Predictive resistance (or: Zwift has no breaks!)


Have you ever encountered the situation where the trainer resistance lags behind the actual terrain? If I understand this correctly, the reason is the trainer is slow to respond to commands sent from the local PC.

However, Zwift knows well in advance the terrain and slope (given riders speed, and no breaks). Hence, it can (most times) send a resistance “script” forecasting, say, 5 seconds of resistance into the future to the trainer, which would act upon it as per time indicated (say “change gradient to 10% in 2 seconds”)

Granted, this requires trainer support (its kind of a chicken and an egg problem - no use having this feature unless there is trainer support, and no use having trainer support unless some app takes advantage of it…) However, I would assume this stands to make an impact wrt the game being more “realistic”, thus an interesting area to explore.

Ride On!

The disconnection between what we see and what we feel is the most jarring thing for me. I really don’t understand why there’s such a long delay between things.

The road goes up; a few seconds later my trainer responds.

I stand up and sprint hard; a few seconds later Zwift notices.

Of course, we learn to live with it to some extent, but give it’s all electronic communication that travels at light speed it seems like it should be so much snappier.

The assumption that the electronics travel at light speed is far from reality. Yes the two way communication between Zwift & Trainer and vice versa is quick but it can be affected by many factors.

So setting a ‘prediction script’ might be ok for one particular day on your set up but on another day it may be slightly off timing. Local internet speeds, local internet congestion may affect a specific users latency levels. Then there is the factor that certain geographic locations are unable to guarantee internet performance levels at all.

I do not expect Zwift to drop a toe in the murky waters of predictive scripting so we will have to live with the tiny delays between visual and physical feedback.

Comparatively, not really. It’s a lot closer to reality than 4 seconds are.

Even playing games over the internet on a 14.4k modem in the 1990s we’d have a 200ms RTT. And that’s travelling hundreds or thousands of miles to a server and back using POTS technology and cabling. And as for input lag, almost no one ever noticed any delay between moving a (rollerball) mouse and the view moving on screen.

Yes, there’s also a mechanical response time depending on the type of trainer. But Tacx themselves claim “the NEO can change the resistance in a split second”.

Of course it won’t be near instantaneous, but a 4 second response time is absolutely ridiculous.

Maybe it’s not 4 seconds with my Neo 2, but it’s at least 2 or 3 seconds. Maybe longer. It’s most noticeable when the road goes up and the resistance doesn’t, or the road levels off and I’m still working against a gradient. The sprint lag is there too. If I remember some time I’ll do a video and measure it precisely.

PC Windows 10 and ANT+

My Neos Ant+ connection is very unstable. I suspect a board failure of some kind and it may not be uncommon.
I’ve got plenty of test data showing regular dropouts but only from the Neo - heart rate data remains active (for months I thought it was a problem my end).
TrainerRoad recently updated their software with a specific fix for Tacx Neo dropouts noted in the changelog - where ERG/Resistance mode needs resetting.

If the Ant+ signal is unstable with high packet loss, it’s not unreasonable for it to take 2-5 seconds to update.

Note: Windows Bluetooth for the last few months has been 100% stable.

1 Like

I mostly don’t use bluetooth because my PC doesn’t have it. And because for some reason I don’t trust it as much as I do Ant+. Dunno why, I just feel - for no valid reason - that bluetooth is a more flaky unreliable technology.

I may well have it back to front.

1 Like

something must be wrong with your set up - i have a neo and it changes resistance almost instantaneously over ant+ or bluetooth (on the android version)

the only lag i get is with the elite drivo but that is a trainer issue, it is just slow!

Don’t see what could be wrong with it. Neo 2, Ant+ dongle about 30 cm from the Neo. Unless something is faulty, but I’ve no way of testing the Neo. Nothing much in the way of dropouts according to the last log I parsed.

have you tried it in erg mode? do the changes still lag when doing a workout?

have you tried the neo using the app or other software manually changing resistance/wattages to see if there is a similar lag?

might be worth trying to rule out a faulty neo.

Nah, not yet. I wasn’t expecting anything much more responsive than I have, so I hadn’t considered there could be a fault. I have 30 days access to the Tacx app to try some time, so I’ll see how that goes.

I was just wondering whether it was a slow PC/software issue with zwift.

the tacx app should be free? unless you mean their PC software.

Yeah, I mean that. I’ve never used the Tacx app other than to calibrate or update the firmware on my old Vortex, so I don’t really know what it does. =)