose it’s both!
Isn’t pedals like the assiomas sending power data far more than once per crank event though?
I don’t have any power meter – just my Kickr Core. I get sticky watts on that and it is quite annoying. I don’t really mind it in races, as you say it smooths the power output a bit and while it might affect supertuck I am usually blowing so much when I am trying to get into supertuck that I don’t really notice! It seems to affect how the trainer performs during workouts – it often hangs up for a while at the start and end of intervals – I find that a lot more annoying. I also run… Read more »
Based on my understanding of the article, you shouldn’t be getting sticky watts from your time based trainer as a power source.
As far as I can tell, the only time you should see sticky watts from a time-based trainer is if you’re getting power signal dropouts.
The computer is only about 1-1.5 meters from the trainer (to the right of my bars) and I don’t get signal dropouts in other cycling apps on the same computer (mainly TR). It definitely seems to be something Zwift is doing when it is processing the power. Obviously, I don’t have a power meter to compare against, but I have dual-recorded using my Garmin before, so I can compare those. I find they don’t match very well. The garmin sits on my bars… I have some dual recorded rides I can compare but I wasn’t doing anything scientific then (intervals… Read more »
My ANT+ signal starting sticking a bit on my Core recently. It used to be smooth, so I expect it’s a signal issue
Eric, does the 3 second sticky have anything to do with the settings page where there is a power reading setting of either 3 seconds or 1 second….. I have changed mine to 1 second as I’m using the Elite Drivo trainer as that has constant power readings throughout each rotation.
It has nothing at all to do with power averaging, Mick. That setting just changes what you SEE, but doesn’t affect what Zwift records and what power it uses to move your avatar.
A different question but sort of similar field, what happens to sticky draft watts, ie. extra watts which you produce, but which don’t allow you past the rider in front. I’m forever trying to find the lowest power that keeps me in the draft, but to overtake the rider in front needs extra watts to break the threshold of the sticky draft. I’m sure there is a wattage window in between these two figures. Are these watts just wasted, do they get banked and used when you overtake?
They don’t get banked. And I’m not sure what the threshold is, and what the threshold is based on. Is it a speed difference? A power difference? Something else? But yes, you can definitely waste some watts sitting in the sticky draft.
It most likely is a speed difference. I am an almost 100 kg rider and i am getting into the sticky draft behind some of the 9 year old feather-weight kids. My absolute power should be much higher when we go uphill. The other option would be w/kg difference.
My initial thoughts are that it’s not cheating. It’s gamesmanship. It’s only cheating if a rider is hacking either their hardware or software to gain advantage or the people that provide the game’s framework (Zwift) explicitly disallow it as part of the rules of the game. The ball is in Zwift’s court to change their framework if they believe it’s cheating otherwise racers are going to try to find and exploit all of the advantages that they can.
Zwift has explicitly disallowed it. In the zwift esport ruleset under appendix B: Sanctions. Here it is mentioned that “Exploiting disconnection / lag / dropout behaviour” and “Exploiting bugs in Zwift software” is a sanctionable actions. These rules are of course developed with the focus on big race events, but my guess is that when a rider rider is reported to zwift, a lot of same rules apply, in checking if the rider is doing something wrong
Zwift is about sportmanship … I wouldn’t exactly qualify trying to abuse a software glitch to gain an athletic advantage as good sportmanship.
It is definitely considered cheating, and it will most certainly get you disqualified (also from smaller events)…
This is fantastic work. It’s a nice demonstration of the difference between reporting present data (what is my power now?), and reporting historical data (what was the power during my ride?) The luxury of hindsight is a big advantage. I had previously thought the reason for E-Racing rules requiring data from the smart trainer to be to require transmission thru the drivetrain before measurement (it’s a 3% or so advantage to measure power upstream of the drivetrain, around 2 minutes saved up Alpe). But this is another reason, and perhaps the main reason: smart trainers can come much closer to… Read more »
Aside from all the sticky watts, potential cheating or losing out stuff it’d be great to see an article covering the best, and most cost effective, dual recording power meter options available to Zwifters especially since it might be needed for ZRL etc data checks – many riders like myself probably don’t have one or any idea of options 👍
Yeah, working on some reviews of the Vectors and Assiomas… and hopefully Powertap P2s also.
So far I’ve been happy with the Vectors and Assiomas… both solid and reliable. Assiomas more affordable though, which is why they’re the most popular power pedals on Zwift!
I’m so glad you published this. I’ve been scratching my head many times wondering why my power takes so long to drop to zero when I want to super tuck. There’s also a weird issue when starting TTT races. I have to pedal for around 3s before the power stops reading 0 when using Vector 3 pedals.
Interesting article, and perfectly timed with my recent attempt to reach out to Zwift Support regarding the exact issue (2 weeks, 4 different people, no response yet). Depending on the event, especially ones with surges, I would see higher power with my Vectors than my Kickr. At first I assumed it was my Kickr, because we always assume we’re stronger than we are, right? Also my Kickr was behaving oddly on a tech support driven ERG ramp test, so I had lost some faith in it. I sent snapshots of the exact issue to Garmin- the plateaued trailing power curve… Read more »
I’ve had the opposite experience with my Vector 3’s: They respond slowly to changes in effort and drop off almost immediately when I stop pedaling. I switched back to the Wahoo Kickr for my power readings and I seem to be faster. I’ve been just using the V3’s for cadence and power numbers on my Garmin 935. The numbers I’m seeing on the watch are usually lower than what I see on the Zwift data HUD.
Maybe I’ll switch to the pedals for the Sprintapalooza tomorrow morning and see how it goes. Film at eleven.
I though I had read about a similar issue with Garmin Vectors a long time ago, but I was remembering wrong. The recording using a Garmin Edge like the 1030 gives an sticky cadence. If you stop pedaling the power reading goes to 0 but the cadence is repeated 2-3 times. I wasn’t able to find the source of the information right away but I verified in my data. Looks like with the Zwift recorded data both cadence and power are sticky.
Why is it considered cheating if I didn’t know about sticky watts, but I’m able to hold 10+ minutes of micro intervals up the volcano climb? If I’m performing consistently like that, zwiftpower will upgrade me and then I’m forced into the next category trying to hold on. Like another commenter wrote, it’s gamification, and if zwift doesn’t like it, then they need to do something in the software that addresses it. You can’t tell me that I need to get a Kickr or neo for $1200+ because the software doesn’t stop listening to the power meters I use. If… Read more »
Zwift needs to address it if they want to be more than a game, which is what they seem to be aspiring to- helping amateurs secure pro contracts, hosting professional races, premier racing leagues. At those levels, accuracy is essential, and making a selection can be about 20 watts. I hear ya, upgrade, but that only works for so long. I used to argue to my friends that the platform was fair enough and not to worry about cheaters, but I def have come to question the crew putting down > 4.7 w/kg FTP’s- numbers the pros can barely produce… Read more »
Context in esports, games take this stuff way more seriously. Its the cycling world that is way behind, not the gaming world.
4.7 wkg are numbers pros can barely produce???
Sure there are some who have terribly calibrated power meters on zwift (like old mate who did 400 watts avg, and in real life strava, avgs 70 lol).
But zwift attracts the best most serious training cyclists, zwift isn’t the local bunch ride.
It must attract some serious talent- as evidenced by the video with Alex Dowsett on how TT,ing in Zwift is so hard. He finished 23rd or so, 369 watts for a little over 20 min. Wiki has his weigh at 73kg (could be wrong) but as a TT specialist and previous hour record holder he is no slacker. His effort in the vid has him at just over 5w/kg, an FTP of around 4.8. I did a race this am, using my Kickr it was 6% below my vectors, only other power analysis in the race with someone using petals… Read more »
Can we agree that according to the spirit of the game, success is driven by athletic performance, in combination with strategic elements intentionally introduced by the developers? Can we also agree that abusing a sticky watts issue inherent to some technical software glitch, present in only a handful of power meters is clearly not a part of this? Zwift has stated abusing this glitch to gain an unfair advantage is against the rules. You will get disqualified when your power output reveals you abuse it. So for the same reason you don’t participate in real life events with an engine… Read more »
Apart from the other aspects already mentioned, I have taken part in A races, where the winner was DQ’d because of the use of sticky watts (luckily at least WTRL can detect it). Upgrades won’t take care of this, when you are already in A category.
May that’s the reason why the Zwift racing rules say that the power source has to be the smart trainer.
Exactly this. One of the recent rider sanctions mentioned that they were trying to make a power meter look like the trainer for the race. I didn’t understand the problem at the time, but it makes sense now!
If the problem is with Zwift receiving a “null” value when watts are zero, could a longer-term fix be that trainers manufacture their units to transmit a decimal, like “0.001” instead of zero, and Zwift turns anything below 0.002 into a zero?
Sure seems like that would work! Not sure why it doesn’t happen.
This has been an issue since… Well, since forever.
But instead of doing something about it and admitting fault Zwift chose to lambast and troll both athletes and hardware manufactures blaming them.
The usual machiavellian, arrogant behaviour. Who cares about community collateral damage with 450 million in the bank?
Who cares about fixing known issues with 450 million in the bank.
Who cares about hardware manufactures with 450 million in the bank.
It is not that simple. Sticky watts are still there fore a reason. And the reason are widespread problems with wireless signal quality. Zwift without sticky watts would be simply unusable for large part of the riders. And this is in fact not a Zwift fault.
What utter protective tosh.
If that were the case they should have provided details to both athlete and brands explaining why this happened.
Instead they banned athletes and blamed them and the hardware.
Zwift being Zwift
I ride with the Quarq Dzero power meter outside all of the time, however I select the Quarq Dzero as my power meter in Zwift then I can pedal fast, hit 600 watts, then abruptly stop pedaling, and Zwift will hold 600 watts indefinitely. Yes, forever. To snap Zwift out of its stuck power stupor, I have to soft-pedal at 100 watts to have it drop from 600 watts. Yes, I’ve reported the bug to both Quarq and Zwift, and the Quarq people tell me that it’s a Zwift bug, and the Zwift support people are radio silent.
I think they don’t want to burst the bubbles of everyone breaking their outdoors ceilings?
I guess this is why Zwift states you must use a trainer as your primary power source for official races and then submit your secondary file for certification
Maybe sort races by power source 😉
For my indoor rowing apps this problem is even worse due to the lower frequency of power events which are once per stroke. The only potential solution I have come up with is to occasionally compare the calculated energy from the real-time sampled wattage events (including the “sticky” dropout protection power) with the (less real-time) reported cumulative power (calories, joules, etc.) and then spread the error as it accumulates. Over time, this makes things fair in terms of total energy applied to the game, so your 20% advantage from sticky power goes away. But, it does not solve the problem… Read more »
Couldn’t you just record on Zwift with your trainer and record on your head unit with your power meter? That would keep Zwift more accurate and still give you a file with readings that match outdoor activities.
Sure, but the secondary file typically isn’t loaded into your main system for analysis so it shows up on power curves, etc. But yes, this would be the way to do it (and the way I’m doing it moving forward.)
I belive Zwift for with the Favero gets the 3 seconds power, this could be as simple as that, it does not gives the instant power.
No, the Favero definitely isn’t averaging its power across three seconds!
This is why I actually prefer our dumb trainers (Kinetic Road Machine). They do not transmit power, only rpm which is then used in a lookup table in Zwift to determine power. This assumes you are using a speed sensor and not the inride sensor. Zwift does have another issue that must purely be bad programming, if you stop pedaling and then pedal backwards at a high cadence Zwift will give you a high power for a long time. This has been a problem for sitting in the pen for the TTT events. If you pedal backwards, you leave the… Read more »
Your cadence isn’t used for power on a dumb trainer – it’s your wheel rotations that are used. And that’s a way less accurate method than using a power meter, even with sticky watts!
Very interesting article ! I’ve recently switched from using my smart trainer as the power source in Zwift to using my Vector 3S (power matching) and have also noticed differences between my Edge 1030 and Zwift. It seems Zwift (or my Garmin ANT+ USB stick) is much more sensible to the quality of the ANT+ signal than my Edge, as I often get signal drops in Zwift which I don’t get on my Edge computer.
Pretty sure this is all down to the head unit(s) and the way Zwift processes data, although some PMs (I seem to recall from the ANT+ power profiles) need three consectutive seconds of zero cadence before they declare zero power. The ANT+ transmission intervals are defined by the ANT+ specs and vary slightly for the type of device (power, HRM etc) as a crude way of handling (avoiding) collisions (of ANT+ transmisisons). ANT+ protocol also sends the same data for four consecutive transmisisons at approximately 250ms intervals so tha if a “packet” is corrupted there is a duplicate to take… Read more »
Could you use an external cadence sensor to correct this for the power meters? One that is separate from the power meter and the trainer.
That’s a good question, Derf. I haven’t tested that! But my guess is the cadence sensors work similarly to the pedals, which means they’re event-based… so it may not help things.
SInce this is such and “old” thread I don’t know if I will get an answer, but really, how hard is it to put a @#@#% USB port on these smart trainers and to quit worrying about all this signal dropout mess for people that can simply “plug in”?
What I’m curious about with your testing protocol is which device received data through BTLE, and which through ANT+. The difference is important, and sometimes staggering. The reason for its importance is the broadcast protocols, as well as recording protocols, and a possible overlap. With ANT+, there are the two types of broadcast as you mentioned… the event based (each rotation), and time based, which updates at 1Hz, or once per second (this is ANT+ standard). Pedaling at 60rpm will yield the same update frequency as the ANT+ standard for time-based broadcasting. Pedaling faster will yield a faster update. The… Read more »
It’s a smart question to ask, j.anthony.
I did my tests with both protocols, just to make sure the sticky watts weren’t restricted to Bluetooth or ANT. But they happen on both on Zwift.
(I believe Bluetooth broadcasts at 16Hz, while ANT+ is at 4Hz. I may be mistaken though.)
Just did a race where this was getting discussed in a very animated way in the peloton as we approached the finish.
amazed at people’s ability to type in the finale.
incoukd see it happening especially on the climbs for the rider I question but not sure where it lies.
Is it in the spirit of the game is the question as the current implementation of the software allows it.
I watched your race on youtube. What do you think should happen to the rider in question? Would it make any difference to you if they came down to 1w/kg instead of 0 in between his pushes? And then is it their fault if Zwift knows about it and hasn’t updated the issue?
i have a new elite direto XR purchased in december 2020. Do you know if the sticky watts / power dropout issue applies to this type of trainer
Far as I know, sticky watts only apply to pedal or crank-based power meters. Never smart trainers.
thank you Eric – when i ride i often go hard for a short interval to stay in the group, bridge a gap or go for a sprint immediately followed by a rest period when i will stop pedalling to catch breath allowing the flywheel to keep momentum going for a few seconds. do you think this would cause a potential a drop-out issue?
Shouldn’t!
thank you again Eric👊. Great site and informative discussion. keep up the good work!
Wow Eric, this is very interesting and beautifully explained. I landed on this page while I was checking why,in zwift analyzer, I get sometimes many searches of the ANT for my trainer and I noticed that I get one search everytime I stop pedaling… Probably this is a side effect of the power dropping to 0 (sure, I’m not pedaling) but Zwift cannot know if this is actually a drop or not… Maybe I’m wrong, but for the 2nd stage of TDZ, I got 10 descent and 10 times I’ve stopped pedaling (looking at the charts in zwift, the watt… Read more »
Thanks for the article Eric. Once again, very interesting, and relevant. Bit of a double edge sword, of course, since I think this issue was a bit hidden away in obscurity. Bringing it out in the open is definitely useful, but I hope it doesn’t inspire more riders to try and abuse it.
So just to be 100% clear to those who are still not sure – because there seems to be some confusing in the comment section: yes, it’s definitely considered cheating, and yes, it will get you disqualified from races when you get caught abusing it 😉
I use Powertap P1S pedals. I have never experienced sticky eats when using my laptop for Zwift with Ant+. Recently I have been using a 5th gen AppleTV with Bluetooth and have experienced sticky watts.
Quick question, Is this the same for powertap pedals? I have never tested, but when I dual record my pedals on zwift and on garmin my garmin AP/NP power is higher than my zwift power because I do get a few dropouts every race. I connect my powertap pedals via ant+. So, I wonder if that makes a difference and that sticky watts don’t happen with ant+
I have a wattbike atom and it really seems like it generates sticky watt behaviour. I paid particular attention to time till supertuck and it was 3+ secs when above 58km/h and greater(smaller?) gradient than -5. The readings are, likewise, slow to drop to zero despite power reporting being set to instantaneous. If this behaviour is not sticky watts,(because trainers don’t generate them), then what can it be? I am using bluetooth and zwift on the ipad. I find it really annoying as you start wondering if the speed/gradient isn’t enough and end up intermittently pedalling to avoid slipping back,(thus… Read more »
Looking at the crude ANT+ datastream from my Assioma DUO pedals, there are around 4 ANT+ messages per second (channel period is 8182 counts, quartz frequency is 32768 Hz, thus the ANT+ message frequency for Bicycle Power is 4.005 Hz) but only every forth ANT+ message contains new data. Therefore, I ask me whether the Assioma pedals are indeed event based? For example, SRM cranks broadcast to ANT+ event based, i.e. the next outgoing ANT+ message after every complete crank revolution is updated. But SRM use the Crank Torque frequency protocol whereas the Assioma pedals use the Standard Crank Torque… Read more »
Before I had a smart trainer at home I used to use Zwift at the gym on the Wattbike trainer, and I had the same issue but exaggerated as I could sprint at 500ws then suddenly stop pedalling and it would keep my wattage there until I began to pedal again.
I think I am getting a version of sticky watts on my new Kickr 2020 v5 trainer. Power sticks (for lack of any other way to describe it) usually at a high output for me but not always and then after several seconds to a half a minute it sticks at a lower power for several seconds. Nothing I do makes a difference. Eventually it will go back to normal. Running Zwift on a 2019 MBP and using ANT + with an extension to connect the trainer. I have not figured out what triggers it but so far it has… Read more »
This is an interesting topic: (because I want consistent readings between indoors and outdoors)…Or do I switch to my smart trainer – which may be less accurate and will generally not match my outdoor power meter?
**Do you have any articles about this?**
Eric, I have been curious about how Sticky Watts occur, since I noticed that a rider suddenly was as fast as me in a time trial with an extreme technique with on off power numbers.
Have done some own research and analysis that might be interesting. Covering different protocols/trainer setups and also including some data on Tacx Neobike and Neo trainer. Both actually giving Sticky Watts, but in another way then for example Assiomas.
If interested, just send me an email.
Regards
/another Eric 😄
I’m hesitant to believe it’s something specific to Zwift and a specific power meter. It sounds to me like you’re computer’s receiver is not working as well as your head unit and you’re getting dropouts in the data into Zwift. Data transmission is not only dependent on the source of the data, but also the receiver. You need to isolate the variables to narrow it down. You should test intervals with Zwift then immediately switch to another training application and perform the same test. Do this maybe 3-5 times and compare the results. If it’s only in Zwift, then that… Read more »
It’s well known and not due to “dropouts”, far too coincidental that it happens and can be replicated every time pedaling stops don’t you think?
Thx for this article…
Just had the issue with SRM power meter, on a race when we made that on the volcano descent (takinf front each in turn on fast rotation with a kind of little sprint each time)… 9 watt out between Garmin and Zwift (235/244w)… will try with kickR as power source, but that’s sucks when you love your SRM for irreproachable precision to make correlation between indoor and outdoor rides … is there any soluce preview from Zwift team?