What about Gradient Lag?

krazyeone said:

My feeling is that Filtering and calculation is done every 20-30 seconds

I think - if it would be like that, an update on the display would also be only every 20-30 sec. But it updates much more often - but with a terrible delay.
As Phil70 says: it‘s all about delayed reading of barometric elevation.

Not if it’s a “moving average”.  
3 factors come into play with this

1) sample rate - how often a pressure reading is taken to calculate gradient change. Every 0.1s,0.5s, 5s or what…

2) averaging rate - over how long a time are samples being taken, the average calculated and a result output. I.e over the last 5 seconds, 10s or what… this is effectively smoothing out rapid changes that could occur.  Also how often is this calculation done.  It must at least be a factor larger than 1).   So if your sample rate is every 1s you can’t average less than this and if you calculate every 2s you don’t get much smoothing  

3) display update frequency.  How often does the display update.  I.e taking the result from 2) and updating. I doubt this is an issue. 

The problem with calculating gradient using pressure is that it relies on high accuracy pressure readings and high stability. The change  in pressure is so small when going up a hill on a bike that it needs a very sensitive device to measure the change, which in the uses to calculate the height difference.  Give the distance traveled is also known and there you go - gradient (good old tangent calculation). Now if that pressure measurement isn’t accurate enough or stable then the readings could be all over the place. So it does need some form of smoothing. 

So here’s the problem.  You have either a fast reacting display that is most likely not that accurate.  But you think it is as it changes fast.  Or you have a slow average that is more accurate on long steady climbs of constant gradient (like a big Alps climb), that doesn’t react well at the start and end of climbs, or ones with a lot of short gradient changes.  Or you try and find some place in the middle.  

It looks like the real answer here for Garmin is to look hard at sample rate and the averaging calculation and see if it could be brought a bit more towards the fast/inaccurate area.  I think that would probably please a lot more people.  I personally am not so bothered if the gradient says 5% when it could actually be 4-6%. But people do need to realise your not going to get it down to 4.8-5.2% and it changing every second.

Yup. Let’s hope DC Rainmaker does the same. To be honest I’m surprised the pair of them have said nothing before now. 

Maybe now garmin might listen. Fingers crossed

Do you think after realizing this issue, this is fixable with a software update or is this a hardware issue?

For me it is about 20-25 sec delay, was not on my 1030 and just this week I bought a new Wahoo Bolt 2, put them both on the bike and rode around to compare and there was a pretty significant difference as you might expect. Going up hills, the Wahoo reflected the hill in their gradient field in about 5-7 seconds while 1040, we'll you know about 25 sec later

Do you made movie picture etc?

no because I did not think it valuable to compare it to Wahoo. 

I thought the better thing would be to compare it to the 1030 which is their own device to show what has badly changed.

I sold my 1030 so cannot do a movie but it seems they know and are not questioning it.

You can send me PM :)

I was interested to buy one wahoo v2 as I don't like training page from Garmin

I think its hard to say.  Garmin changed it for a reason. Was it a new type of pressure sensor?  Was it to try and save battery life.  I don’t doubt something appears to have made them change the software.  It maybe they can improve the situation, but not to the same state as a 1030.  You might find that although the 1030 updated faster the value it gave was not very good and this why why Garmin tried new hardware out.  Maybe that was too sensitive and they found they had to smooth it out more to stop it going all over the place. Who knows except Garmin?

it may be with this type of measurement (at this price point) your choice is

1) fast response to gradient change but inaccurate reading (inaccurate barometer)

2) fast response to pressure changes (gradient varying a lot on what appears to be a steady gradient) 

3) accurate gradient, but delay 

Anyone ever questioned the actual value displayed across devices on a long and even gradient?  Might find that an interesting viewpoint also. Your fast acting wahoo or 1000 device may be fast, but is it displaying 8% when reality is 10%?   I do know when riding with someone and we both had either a 1000 and then a 1030 that whenever we called out gradients being displayed when riding together more often than not it was between 1-3% different. 

I think what makes the gradient lag so annoying on rolling terrain is that so obviously wrong, you know you are descending and it shows your climbing. On more hilly routes it's difficult to tell if it's 6% or 8% but you know your climbing so not so irritating.