# Tacx FLUX 2 Direct Drive Smart Trainer // GPLama Loop-Back Review

**Channel:** Shane Miller - GPLama
**Source:** https://www.youtube.com/watch?v=nJpmRLRiVL0
**Transcript page:** https://www.withtranscript.ai/video/nJpmRLRiVL0

## Chapters

- 0:00 — Loopback Review Format And Trainer Specs
- 3:24 — Setup, Assembly, And Ride Impressions
- 7:48 — Data Analysis Reveals Erg Issues
- 13:32 — Deeper Testing Across Workout Scenarios
- 20:33 — Final Verdict On The Flux 2

## Transcript

**[0:00]:** in the absence of any new smart trainer hitting the market at the moment and an increasing number of people holding onto their current smart trainers for a number of years i've decided to do what i'm calling loopback reviews on trainers that i may have reviewed in the past and some that i haven't now in these reviews i'll be looking at the current state of play with these trainers and that'll cover an overview of the trainer and the hardware specifications any updates that have been made since release to either the firmware the hardware support or the management applications i'll be running a llama lab test or two to see how they hold up with that current firmware and i'll be making other observations over a number of hours using these trainers in the llama lab follow-up reviews like this may be useful for existing owners of the smart trainers that i'm looking at to see how

**[0:40]:** they stack up in the llama lab with the latest updates maybe if you're buying second hand from someone who's selling their smart trainer to understand what you're getting into or maybe these videos just serve as a warning to keep well away from the trainer that's for sale on that facebook ad first cab off the rank the tax flux 2 smart now this training was first announced in 2018 it was on sale in 2019 and in early 2020 it was silently refreshed by tax but retained the name flux 2. so it's kind of hard to tell which ones which the newer version is often referred to as the tax flux 2.1 and that's the version that is on the shelf if you were to purchase new at this point in time the updates to that model compared to the one that i'm reviewing today is that it has native through axle support so you have the throax on your bike and you put it that's a pen but you

**[1:28]:** know what i'm talking about it doesn't use this hokie quick release adapter which is the one that i have and it has some redesigned internals for it to have better resistance control and you'll see in a moment in the llama lab tests why better resistance control is a necessary upgrade to this unit that i'm reviewing today quickly covering the technical specifications of this unit direct drive smart trainer so rear wheel comes off bike goes on compatibility is across the board quick release 130 135 3x142 and 148 with this hokey pokey adapter as i mentioned just before the flux 2.1 has a native through axle support which is a much better solution the free hub it comes with is the shimano sram hyperglide compatible freehub with no cassette supplies you need to supply your own no rear pivoting axle which is an attribute i guess we now have to look into for trainers which means no

**[2:12]:** elite riser compatibility or any other device that goes up and down on the front wireless protocols both amp plus and bluetooth you'll get fec and ftms compatibility with that so up to date with all the current standards on any platform the data you'll get over both amp plus and bluetooth from this trainer is power and cadence you'll get up to 2000 watts of power if you can crank that out power accuracy plus or minus 2.5 putting it outside the range of accuracy requirements for zwift pro level racing not unexpected for a mid-range trainer calibration is done via a spin down max gradient 16 flywheel 7.6 kilo which is a little heavier than the flux one and the flux s and that does have a big positive impact on the ride feel which i'll discuss a little later noise level the thing's quiet but not silent the power source is via a mains power with no

**[2:58]:** power pack required it's kind of handy you plug it into the wall into the unit and you are done no power packs to get confused with firmware upgradable via their app and price upon release us899 and in australia it was just under 1200 these have been seen on discount and including the updated version which is the flux 2.1 the latest firmware update for the original tax flux 2 was released in april 2020 and that's version 3.3.40

**[3:24]:** 1.1.6 and the updates within that update included ftms compatibility improved cadence power stability and power accuracy now one thing to note is the tax util app which we've always used in the past for tax trainers has been replaced by the tax training app you'll get a little notification if you're trying to use the tax util app to update your trainer you can skip that little warning box and continue on best not do that though i did try and upgrade the firmware with the older app and that failed dismally i needed to then jump over to the tax training app connect to the trainer wait for it to reboot and sort itself out and then do the firmware update that worked flawlessly so hot tip don't use the tax util app anymore it's the tax training app quickly flying through the unboxing and the assembly of the tax flux 2 comes in two separate components in

**[4:12]:** the box the legs and the core unit itself

**[4:26]:** just the two bolts required on the underside of the unit and the hex wrench or allen key is supplied with the unit

**[4:42]:** cassette on bike on with the quick release configuration and onto a few quick sound tests

**[4:51]:** with no fans on

**[4:56]:** and there's a little bit of resonance in this room it's always difficult to capture that exact sound that you'll hear in the room yourself on camera as a reference here's a gear change you can hear the gear changes quite clearly above and beyond that trainer and the free hub noise so not silent but definitely not a loud trainer either now onto how the trainer rides in sim mode on zwift and it's one thing i will point out that this trainer does very very well when connected via bluetooth now here i am on titan's grove which is known for its up and down roller coaster terrain and the experience on this was absolutely brilliant using bluetooth now why i say using bluetooth only because without plus fec it was a little laggier with these sim gradients and you'll see here as i roll down and as soon as that gradient kicks up a little bit the

**[5:52]:** resistance kicks in and the power numbers go straight up no delay no lag again over the top of this resistance let's go right here and then kicks in right there again as the gradient goes up on the road now if you're using app plus fec it's a little slower a little laggier on the road so you'll be already into the transition on the tarmac before you feel that resistance kick back so full credit to this trainer for doing very very well in sim mode over bluetooth and with that in conjunction with the larger flywheel and the better gearing feel of this trainer compared to the flux one and the flux s this gets a pretty high rating in my books for sim mode within zwift

**[6:39]:** jumping now into erg mode with attacks flux 2 and you can see along the bottom there there's one interval that i've performed for five minutes at 270 set point but it doesn't quite sit on 270 for the entire interval very strange i'll dive more into this in the data analysis in a moment but the interval that i'm performing right now on screen is 300 watts and it's doing pretty well sitting on 300 watts and the data recorded on the bike whilst different to what was being reported by the flux 2 was quite consistent so it held a stable set point was it the correct set point well it really depend on what sign i was in and i'll dive more into that again in a moment and finally back to sim mode for a quick sprint test to check one the stability of the unit and two does the flywheel spin out under really heavy

**[7:30]:** load

**[7:36]:** although that was a short sprint it does get a pass in my book it pushed back a lot of resistance and i did not spin that thing out i just didn't have enough legs to keep turning it over at those watts

**[7:48]:** now to dive into the data analysis side of things with the flux 2 up against the asean meduros and the power to max ngco over seven hours of riding indoors this week using both zwift and trainer road as the trainer control now what i'm looking for in this testing and analysis is is the trainer accurate within manufacturer's specifications give or take a little bit at very very slow flywheel speeds and very very high flywheel speeds where most trainers really do fall apart for accuracy i also look at the trainer's ability to hold erg mode set point watts and that's separate from accuracy what i mean by that is if i set the train to 300 watts can it hold 300 watts or what it thinks 300 watts is accuracy is a separate thing which is on top of that again literally with this trainer it is on top of that when it comes to accuracy

**[8:33]:** you'll see what i mean in a moment is the trainer able to hold that set point stable give or take some human variability in your legs or is it all over the shop so i look at that and we also look at cadence if the trainer reports that so they're the areas i dive into let's jump into it right now and this is a rabbit hole and here we are on my favorite website on the internet the dcr analyzer tool where we can compare multiple power meters and other things as an overlay and see how they stack up things are looking pretty good with the initial alarm at lab test the first 10 minutes let's throw that out that's blowing the cobwebs out this trainer had been on the shelf in the llama garage for over two years before i looked back to this one and there's a spin down calibration right here which really

**[9:13]:** threw the numbers out so let's discard that we have the flux too up against the asiana duos upon visual inspection everything looks pretty good let's dive in not bad what do we got power wise we have 222 versus 226 okay give or take a little bit of drivetrain loss there happy days until we look a little bit deeper the 200 watt zone the trainer had no problems at all holding 200 watts in fact it was probably 198 199 but i'm really splitting hands for that speaking of splitting hairs that aren't as fine as the top of my head uh here this zone here now i'm diving a little closer so the scaling is out but this set point was 250 watts it's averaging 239.5 the pedal is reading 242 which i'm pretty happy with give or take 2.5 inaccuracy um something's up it's not able to hold that 250 watt set point in erg but

**[10:06]:** not only is it not able to hold that it's not even reporting 250 watts set to 250 it's doing 240 and reporting 240. it has some behavioral issues in erg early on in the sprint pulling up that information here and it acts as you'd expect for a mid-range smartphone the pedals are super responsive they're connected to my feet they're responding like that uh peak power a little bit lower on the trainer it does come up after a little bit but overall the sprint i'd probably give that a pass given it's a mid-range trainer and how it measures power so not too bad in the sprints but not really something you'll be wanting to do pro level races with on zwift into the overs and unders 20 seconds on 20 seconds off um again only you up against the assignment duos astronomer duo's again a little spikier because they're measuring almost instantly the power i'm putting down on the pedals the trainer taking a little bit to come up and still reporting a little bit under power wise there so

**[11:02]:** especially kicking up to 450 watts it does apply the resistance probably a little bit too much because i am kicking over 450 for that on average um again not too bad for a mid-level trainer but the higher we get with power zones the further it's out with power accuracy next up is onto the flywheel speed test now this is an interesting one this part of the llama lab test is designed for trainers to fail and sometimes that's okay the first test here is with a very very slow flywheel speed so this bike has a 3928 on the back so very very slow flywheel speed the set point for this section through here is 225 watts for the entire four and a half minutes and things do change based on the flywheel speed so early on it sort of dips down and comes back up as it stabilizes over the 90 seconds with the astronaut duos

**[11:48]:** and we have a split there of around 15 watts or so through this section here the trainer believes it's doing 225 watts it's reporting 225 watts so that's a good thing but the actual power it's making me do is a little higher spinning up the flywheel speed a little higher again 53 probably 21 on the back i think and again that separation is still there but the trainer still thinks it's holding 225. um still not the best and then finally spinning the trainer right up into the 5311 and the set point goes out the window it's uh set to 225 and it's reporting 269 270 and the power that i'm actually doing on the pedals is a little lower now at that speed there's a lot of vibrations happening if it's just a little bit of imbalance in the flywheel so there's vibrations happening through the bike it's chaotic it's really

**[12:39]:** an edge case for using a trainer you should never use erg mode in the 5311 things don't look too pretty but it's not too bad it's not 50 to 60 to 70 watts out anyway finally into a short little kick which i guess replicates an attack in a race and the trainer again reporting just a little bit under by a little bit here we're looking at what nearly 45 to 50 watts under right where it matters right when you're going really hard you just need to hold that wheel the train is not reporting what you're really doing on the pedals quickly jumping to cadence and it is the only data set that i look into cadence because the storage is the same across the board it's not too bad if you're smooth on the pedals but it still jumps around a little bit overall average is 84 84 so it's spot on for overall averages

**[13:22]:** but sometimes let's just say the high flywheel speed test the cadence goes out the window and there's a few jumps up and down that i really wasn't doing on the pedals onto llama lab test number two and things

**[13:32]:** look okay but when we dive into the details of my actual experience on the bike there's a few wtf's that had me wtfing starting off with some coco cadence rides into the b plus ride which i believe is bowie on the pedal is 253 260 261. so again the trainer a little bit lower if you do the math on that that probably comes out very close to 2.5 percent difference but again what i'm having to do there is go to the extreme of 2.5 oh it's lower because it's 2.5 and the power meters could be here on average things should line up pretty well um it's just a common occurrence that we always have to go to these extremes to make excuses for trainers reading under do we expect it to read under a little bit that much what a coincidence that it's almost on that borderline anyhow a little lower was i impressed not really not really into the llama lab test

**[14:24]:** again with this and this is with two power meters on the bike not looking too bad numbers good still unable to hold that set point wattage at 250 watts it dropped in at about 240 on average all the way along until and let me back back out a little further now if i zoom in at a different level you'll see what i'm talking about here had a bit of an x-files moment right there at the end it jumps up to the set point of 250 and exceeds it by a little bit it was reporting about 255 so for the last minute of that erg session after spending around nine minutes at 240 with a set point at 250 it jumps to 255 and that's real power that's what i was doing you see both power meters respond to that and you see the trainer reporting at about 255.

**[15:10]:** i have absolutely no idea why it did that diving back through the zwift logs there was only ever one ftms command sent to the trainer and that was do 250 erg but something happened right about here that flipped it up into another mode it's not the only time that happened either the sprint on this one again not too bad the fervero aseam is winning that competition with the spike there at the start and they all averaged it out not too bad so you can see why in the previous llama lab test i gave it a pass or an ok ish on the sprint that again pretty much what you expect for one of these trainers jumping in again to the over and unders with two power meters on the bike this time same story as before the power meters on the bike reading a little higher a little more responsive and the trainer just being

**[15:53]:** a little bit laggier especially in the 350 and the 450 watt zones down there in the 150s the thing just hums along nice it's quite happy but in the higher power zones the trainer accuracy just starts to go out the window a little bit which leads me with excitement to the next three data sets and i do apologize for going deep down the rabbit hole i think this is interesting though of how this trainer actually responds or doesn't in erg mode at certain zones first 10 minutes let's forget about that let's warm up into the first two intervals here not too bad flux two up against the powder max and g co 226 228 call it a day quite smooth the second one was a little rougher it wasn't able to hold the exact set point i was after but it sort of got me into the ballpark zone for the warm up to get

**[16:38]:** me to the 300 watt zones and that's where things fell apart with this test you can see straight here flux too up against the powdermax ngco not quite hitting the power accuracy that is required so 300 watts for the trainer itself it was reporting 300 it was set to 300 was reporting 300 it was actually doing around 309 to 315 so at 300 watts this trainer appeared to be around 12 to 14 watts out on average uh the spin down was performed so i'm not going to continue doing spin downs after every single interval so it is what it is it shouldn't vary too much at all again what we're seeing is pretty good at 350 zones and not pretty good at the 300 zones so i soldiered on through that session doing more power than was actually being recorded on zwift because i had the trainer as control and surprise surprise the training completely

**[17:26]:** blew up and stopped working and you'll see that right here no amp plus no bluetooth lights flashing on the unit nothing connected power meter on the bike working just fine um i don't know what happened there was no smoke there was no smell so i didn't completely blow it up i had to power cycle the unit off power it back on and then things came back alive so strike one for unhappy llama for that the trainer just failed to work after that just riding along just running along and so on average overall at that lower zone not too bad after that but definitely not a pass market the 300 watt erg and definitely not a pass mark with the trainer just failing to transmit bluetooth plus for whatever reason almost done we're on to data set number four five by fives with an extra interval at the end because i was angry and i was

**[18:13]:** doing some power match testing again with the power and the smart controller paired to the flux too everything's looking good everything's looking good kind of wasn't quite able to hold the set point of 270 until about halfway through this interval where it did that magic step up again but no need to step up it stepped further the unit then held 280 or 279 watts for 2.5 minutes of that 5 minute interval again going back to the zwift logs the only command sent to the trainer was do erg at 270. this trainer with its behavioral problems in erg decided to do its own thing for the rest of that interval after the halfway mark hmm down to the 150s all looking good up into the 300s exactly what we saw before i'll dive in a little closer to all of these here we go and the power offset about the same again so it thought it was doing 300 it wasn't doing 300 it was doing around 300 and well 319 320 for that section through there and around 15 to

**[19:13]:** 20 watts too low on the train which made for a pretty hard erg session given the target zone was 300 for my training second 300 watt interval and a dropout boo for that and the same power difference between what was happening on the bike what i was actually doing and what the trainer was reporting now before the next five minute interval you see on the screen with another power dropout there i stopped and did power matching on zwift to the power to max that was now the power source with the flux 2 being the controllable trainer so the power accuracy just shouldn't matter and it didn't it brought it down to 300 watts on the pedals or on the spider i guess that's where the power max is and it was all happy days and as expected the trainer reporting around 12 to 15 watts lower than 300 watts which is consistent with what it's

**[19:58]:** doing in that zone okay jumping back out and the last interval there that i did was power matching with the asean meduo so this trainer using power match with the power meter on the bike works quite well quite smooth power numbers it's reporting though just something else and finally a short sprint and some just riding along after that workout are done and it kind of missed the peak on that but that was a very very short sprint so i won't deduct too many points from that so that's four data sets indicating this trainer that i had in the llama lab had behavioral issues within zwift was it erg was it ftms was it zwift's implementation

**[20:33]:** of those let's put that aside and load trainer road to find out i loaded the jefferson workout and i really don't even need to go through the data it's exactly the same thing as what we're seeing on so if that offset at about 300 watts was exactly the same and the lower zones it was pretty good so yeah can't quite put my finger on this trainer it's definitely not one of these just scale it up a little bit and problem solved as you can see there it's just inaccurate across different zones but good to know it wasn't isolated to just swift it's pretty much everything alrighty extracting ourselves from that rabbit hole of data which was i must admit only the tip of the iceberg of what i was digging into with that whole seven hours of data and looking at the log files and the conclusion was it wasn't really that great but on

**[21:16]:** the pros and cons of this training because there were some pros i did have to search high and low for these days but the ride feel is good on the pedals turning the cranks over i could get my workouts done swapping from the say the sb 20 the kicker bike the kicker 5 the tax neo2t which i can get my workouts on no worries some trainers i just put them aside they just feel like absolute crap this was actually pretty good much better than the flux one much better than the flux so there's a tick there the unit is relatively quiet so you're not going to be waking up your neighbors or your small children sleeping in the next room it's great in sim and erg response mode with bluetooth as you saw early on on zwift using bluetooth those hills really kicked in really fast and without the three second averaging on screen it

**[22:00]:** was really punchy so that was a good thing and whilst i'm grabbing at straws of what's good with this unit it does work well with the power meter on the bike used for the power source but that's going to cost you more money and you may max out your apple tv bluetooth connections if you're pairing a separate trainer a separate power meter and a separate heart rate monitor game over if you're using the apple tv for that on the cons this is an easy list to get through the eggmo performance that was pretty horrible i wasn't hitting the setpoint wattage in a number of designs that i needed to it was smooth but it just wasn't accurate nor was it precise nor was it what i expected from this trainer the accuracy from the power itself just wasn't consistent the 100 200 wasn't too bad 250 out the window 300 don't even worry about it

**[22:44]:** it was well outside that 2.5 percent no guesses why these trainers aren't marked as pro level racing devices for zwift and plus and bluetooth crashing well what do i say trainers shouldn't just shut down requiring a reboot halfway through so that was a little concerning now i was connected via bluetooth and amp plus at the same time that's not really unexpected it's not really any different to how i treat other trainers this one had a mind of its own and decided to go to sleep through two of those sessions the legs don't fold in the unit is heavy and there's no handle to move it so it's kind of awkward and you possibly stub your toe on this if you're not packing it away in smaller areas now i don't expect perfection from mid-level trains but what i do expect is to at least match the specifications written on the side of the box and

**[23:28]:** the unit that i tested this week didn't do that i gave every opportunity to meet those expectations and those specifications across multiple rides in multiple tests using multiple software platforms it didn't it simply didn't and because of that it's very difficult to give this trainer a pass mark when it behaves as it did in the llama lab and on that low note i'm going to have to leave it there for today thanks for coming along for this one it has been a bit of fun looping back to this trainer which didn't quite meet my expectations and really didn't meet what was printed on the side of the box either is the 2.1 any better all indications are maybe not but if i can get a hold of one i'll put it in a loopback review thanks for watching

**[24:19]:** you