Direto XR problems with ERG mode

[JohandeWit]

Hello to all,

Please apologize for my english

I'm new to home trainer and therefore to direto xr that I acquired a few days ago. So I'm not able to compare to others ERG mode behaviours.
That's why I'm writting in this post to get advices from you with more experience on HT.
My direto XR is in version 073 already.
From now on, I only activated rouvy 30 days trial so I'm not able to compare with Swift.
Therefore I made the test recommanded by rouvy and send them the results : https://support.rouvy.com/s/article/The-Trainer-Resistance-Seems-To-Be-Wrong?language=en_US
I post you an image of the excel graph I made to let you tell what you think of power variations which seem quite big to me. FYI I was using 34/21 gear for the test I think. That's the result :

The test was like this : from second 15s to 195s, 1 minute at power 150/200/250W and from seconds 285 to 465s 1 minute at slope 0/3/6 %. What do you think ? Should I create a ticket to elite too, or the results seem quite good to you ?

Also a lot of you are talking about smoothing calibration ? I didn't find this option on rouvy ? Is it only on swift ? Therefore I can't tell you on which smoothing value was made the test.

Thanks in advance for your answers,

What strikes me here in your graph is that (apart from the variation in the power) the actual response to the power step is rather quick. I mean for example the step from 150 -> 200 w. You see a respons within seconds. That is not what I experience using ERG mode. So I looked up your Rouvy link. Do I understand correctly that you make the power adjustment yourself? I mean after 1 minute at 150w, you manually set the requested power to 200 using the Rouvy application right?

[urdel62]

Hello to all,

Please apologize for my english

I'm new to home trainer and therefore to direto xr that I acquired a few days ago. So I'm not able to compare to others ERG mode behaviours.
That's why I'm writting in this post to get advices from you with more experience on HT.
My direto XR is in version 073 already.
From now on, I only activated rouvy 30 days trial so I'm not able to compare with Swift.
Therefore I made the test recommanded by rouvy and send them the results : https://support.rouvy.com/s/article/The-Trainer-Resistance-Seems-To-Be-Wrong?language=en_US
I post you an image of the excel graph I made to let you tell what you think of power variations which seem quite big to me. FYI I was using 34/21 gear for the test I think. That's the result :

The test was like this : from second 15s to 195s, 1 minute at power 150/200/250W and from seconds 285 to 465s 1 minute at slope 0/3/6 %. What do you think ? Should I create a ticket to elite too, or the results seem quite good to you ?

Also a lot of you are talking about smoothing calibration ? I didn't find this option on rouvy ? Is it only on swift ? Therefore I can't tell you on which smoothing value was made the test.

Thanks in advance for your answers,

What strikes me here in your graph is that (apart from the variation in the power) the actual response to the power step is rather quick. I mean for example the step from 150 -> 200 w. You see a respons within seconds. That is not what I experience using ERG mode. So I looked up your Rouvy link. Do I understand correctly that you make the power adjustment yourself? I mean after 1 minute at 150w, you manually set the requested power to 200 using the Rouvy application right?

Hello,
Yes that's it. It' s in free ride mode, I manually adjust power level by clicking 5 times on the + button (adding 10W). And it's similar on slope mode ( 3 clicks very minute to add 3% slope). It was not a program session, maybe that's why the response may seem quicker I don't know. I join you an extract of a programmed workout made on rouvy and it seems less responsive, maybe my cadence is not that good :

add photos to google album
Thanks,

[jhankey]

Urdel62, now there was I thinking that your erg step responses was actually a lot better than what I see. So glad to understand it wasn't erg... which is a whole different issue with the direto.
In free mode, I expect the direto is one of the most 'honest' trainers . If you free ride in zwift or Rouvey I would expect the direto sensed power and your power meter pedal traces will line up perfectly... it seems very good at this. Other machines tend to smooth even this a little, and it effectively hides what is actually going on. I come from a tacx flux, that reports very stable power figures in all sorts of ride conditions. Massive lags though on changes of power etc.... it is pretty much a guide rather than actual data and the averages may be close to actuals. Where as the direto is telling you exactly what you are doing. However, what we are all finding is that in erg mode it is practically useless. The power variation response is almost too quick and never settles, or doesn't settle in a time frame that humans can regulate against. It needs near perfect cadence for it to work anywhere near right. Additionally, the direto calculated cadence seems to be all over the place. I was zwifting zone 2 yesterday, and if I soft pedalled after a period of say 200w down to 100w on a decent, keeping the same cadence, the report cadence numbers where jumping all over the place, 20 rpm or more.
I have thought about connecting up a cadence sensor, but this will only interact with zwift, and the direto is still only going to calculate its own cadence. What I will do guys is ask the local Elite rep up for the Direto External Cadence sensor and to see if that improves things. Has anyone else got this fitted. ( Note: Tacxs neo can have a similar problem if the crank length is not long enough)

[urdel62]

Urdel62, now there was I thinking that your erg step responses was actually a lot better than what I see. So glad to understand it wasn't erg... which is a whole different issue with the direto.
In free mode, I expect the direto is one of the most 'honest' trainers . If you free ride in zwift or Rouvey I would expect the direto sensed power and your power meter pedal traces will line up perfectly... it seems very good at this. Other machines tend to smooth even this a little, and it effectively hides what is actually going on. I come from a tacx flux, that reports very stable power figures in all sorts of ride conditions. Massive lags though on changes of power etc.... it is pretty much a guide rather than actual data and the averages may be close to actuals. Where as the direto is telling you exactly what you are doing. However, what we are all finding is that in erg mode it is practically useless. The power variation response is almost too quick and never settles, or doesn't settle in a time frame that humans can regulate against. It needs near perfect cadence for it to work anywhere near right. Additionally, the direto calculated cadence seems to be all over the place. I was zwifting zone 2 yesterday, and if I soft pedalled after a period of say 200w down to 100w on a decent, keeping the same cadence, the report cadence numbers where jumping all over the place, 20 rpm or more.
I have thought about connecting up a cadence sensor, but this will only interact with zwift, and the direto is still only going to calculate its own cadence. What I will do guys is ask the local Elite rep up for the Direto External Cadence sensor and to see if that improves things. Has anyone else got this fitted. ( Note: Tacxs neo can have a similar problem if the crank length is not long enough)

Ok thanks for your infos ans sorry to let you think than my direto xr behaviour was better than yours ..
However for me I don t think that with rouvy erg mode is unusable. The power is all over the place on the graph but I have the sensation that if I keep a constant cadence the trainer résistance is keeping constant also. It may be than rouvy is making some smoothing on power measure received from direto before doing some régulation. And maybe than Swift IS reacting directly to values received from direto and is making too much régulation. Just a thought of mine. Or maybe I m all wrong and the full job is made by the trainer and rouvy is just reporting data.
But that s sure that power send by the direto seems way too volatile to make proper régulation. I compared my workout session to one made by a tacx trainer on rouvy and power graphs are very différent.

[JohandeWit]

Ok thanks for your infos ans sorry to let you think than my direto xr behaviour was better than yours ..
However for me I don t think that with rouvy erg mode is unusable. The power is all over the place on the graph but I have the sensation that if I keep a constant cadence the trainer résistance is keeping constant also. It may be than rouvy is making some smoothing on power measure received from direto before doing some régulation. And maybe than Swift IS reacting directly to values received from direto and is making too much régulation. Just a thought of mine. Or maybe I m all wrong and the full job is made by the trainer and rouvy is just reporting data.
But that s sure that power send by the direto seems way too volatile to make proper régulation. I compared my workout session to one made by a tacx trainer on rouvy and power graphs are very différent.

I do not think that in ERG mode the software applications Zwift or Rouvy are actually controlling the trainer. Neither Zwift nor Rouvy nor another application will be the source of the bad behavior in ERG mode. I attached a quick whiteboard sketch to clarify what I mean.

in ERG Zwift and Rouvy only sent a setpoint to the trainer. It asks the trainer to maintain for example 200W. The trainer than adjusts the resistance based on the power it measures internally. So there is a control loop in the trainer that regulates the resistance in such a way that you produce 200W. (I guess it's more complicated than my basis sketch of course).

And it's logical to do it like this. Otherwise, if Zwift or Rouvy would be part of the loop that controls the resistance in ERG it would introduce quite a big and unknown delay in this loop. This because of the Bleutooth/ANT+ transmission lag and your PC or tables processing time. They will vary with each setup of every customer. And if you know a bit about control engineering you know that delays (or death time) in control loops is killing

The trainer also reports back the measured power and cadence. Those are the number Zwift or Rouvey show you on the screen. I expect that the so-called "power smoothing" is an additional filter to that reported power, but it also affects the behavior of the control loop.

I strongly feel that our trainer can be improved by optimizing the control parameters inside the Direto control loop. I also read in other threads (of previous Direto's) that Elite actually gave some users files to update those parameters. I also asked for that, but till today we did not get any respons from Elite admins.
(Me being an engineer as well can even envision a special app to adjust some parameters, that's how I would do it anyway )

sketch2.jpg (232.39 KiB) Viewed 11373 times

[urdel62]

Ok thanks for your infos ans sorry to let you think than my direto xr behaviour was better than yours ..
However for me I don t think that with rouvy erg mode is unusable. The power is all over the place on the graph but I have the sensation that if I keep a constant cadence the trainer résistance is keeping constant also. It may be than rouvy is making some smoothing on power measure received from direto before doing some régulation. And maybe than Swift IS reacting directly to values received from direto and is making too much régulation. Just a thought of mine. Or maybe I m all wrong and the full job is made by the trainer and rouvy is just reporting data.
But that s sure that power send by the direto seems way too volatile to make proper régulation. I compared my workout session to one made by a tacx trainer on rouvy and power graphs are very différent.

I do not think that in ERG mode the software applications Zwift or Rouvy are actually controlling the trainer. Neither Zwift nor Rouvy nor another application will be the source of the bad behavior in ERG mode. I attached a quick whiteboard sketch to clarify what I mean.

in ERG Zwift and Rouvy only sent a setpoint to the trainer. It asks the trainer to maintain for example 200W. The trainer than adjusts the resistance based on the power it measures internally. So there is a control loop in the trainer that regulates the resistance in such a way that you produce 200W. (I guess it's more complicated than my basis sketch of course).

And it's logical to do it like this. Otherwise, if Zwift or Rouvy would be part of the loop that controls the resistance in ERG it would introduce quite a big and unknown delay in this loop. This because of the Bleutooth/ANT+ transmission lag and your PC or tables processing time. They will vary with each setup of every customer. And if you know a bit about control engineering you know that delays (or death time) in control loops is killing

The trainer also reports back the measured power and cadence. Those are the number Zwift or Rouvey show you on the screen. I expect that the so-called "power smoothing" is an additional filter to that reported power, but it also affects the behavior of the control loop.

I strongly feel that our trainer can be improved by optimizing the control parameters inside the Direto control loop. I also read in other threads (of previous Direto's) that Elite actually gave some users files to update those parameters. I also asked for that, but till today we did not get any respons from Elite admins.
(Me being an engineer as well can even envision a special app to adjust some parameters, that's how I would do it anyway )

sketch2.jpg

Hi,
Thanks for the nice graph
Yes after second thought I think that my hypothesis was not right, and you're right applications must not be part of the regulation loop.
There is another "newbie" question I guess I'm asking myself :
In all the data we can find in the .FIT, is there any kind of indicator representing the resistance imposed by the trainer. If so, we would be able to see if indeed the trainer resistance is allways moving or on the contrary if it remains quite stable if cadence is quite good.
PS : to get the json param file, you create a tickets to elite ?

thanks,

[guiperpt]

Quick update and a wild theory.

Today for the first time was able to get all the stars in a workout, which impressively was the ZA #7 - VO2 Development. Pretty impressed with the ERG mode performance today, honestly no complaints even about the multiple changes (Smooth set to 2).

The only thing I did since last workout with ERG, despite putting a couple of kms on the trainer (probably around 600km, and I warn no idea if it's related, and not even sure if it's safe, was to perform a calibration with the Allen key adjusting the belt tension so there is no difference between the calibration value and the value written under the trainer.

In my case, the value written under is 6791 and was getting 6787. Following the procedure from Elite on this video, https://www.youtube.com/watch?v=hf0xsbVd91I managed to get it to 6791 (even though after a couple of kms it's back to 6790). Please note that Elite says clearly in the comments that you shouldn't do the procedure!

A couple of screenshots from the workout are attached. Please note that on the 135W sections, I did reach for water, for a banana, etc. But if I want, focusing on the cadence (you must really focus to keep it mega stable), I can keep the power mostly within -5W and +5W from the target, without any huge + or - 20W spikes.
If it matters using the ERG mode on a 36x23 gear.

Maybe the calibration offset is somehow related to the ERG mode performance?

[roilev]

...I come from a tacx flux, that reports very stable power figures in all sorts of ride conditions. Massive lags though on changes of power etc....

jhankey same here, I came from Flux too. And though it is very slow to react in simulation mode, it was actually quicker to change power in ERG mode. My Drivo2 takes forever (13-18 seconds) to change power from 100W to 300W. Or back. When doing tabatta intervals, most of the interval time is spent in different power levels than prescribed.

[rederik1979]

Hi guys, I have buyed the XR two week ago. I follow a structured workout and the erg, for me, without a big experience,working good, but today I maked a workout with cadence and power target, es. 250w for three minutes at 53 rpm, so the power was every time low, the trainer never reached the expected power. Very bad.

[elite949]

In my Training Sessions in Zwift using the ERG-mode it is +-5Watts with some spikes +-10Watts.
I have experience with other trainers, but it seems OK to me.

It is essential to ride in a low gear. I ride on the small chain ring and the second biggest cog (25).