whitemax

I have one of my metrics set to gradient in order to see percentage of gradient. Problem is, it doesn't give me the instant percentage at any given point like my 800 did...seems to be delayed by 4 or 5 seconds for some reason which is frustrating. The 1000 is new to me...I've looked for a way to adjust it to get it do what I want to no avail. Anybody know?

JohnJ80

Every bike computer I know of computes grade by rise over run. So you have to go a distance before there is any data to calculate the grade therefore it cannot be instantaneous.

J.

whitemax

Perhaps I'm riding so fast over a 12% grade that I'm well past it when I get the reading (no, that surely isn't it )
The 800 I had gave a pretty quick reading so I knew pretty much was a gradient was while I was still on it. With the 1000, I'm well past it so it's hard to tell what's what. Kind of defeats the purpose

whitemax

I may have found a solution. I went into settings, then system, then data recording, then changed recording interval from smart to 1 sec. I don't know what "smart" relative to data recording means and it's raining so I can't test it out at the moment to see if it gives gradient readings quicker, we'll see.

Trsnrtr

I think smart is 2 seconds.

whitemax

I was thinking that perhaps "smart" meant that the unit was set to perhaps 3 seconds in order to save some battery (though you might be right about 2 seconds, I haven't found anything about what "smart" means at this point). In that case, setting it to one sec. as I did may give at least a little quicker reading on the gradient. Anybody know more about smart, please chime in.

NealH

The "Smart" mode means that when data is not changing much, it slows down the data record time - maybe every two seconds, maybe three. If you start accelerating for example, it will read progressively faster - down to a 1 second interval. The Smart mode supposedly saves a little battery life but in my experience, I tried it extensively both ways and never really could tell any difference. Therefore I always set mine to 1 second.

Even on the 800, with the data record mode set to "1 second", there is often a 1, or maybe even 2 second delay on gradient. When gradient is pretty constant, of course you don't see any delay. A 1 second delay is no big deal. How fast you ride a hill with changing gradients will also have an impact on the perceived delay. But its really not something important. I now have the 1000 and have not bothered to watch this to see if it's any different from my recent 800.

gregf83 

'Smart' recording was conceived when memory was expensive to conserve storage. It only stores a point when you move a certain distance. If you are moving slowly the points will be recorded less frequently.

If you have a powermeter the recording interval is automatically set to 1 Sec. There's no reason to use anything other than 1 Sec recording.

Silvercivic27

Every second, psh...Pioneer Power Meter records force and direction data from each side 12 times per revolution of the crank. Now that's fast.

JohnJ80

Actually, that would probably help reduce the delay (in time) because you're covering the "rise" and "run" distances faster making the calculation easier. Otherwise, quick calculations are almost certainly going to be error prone due to sensor noise - you get a number it just isn't an accurate one.

J.

whitemax

That's just it, on a 12 percent hill, I surely wasn't going all that fast. Seems like it took forever to read 12% and then only after I was well past it. The 800 I had before got to it much quicker. With the 1000, I am starting to wonder if it would even register the max gradient on say a roller where I could get over the hill much quicker.

JohnJ80

Either you're not understanding me, or I'm not understanding you.

The way grade is calculated is by moving a distance horizontally (i.e. wheel rotations or GPS) and a change in air pressure with the barometric sensor indicating you are going up. So presuming the barometric measurement is largely instantaneous (although subject to a lot of noise in the data), the faster you traverse the horizontal distance the faster you can get the measurement. Inaccuracy comes because of the "noise" in the data, so the readings would benefit by having some sort of averaging or selection of proper data points. All that takes time, hence the lag.

The time lag decreases if you are going faster even though the distance to measure the horizontal distance stays the same, you just cover it faster. Because you have to go the horizontal and vertical distances, within the tolerance of the sensors, it's going to take time and grade will not ever be an instantaneous measurement using these types of sensors.

If the software in the unit does less data selection (averaging) and puts data up faster it may or may not be correct. So you can get a less accurate reading faster or a more accurate reading slower. I suppose unless someone actually measured the grade (i.e. other means, say, surveying), we don't really know which one is more accurate. If you've done that, I suppose, then we'd be able to answer that question.

J.

Seattle Forrest

It stores a point when your direction or speed changes enough to need a new point to show the change. Could be every second if you're going through switchbacks, or every few minutes in a time trial.

"Smart" recording probably uses more battery power than 1 second because the CPU has to work so much harder to figure out when a new point is needed. It means fewer writes but writing to a solid state device doesn't take much power. Like you said, it's purely about file size. Also should not affect stats during the ride, only the way they're saved.

njkayaker

The units are doing all sorts of things with data that the extra effort to determine whether to write a point might be an extra compare or two. Calculating the smart recording might not cost much extra power at all. It might even be less than writing the data to the external memory. Even if there was a difference between the power of the two, it's not likely that the difference between these operations would have any real impact on battery life.

It seems the only reason to use the "smart recording" is to have smaller file sizes.

whitemax

I'm understanding you to say that the faster one goes over the grade, the faster the read would be....got it. On the 12% grade I referenced, I of course wasn't going all that fast, perhaps 9 mph as I really wasn't warmed up and had to start from a dead start given the hill came at the end of a road. I was well past the steepest section before I got the reading. I used to get it much quicker on the 800 (which I lost) and know this hill to be 12% at it's steepest using it.

JohnJ80

The other point I was trying to make is that the accuracy is suspect if the only way you know the grade if from using a bike computer. I have a couple, and they tend to agree in a broad sort of way. At the time I get the reading on my computer is not necessarily at the same point in the ride (i.e. over the same terrain) AND it varies from computer to computer. Based not that, we still don't know that it's a 12% grade just because your 800 said it was. About all we do know is that it is (a) a different sensor than you have in your new computer (maybe even a different component part number) which will behave differently than the sensor you are using now and (b) the 800 has a different algorithm to calculate it than the computer you are using now.

Barometric altitude sensor are affected by air pressure changes. So wind, terrain, cars passing etc... all can have impacts and can either introduce noise into the data or be inaccurate and I presume that Garmin is filtering this with software (i.e. different algorithms). Depending on how short the pitch is and other factors is going to determine how this actually shows after filtering/averaging occurs.

Hopefully each device is repeatable even though the "answers" are almost certainly going to be different.

njkayaker

I don't think Garmin is doing much to filter the barometric data. Garmin doesn't even adjust the barometric reading for temperature (from a Garmin user support answer that I'm not going to go look for).

Anyway, the things that affect the barometer appear to act over a much longer time period than the grade is being measured.

It seems that these things act slow enough not to affect the elevation gain very much.

I have no idea why newer units would be slower than the 800 to react to changes in grade. My 800 is somewhat laggy with grade measurements. That might be due to trying to get a moving average to make the grade number more stable (a moving average is some sort of filtering but it's a very basic kind of filtering).

As some one pointed out above, the units measure "rise over run" (barometric change over horizontal distance), which means it measures the grade of a position after you've passed it.

JohnJ80

That's sort of what I'm in the process of trying to figure out. I recently bought a Wahoo Elemnt and have been using that. The grade measurement there seems to be much more unstable in that it changes faster and especially at low angles of grade. In some cases, on a 2% grade yesterday (estimate) riding on the shoulder of a large 55mph highway, it wold to from around 2% to zero when a large semi would pass at speed (at least that's my current theory). I noticed later in the ride in places that were more consistent (say more open, so no pressure changes as wind passes over a sharp ridge, for example), the grade reading was much more consistent. So I think they do some filtering both in hardware around the sensor and in software with some sort of a low order difference equation. But what is interesting, is both my Elemnt and my Edge 1000 seem to be repeatable over the same terrain, with the exception of on the highway shoulder, but they don't agree with any precision.

I also think that at low angles over the relatively short distances that they would need to measure horizontal run to get a result in a timely fashion, the delta to compute the rise is going to be so small that it just has to be inherently a "noisy" signal from the barometric sensor. So if its going to be inaccurate, it's likely most inaccurate over short rises and with low angles. I'd guess any short rise is not going to yield a great answer. Long rises on consistent grades should work the best.

I think they would *have* to do some level of filtering even though rudimentary. That all said, and also because of how they are trying to figure out grade with a barometric sensor over a horizontal distance, this measurement is not going to be particularly accurate and certainly probably not timely.

I suppose there are other ways this could be done where you would use terrain data with a GPS fix to get to a closer fix on the vertical rise after knowing something about the terrain. This would probably (within the quality of the map data) help the software rule out extraneous data that is "out of bounds." That might be one way this could be done but I doubt anyone is using that and I'm not sure the maps in the units are that good either.

So, the conclusion I've come to is that grade is not a great measurement to be looking at real time and given that we do care if we are on a 2.5% vs 3% grade from an effort perspective, it's still only a pretty tiny difference in actual angle but which would still be difficult to measure.

J.

njkayaker

Yes.

It seems quite likely they are using a moving average (or something like that). That's a "rudimentary filter".

I believe the terrain data doesn't have the resolution necessary to be able to do this. The 820/1000/Touring are shipped with terrain data (it's in the "DEM" map). Doing something might cause a small impact on battery life (and it's another place to introduce bugs) for not that important a number.

I think the basic problem with grade is the same basic problem with elevation gain. People expect the measurement to be "perfect" when it's really only an estimate.

GPS data is so good that people can expect that every measurement will be equally as good (that isn't necessarily going to be true).

This. I only start to care (a little) with much higher grades. Where I've found it useful is in places that don't seem to have much grade but are hard.

JohnJ80

Yep, we agree. With current sensor sets, grade is kind of a lousy measurement. We'd need a different instantaneous way to measure grade (tilt) that was accurate to a fraction of a degree and did not need calibration (i.e. pitch/angle sensor). Pretty much means that grade on a bike computer is not going to be great at all and - at best - a figure of merit and after the fact.

What's interesting though, is that if the bike computer just threw a number up on the display, even though it was largely inaccurate, users might find that to be more acceptable. It's the time lag that seems to be the most irritating to most people not the lack of accuracy.

J.

njkayaker

Outside of excessive lag, I don't think there's​ any indication that it's a "lousy measurement".

"Fraction of a degree" seems overly precise.

People notice lag. It's a normal thing.

They have no real way to notice inaccuracies (unless they are really obvious).

If the 1000 is more laggy than the old 800, it had better be more accurate!

JohnJ80

Yes. This whole exercise got me thinking about grade. Never really paid attention to it before. But I'm glad I did so I know what to expect of the grade measurement (or how little to expect of it, actually). I'm also going to pay some attention to the repeatability of the measurement. Be nice if the measurement, while probably inaccurate, was at least repeatably inaccurate.

J.

njkayaker

You can get an inclinometer and strap it to you bike (if you are so inclined).

The Garmins get grade as a byproduct of other uses. Specifically targeting measuring grade means increasing the cost (or the size) of the units.

springs

My preference is for bike computers to have a pitch sensor like most cell phones have so that you can have instantaneous pitch (grade) readings in addition to lagging rise/run calculations. Pitch sensors need to be calibrated of course, so your computer mount can't change position during a ride unless you recalibrate. Any guesses on what the combined tolerance of a barometric altimeter + gps distance over a short distance is? That is all your pitch calibration + sensor tolerance needs to be within to be at least as accurate as rise/run.

So a pitch sensor isn't perfect but I'd like it as an additional grade sensor on my bike computer.

njkayaker

Yes I know about those.

Pitch sensors are required for measurements for devices that don't move distances.

It wouldn't be useful for elevation gain.

It might not be that much better than what is already being done.

It's another thing that consumes power.

Currently, you get grade "free" with elevation gain.

What are you willing to give up? Size? Cost? Battery life?

Is it worth Garmin supporting it if every user has to calibrate it?