The resistance swings were extreme. I was just trying to warm up for a while at 150W. The resistance was swinging from 0 to "barely turn the pedals". Not rideable.

---

Books @ Amazon
"If only he had used his genius for niceness, instead of Evil." M. Smart

The tacx flow and tacx neo are two really separate beasts. The issues that you're experience with the flow likely deserves it's own thread. I own both a Neo and a Vortex (similar to flow) and can speak from experience.

Your experience with calibrating the roller tension on a flow with a power meter was similar to mine (on a vortex). I too was very much on the loose end of roller tension to get my vortex to 'match' my power meter. I say match because these lower end smart trainers really don't have a power meter and are just estimating power based on roller speed.

I also noted that when I used my PM to control the trainer that the resistance dance around a little (not quite what you describe though). I wonder if changing the setting in trainer road to average power across a 3sec average would help you to manage the sudden changes in resistance?

Ultimately, I ended up just using the power estimation from the vortex and didn't worry too much abut controlling the trainer through a PM. I got the workout that I wanted, albeit at the expense of precise/accurate power feedback.

These entry level smart trainers all sacrifice accuracy and this is not a problem that is exclusive to tacx. ...The Neo, on the other hand, is in a different league of trainers.

Ditto beston... The Flow Smart is a dirt cheap, entry-level trainer priced hundreds below typical wheel-on trainers (including the Vortex). If I were going to guess, your problems were that the Flow is an el-cheapo, too-good-to-be-true trainer, not smart trainers as a class of device.

Of course. But the problems added up fast.
1) Extreme variations of watts from 0 resistance to prob 4x the watts that it was directed provide.
2) Wouldn't communicate with either TrainerRoad or Wahoo Elment calibration routine.
3) Tried to calibrate using Tacx software and tensioner assembly unthreaded itself at the moment I got it "barely inside" of the broad acceptable calibration zone. This is a mechanical design issue so completely unrelated to the first 2 issues.
4) There is a plastic lever that moves the brake against the wheel. Flip it down, adjust the tensioner, and the firmly pull the plastic lever back up. When I still had insufficient tension pushing the electronic brake against the tire, the plastic lever required so much force that it was at risk of breaking. That made me very concerned that once I got the tensioner tight enough to be in the middle of the "correctly calibrated zone" as shown by the Tacx app, the plastic lever would break. This is an mechanical design problem unrelated to the other issues.

I figured that the only way I was going to fix all those issues for certain was to buy a smart trainer of entirely different design, or just go back to my dumb trainer that's been a reliable friend for decades.

I talked to Amazon today and they'll let me send it back. Fortunately, I hung on to all the packing materials. I think I'm going to let another year or two go by and let smart trainer design mature a bit.

---

Books @ Amazon
"If only he had used his genius for niceness, instead of Evil." M. Smart

Likely a reasonable call to return it. Just a couple of points

Points 1 and 2 are both software issues with Trainer Road. I'd contact TR directly and I have found them to be very helpful.
I'm not even sure I have ever successfully used the calibration routine in TR. I have always just relied upon the Tacx app for that kind of stuff.

Point 3 and 4 is related to the set up of the trainer. The roller can be initially installed/positioned in 3 different places according to wheel size. Perhaps you need to adjust which position its in.

There's certainly a chance that there is a problem with your trainer, but I'd certainly inspect how you initially mounted the roller to the to the frame.

Re. 2. Wasn't just TR, the Elmnt wouldn't calibrate with it either. The problem there was probably that the Tacx trainer needed to be first calibrated with the Tacx software. Tacx, so my theory goes, hadn't worked hard enough with TR and Elmnt to ensure that their apps provided useful info as part of the calibration fail. Like an error message in TR and the Elment that said "Calibrate device with OEM app first, before attempting to calibrate with us". Or the trainer could have shipped with a tag, attached to the electronic brake, that said "download Tacx Trainer app and calibrate before first ride".

The Tacx arrived with no documentation at all. So there was no clue that it might require some unique calibration routine prior to trying to get TR or an Elment to communicate with it.

I neglected to mention that the trainer didn't want to calibrate with a Garmin 935 watch either.

Re. 3. I didn't mount the electronic brake in the correct position. The device ships with the brake already mounted. Could have had a tag on it that showed different mounting positions for different size wheels, but it didn't have any of that. This is my 4th trainer in as many decades. I've not had to change the mounting positions of any of the resistance mechanisms before, so it was not intuitively obvious that it was necessary this time.

As an engineer, I read documentation. So altho a tag fastened to the electronic brake would have been best, had the issue been addressed in accompanying documentation, I'd have figured it out.

Re. 4. Flimsy electronic brake positioning lever could have been a product of me mounting brake in wrong position. As discussed, I didn't mount it. That aside, I think it's just a matter of a weak design. Had the tensioner stayed together, that plastic lever would have been subject to helacious force in order to push the brake against the tire. The lever was already bending and I didn't have the tensioner tight enough. Sooner or later that lever was going to snap. My old Computrainer was a tank. My TravelTrac, from the early 90's, functions just fine. No plastic on them that bends alarmly in casual use.

---

Books @ Amazon
"If only he had used his genius for niceness, instead of Evil." M. Smart

Hmmm. The Flow trainer usually needs to be constructed. Did your Flow come in the Tacx box?

This site shows basic video instructions of how to put the Flow together out of the box. At 1:20 into the Flow video, they show how to adjust the installation of the roller.
https://support.tacx.com/...up-your-Tacx-Trainer

Basic manual for the smart trainers can be found here. It includes instructions to get the app. I didn't see this manual unpacked with the box in the video above.
http://tacx.com/...iners_2016_en-GB.pdf

I have to say, it doesn't take long to google search for these things;)

I'll admit that Tacx isn't the best in terms of in box instructions. With the flux, it had the basic assembly instructions, and mentions of the gap, but that was it (that said, it's so easy to put together, you can easily skip reading them and figure it out). The videos on their site are much more helpful. The App works well to calibrate it, I've used that with some of the older smart models (wheel on) with good success as well. I've also calibrated various models within the tacx software as well, which works similarly well. in their wheel on models, it is finnicky when you do the spin down, and very sensitive to tire pressure, so you do need to play with tension some, but get the hang of it pretty quick... I've never tried to calibrate my trainer within other platforms though. Even for Zwifting on my Flux, I just calibrate from my app before riding.

I haven't had the issue mentioned with the Neo. That said, I always run a small gear in Erg mode. I run the smallest gear that can reach the power that I am interested in. The reason is that the larger the gear you use the smaller the amount of brake force that is necessary at a specific power output. This is because the torque at the pedals is reduced by the gear ratio at the cassette. So, at a given power the trainer's brake might be operating 3 times harder in a small gear than a big gear. So, suppose you want to control 300Watts to 2% stability (6W). That means that the power measurement, signal to control the brake, etc. needs to have a total noise level that is less than 2% of the level at which the brake is operating. So, if you are in a big gear where the Neo could hit 2000W the brake is only operating at 300/2000= 15% of maximum. 2% of that is only 0.3% of its maximum which is a really small noise level. If, instead, you pick a gear where the trainer can only hold 600W all of those numbers are 2000/600 = 3.33 times larger. This gives the Neo (or any other trainer) electronics and software a much easier time to control the power. The moral of the story is to ride as small a gear as you can for the intervals that you plan to do, If you have some sprints in the middle of the workout then shift gears up before the sprint starts and then shift back down when they are over.

I will try that, but tacx say to run big gear with a straight chainline.

Don't use the Tacx tyre, they are crap and slip and get worse and worse as they age.
A Conti gator works really well and you won't have to tension it to the moon.

Not replying to you specifically, but this thread is a disaster. 2/3 of the posts are not about the subject trainer.

Tacx run the same software for all of them and no doubt a lot of the firmware protocols are cross platform, so yes it is relevant.

That's interesting logic, but it doesn't match reality - for my Neo, at least. I used to run small chainring because it's quiet and felt smoother. Then when I had my P1 pedals on the trainer bike for a while I discovered my Neo was over-reporting power; threshold-type efforts were recording a good 20-30W lower on the power meter than the Neo's target power. Changing to large chainring brought the Neo and power meter into agreement within a couple watts. So my experience is that the small chain-ring produces less accurate results. YMMV of course.

I have only checked power accuracy a couple of times since the bike that is on the trainer currently does not have a power meter. In the past I used a different bike with a Stages powermeter. At that time the match was as good as can be expected- the difference between my Stages meter and the Neo was around 15W at 300W (5%). I felt this was reasonable considering the accuracy of the Neo and Stages, the fact that the Neo measures after drivetrain losses, and my left/right power imbalance that the Stages doesn't take into account.

What I was referring to in my post was how well the Neo controlled the power. Its stability of the power under (nominally) constant cadence and gearing. I have seen less wandering around of power in a smaller gear under these conditions. Particularly at lower powers like the recovery phase of intervals.