Trainer Accuracy - Calibration - Software Bugs - Real Turbo Muin B+

[ChrisH]

Hi All,

I've owned a Real Turbo Muin B+ for a few years now but like many, it's had more use this year than ever before. Because of this I have wanted to make the trainer as accurate as possible. As detailed experience of this is not very often communicated - there are more frustrations than successes - I thought it would be helpful to post my results. Also though... I have found several key bugs in the software that hopefully Elite will pick up on to prevent frustration for others!

WARNING - LONG AND DETAILED POST TO FOLLOW - BUT I PROMISE IF YOU ARE INTERESTED IN THIS SUBJECT IT WILL PROBABLY BE ONE OF THE MOST HELPFUL YOU EVER READ

General Summary

Trainer initially setup with Ant+ connection then later Bluetooth to a Windows 10 PC. Software used includes My eTraining, The Sufferfest and Zwift. Bike is fitted with Garmin Vector 3 pedals (recording in parallel during tests to a Garmin 530 head unit) which I have targeted my calibration against - i.e. I wanted to get them as aligned as possible.

Belt Tension

First a primer on this. Does changing the belt tension affect the resistance of the unit and therefore the power required to drive the trainer at a given resistance. Yes. Should this be your primary method of calibration. Probably not. Like many I experimented with this a lot initially as I was struggling with the software calibration, but knowing what I know now... read on... really this should be set at the factory spec because as long as it is in tolerance, it shouldn't particularly effect your ability to accurately calibrate the trainer. Personally I needed to get this 'reset' because I had to strip down my trainer previously - you can search another thread on here - so I settled with a torque of just less than 2nm on the tensioning bolt. This also 'felt' about right providing enough tension to ensure the belt wasn't really at any risk of slipping whilst also not placing excessive strain on the pulleys. Note that subjectively, my personal logic would be that it would be best to run the minimum belt tension you can whilst avoiding slippage... any tension above this is just going to equal more stress on everything which in turn is going to shorten the life of your trainer. Ergo... if you are tightening and tightening and tightening... somethings probably wrong.

Calibration Routine

I would definitely suggest running the calibration routine to start with to get you in the right ball park. The calibration routine observes the power at resistance step 2 and 6 of 14 (on the Real Turbo Muin B+ at least) at a 'speed' of 30 kmph. Note that saying the speed is 30 kmph is probably actually a specific internal flywheel speed but one which translates to 30 kmph using the default wheel circumfrances... which for base calibration purposes... need not be changed in my opinion (especially as end apps like The Sufferfest and Zwift ignore the reported trainer 'speed' anyway). For my purposes, my factory calibrated values were 132 and 342 and then following the re-build and tension as per above, the result of the first calibration test was 141 and 316. In the long run this proved to be pretty close actually but read on. First, two specific observations here:

1 - you need to pedal smooth. A true power meter will bob around quite a lot... the power reading will jump up and down by maybe a 20w range in my experience hence we often use a 3 sec or 10 sec or even 30 sec average. The calibration software seems very sensitive to this... it will hold you on the step i.e. PA... until it sees a relatively steady power reading. If you're not smooth and your power meter is outputting all over the place... you'll never progress. I found this easy enough when setting PA... it's harder when setting PB and smaller / lighter / less powerful riders will probably really struggle (my FTP is around my PB value by coincidence... if yours was well below... I think you'd be a tad stuffed unless it got a reading very quickly). So anyway, pedal smooth!

2 - I think the values you get from this relative to your original sticker values are a good indicator of whether your belt tension is where it should be. If I was way lower or higher on PA (like maybe 30w or more away from the factory) then I think I would be tempted to adjust the belt tension before continuing to get it a little closer to stock. Tightening increases resistance so will increases the power required hence higher number... loosening will decrease resistance hence will lower your effective calibration value. In my case... my calibrated numbers were inside the factory ones... PA was a bit higher but PB quite a bit lower... so I was happy I was nicely in range. Plus I torqued the belt with a pretty decent torque wrench to begin with.

Software Bugs

So you're probably thinking that's great... very obvious. You will have either done that and been successful, happy, and probably not reading this... or if you are reading this you're probably thinking... yeah... I've done that... but it doesn't *$$*** work! Well that's where I was for a long time.

I tested soooo many times making small and huge manual changes to PA and PB values to observe what difference it made in actual use - i.e. with Zwift, with The Sufferfest - and I constantly found that regardless of the values I set, it seemed to make no difference to the wattages reported in those apps. What I now know is that there are two major bugs - Elite - please observe this and get them fixed because it is a nightmare when you don't know what is going on!

Bug 1

Bug 1 for me is that any write of PA and PB values from my Windows 10 computer, whether connected over Ant+ or Bluetooth, whether done as a manual update or at the end of the calibration routine, just plain old doesn't work. The software reports a successful write, it says 25%, 50%, 75%, 100% / Operation Complete... but it's lying. No update is made. I was able to prove this by using an older version of the My eTraining software which allowed you to 'Read Trainer Map'. This meant I could download the full calibration map and from that I could see that no update was being made! Regardless of what was entered... I always just had the default map hence, no change in an app.

The solution? I found that applying the PA and PB values manually, from my Android phone connected by Bluetooth, worked! The first time I did this and observed the results was such a breakthrough, it was immediately obvious that a change had occurred and it gave me hope for dialling in the accuracy of the trainer. The issue with the Windows bug above is massive, especially as it inaccurately reports as a success. Basically this in my opinion is why so many people post that the calibration routine 'doesn't work'... because regardless of an accurate detection of PA and PB values... the software fails to actually upload them. So rule 1, always use a phone connected by Bluetooth, for me at least this worked... mostly...

Bug 2

Bug 2 is more nuanced but also potentially very very confusing. This really threw me off track with dialling in my trainer. To cut to the chase, I have found that if either PA or PB is one of the default values, then the default MAP just gets loaded. So in my case if I tried to load PA 132 and PB 300... the fact that PA 132 is the default value means the trainer automatically adopts the default map and thus my request for PB to be 300 is ignored and it sets it to in my case, 342. This then makes such a massive change to the power curve that if it happens without you knowing... you just go round in one big confused circle because all of a sudden you go from being close to perfectly calibrated to being miles away even if you only changed PA by 1w but were just unfortunate enough to land on that default value. So Elite... that is 2 pretty major calibration bugs you need to fix... no wonder so many people struggle because the combined impact of both would drive most people totally insane before they managed to understand them in the way I eventually did.

Ending on a Positive

Once I understood all this and adopted a rigorous routine of loading calibration values on my phone, downloading the maps to confirm using an old software version on a laptop, then testing using defined training routines - mostly in The Sufferfest but it could be Zwift or any software I suppose - I was able to achieve a really good result. How good? Well I have tracked average power over a 30 min interval workout to within 1w with an average power of 230w and a NP of 270w. So that's less than a 0.5% discrepancy which I think is damn good. So... it goes to show... you can teach an old dog new tricks. This can be a very accurate trainer... but I would say it took me months to fully understand this... and even once I started to (the point I started downloading the maps using the older software) I still went through 21 different tests and training programmes - probably at 1 hour+ per test) to get to this point. But, hopefully with the above bugs identified... others will be able to achieve this with far less... err... suffering.

If you're interested here is the final result: https://analyze.dcrainmaker.com/#/publi ... 74e6c0e3e1

Not bad eh?

Signal Drop Outs

A quick word on signal... I believe it is imperative you have a really solid signal. I found Bluetooth much better for me but I ended using a dedicated powered extension lead and a new 5.0 spec receiver to be sure it was as good as it could be. Whilst Ant+ seemed ok... I could see I got some drop outs at times. This just made testing a real lottery because different software handled the drop outs differently. If you were shooting for +-10% then maybe this wouldn't matter... but I was always aiming for 'perfect' or as near damn it as I could get... and to do this... you NEED perfect signal. So make sure you have this to start. Otherwise you won't be comparing Apples to Apples so to speak.

Using the Trainer

As a final tip... which probably applies to all trainers but especially the Muin... it is very 'speed' / 'cadence' sensitive in ERG mode. Because the flywheel speed has such a profound effect on resistance and power... if you're all over the place with your cadence... it will constantly be 'hunting' to set your power correctly. The smoother you are... the better the trainer can be... and the better your results will be. To be honest... pedalling smoothly is efficient and good practice... so I don't see this as a limitation really... I'm sure its 'better' on newer more expensive trainers. but this is just an observation.

A similar observation here is that because of this... more than likely you'll only achieve 'perfect' in a finite ish gear ratio range. I haven't fully tested this but I know if I move to extreme ratios and push more towards the edges of the trainers 'map'... it's probably going to lose accuracy. I don't think this matters too much... the key is calibrate it in a sensible gear... probably the main one you will train in. So think straight chain line... I chose big ring... it works well for me.

Conclusion

I hope that helps. Elite... I kinda feel like you owe me a brand new Directo XR for my time here... so if you did want to 'sponsor' me for further product and software testing please let me know. I have all the data for this recorded (as I say, probably months of testing, all logged with comparisons of every stage on DCR etc)... but anyway... regardless of that... I hope this helps people out. Obviously it very definitely applies to Real Turbo Muin B+ users... but probably it will also apply to many other trainers as well.

Good luck. Good training.