Support for Tacx older smart trainers

(Kjetil Kirkholt) #161

Hi All.
I have been testing Antifier long before it got its current name and reported bugs back to John. Now i would like to ask those of you that have started using it about your experience?

I have during development experienced issues with slope delays between Zwift and Antifier. The delay can vary from 0 to 6 seconds. When i get to a hill i can see the slope increase on Zwift, but its delayed to my trainer. The same delay i get when the hill is finished. I have discussed this with John for some time, but no errors have been found and no other users have said anything about it. Maybe its my setup or maybe others have not noticed or dont care. Please share your experience.

(Bernard Walther) #162

Hi all,

I have a Fortius T1932 and have the same problem with the delay around 4-5 seconds and we’re not alone Kjetil. A lot of other Zwift users with different brands or models of trainers have the same problem.

I tried with two computers, tried to use a cable to connect to my network, change the USB ports , no change for me

(John Taylor) #163

It would be useful to look at the grade displayed by Antifier. This is a direct display from Zwift. Does it change immediately or does it delay by 6 seconds?

(Kjetil Kirkholt) #164

Antifier experience the same delay. When Antifier updates slope, the trainer get updated with no delay.

(Domen Rogina) #165

John, I tested today, but I have a bad news. With this power curve I didn’t speed up to 40km/h and I didn’t to finish the runoff test. :confused:

It is harder than your iMagic power curve.

I do not believe that such a difference can be with power-meter data and estimated calculated power from trainer? :confused:

(Kjetil Kirkholt) #166

I have done some reading and it seems like the delayproblem is well known and experienced by lots of people on various trainers. The thing i noticed was that those who are using a powertap dont experience the delay issue. Also Wahoo trainers using proprietary Ant+ comms or bluetooth they dont see delay, but when using fe-c protocol the delay is back.
John, could you test the powertap theory since you own one, do a ride with and without?
Also, could the ant sticks buffer messages when using fe-c and cause delay?
I have two friends riding with Zwift locally, both using Wahoo trainers, one using powertap the other Bluetooth and none of them have any issues with delay.

(Jo-) #167

Kjetil …Today at around 1pm United States Eastern Time I noticed significant delay when riding Wattopia … But previously when riding the London Course at around 8pm EST I experienced very little lag… ((Using the climb out of the Subway as a reference, there was no significant delay))

Could it be - number of users? Zwift reported 3000 users. At times today the delay/lag was becoming significant. When this occurred, I slowed to around 20kph for a few seconds, let Zwift catch up to me and then all would be Okay, not perfect but OK. 


(John Taylor) #168

I did a ride yesterday with very little slope change lag. My powertap hub simply reports power- I need the FE-C trainer to change resistance so I’m not sure what test you want me to do. Have you tried looking at your trainer difficulty setting Kjetil? You need to make sure the range of gradients reported by ZZwift matches your power curve file.

To all fortius users- I can’t progress with the resistance settings for this unit until we get a power curve. As a result expect it to be unreasonably hard or easy! If someone wants to manually edit the power curve file as per the readme until they get something reasonable then I’d be interested.

(John Taylor) #169

My ride:

(paha SBG [ZHU]) #170

@John and all Fortius users: I’m working on finding and borrowing a power meter.  

Isn’t here anyone who is considering buying a power meter?  They cost so much and I can’t risk buying and just sending one back.  :(

(Fabi An) #171

 Hi all, i was just wondering if my old Fortius would work with Zwift and have found this thread. Seems to be great work with the antifier. Thank you John and the others who helped for creating this programm. Unfortunately, i could not test it as i am still waiting for the ant stick. When it arrives i could do a power curve.

(Kjetil Kirkholt) #172

Hi Fabi.

Your Fortius will work on Zwift using Antifier if you have a T1932 head unit.
But be aware, you will need two ANT sticks.

If you can help in creating an power curve lots of other users will be happy.
As you can see in the thread above, a powertap or power meter is required to be able to create a power curve.

(Dave Miller) #173

Just curious because I know there are other priorities.  Any closer to a Mac solution?

(Joe Fultano) #174

Jumping in looking for a Mac solution/combo solution as well. I have a Tacx Flow trainer with the upgraded headunit. (images attached as per previous request). I have a new Macbook that is my primary machine but as it stands, like everyone else with OSX, I can not use it on Zwift. I’ve been kicking around the idea of getting one of those mini-PCs just to control my Tacx and run Zwift on my Mac but that’s less than ideal. I’ll be keeping a close eye on this thread as cold weather/darkness is here and I’m stuck inside most days. Will happily Paypal “beer money” for hard work to anyone that can help :wink:

(Robert Klinkhammer[TriFury]) #175

Hi – I have been using a MacBook Pro successfully with TACX T1932 and Zwift.  In this case, I run a windows virtual machine to run the TACX software in parallel with Zwift.  I have used this for about a year now.

I have ported the antifier to MacOS and have gotten it to communicate successfully with my TACX.  However, I lack a second compatible ANT+ dongle to connect to Zwift.    I do have an older ANT+ dongle from my Garmin 310xt, however, it does not seem to be supported correctly by pyUSB.  There may be some USB non-standardness involved.

I  have vector pedals, and have used this for power on Zwift.  The creation of the power curve is interesting due to the calibration step.  I have noticed differences of 30 watts or more between the TACX and vector.  

In any case, it would be nice to not use the cadence and power from the trainer, and just set the resistance.  I want to keep using the vectors.


(Robert Klinkhammer[TriFury]) #176

I haven’t dug too deeply into the GIT repo, but can I create the power curve without being connected to Zwift?  If so, this is something I could do.  But no matter what, due to the roller tension on the rear wheel, the power measurement is affected.  I am not sure how you account for this without a second power meter by which to calculate some bias.  I would expect that this bias factor will need to be added into antifier.

I suppose if you still have access to your original TACX software, you can calibrate the trainer, do some ride at a set speed, and note the power.  Afterwards, you connect to Antifier, ride the same speed, and enter the bias.  I am assuming the power curve itself is not affected by this calibration, other than a bias. 

However, I have found that TACX reads higher watts when calibrated than the Vector pedals.



(Robert Klinkhammer[TriFury]) #177

I have been looking at the structure of Antifier.  It may make sense to separate the main loop into two threads.  The first thread is controlling the TACX.  The second thread is dealing with the ANT interface.  

The ANT thread would send resistance messages to the TACX thread queue.  The TACX thread would send current data to the ANT thread queue.  At the same time, each thread would maintain the correct timing and state (hopefully) between the two interfaces.  In the case of dropped messages, either thread could interpolate the data until new data is received.  I believe that this decoupling of the two interfaces may alleviate the “hanging” issues but its just a guess.


(James Budge) #178

Robert what motor unit do you have? If it were the fortius one would you be able to generate a power curve for us as I note you have a power meter?




(Håkan Samuelsson) #179

John Taylor! Thanks for some stellar work! This is truly awesome!

Thanks all other that has contributed! 

Just to make 100% sure, for the iFlow we are supposed to use the iMagic power curve? I can see that iFlow has a power curve and that is what you are using John, but I can not find a ref saying that the iMagic is the curve for iFlow.

Kjetil Kirkholt, as for the delay, i can not say that I see that with a iFlow and 1932 head unit (only did 35km yesterday, so not too much data to wok with). However, I remember when I got mine, a few years back, I had multiple issues in TTS4 where the correct power was not transmitted between TTS and the wheel unit. That turned out to be a problem with the head unit and it seemed that a lot of 1932 units with serial number starting with 12 (if I remember correct) had this issue. I managed to get hold of a 1932 head unit with serial starting with 11 and have not seen any lag since. Tacx did not acknowledged this as an issue, at least not at the time when I had the problem.  I’m not sure if this is related, but might be worth looking in to… 

(Håkan Samuelsson) #180

Ohh and just a note. The code does not support Python3. Many distros probably are using Python3 as default version and if so the will not start after is ran… 

I added a issue report for this in Github with a solution to how to solve this with Arch Linux, command by command. For other distros with the same problem I believe similar solutions will apply.