Old Specialized cyclocomp vs. new Cateye Astrale 8: interesting numbers for numbers geeks

[Doug]

Hey all,

I finally am replacing my incredibly tried and true 1993 (or so) vintage Specialized cyclocomputer
on my main road bike. I wanted cadence, plain and simple. The old Astrale was butt ugly, the new one
slick. That coupled with a 20% coupon forced me to "upgrade."

I mounted the Astrale 8 and rode a direct comparison ride. Sensors and magnets were on opposite
sides of the front wheel to eliminate crosstalk. Wheel sized programmed the same on both.

Not to my surprise, the comps didn't slew identically as speed changed. But to my bewilderment, the
old Specialized slewed faster! Not by a sneeze either. By a good second or fractionally more. The
comps didn't even agree on the top speed I reached, 21.2 vs. 21.4. That's a 1% difference, probably
within specs, although there are no written accuracy specs in the Astrale docs. 1% rated is a good
bet I presume.

The automatic timer mode, which starts and stops the clock as the wheel starts and stops, was much
better on the old Specialized. The Cateye started instantly, but it overran the stop by a consistent
3 seconds. The old Specialized started 2 secs late, but overran by the exact same time when the
wheel stopped. Hence it's timer was dead on, all according to short stopwatch tests.

Steady state speed readings and distance matched. The timer error on the Astrale messed up the
average speed calculation on my short ride. How much so on the real 60 mile rides I do on the
weekends? Based on 40 wheel stops, which probably happens since I hit areas with stoplights and
signs and take in sights on the grades, that's a 2 minute error in ~5 hrs of riding. That yields
around a 0.1 mph average speed error.

As an electrical engineer with embedded controller experience, the Astrale start/stop overshoot is a
trivial matter for which to program. Yes, there are other errors outside the control of Cateye. Tire
pressure, incorrect tire size, etc. But there really is no excuse for allowing such a correctable
error to sneak in. A large part of engineering is about controlling that which can be controlled,
and I think Cateye dropped the ball a bit. Call me a numbers geek, quibbling over 3 seconds and 0.1
mph errors, but I find these little gadgets fascinating.

Doug

 

[Bill Kingson]

On Mon, 12 May 2003 21:16:04 GMT, Doug <[email protected]> wrote:

>Hey all,
>
>I finally am replacing my incredibly tried and true 1993 (or so) vintage Specialized cyclocomputer
>on my main road bike. I wanted cadence, plain and simple. The old Astrale was butt ugly, the new
>one slick. That coupled with a 20% coupon forced me to "upgrade."
>
>I mounted the Astrale 8 and rode a direct comparison ride. Sensors and magnets were on opposite
>sides of the front wheel to eliminate crosstalk. Wheel sized programmed the same on both.
Wow, Doug, that's quite a detailed analysis. I'd love to see that kind of comparo with many more
bicycle cyclocomputers measured against each other, or a GPS perhaps.

Bill Kingson Caribou, Maine

 

[Luvjava]

I have one idea to offer about the start stop issue. You stated that "the magnets were on opposite
sides of the front wheel to eliminate crosstalk."

I guess that to mean that they were 180 degrees apart when it comes to wheel rotation.....and also
sensors were mounted one per fork blade.

The difference is start stop times my be due to where the magnet was located when you started or
spopped your wheel roll. If the magnet just passed the sensor at the time of wheel stop, then it
would take almost a full rotation of the wheel before it would start again. ( that computer would be
slow to start) or if it was coming up to the sensor at wheel stop (fast to start)

all in all you would think that a 40 start/stop ride, luck of the draw would average them out.

I love my 2002 Model Astral and the extra second or two for auto start/stop doesn't bother me. It's
the fact that the older astral has rear wheel sensor for speed that hooked me, with it I get to
count my trainer miles I ride in the winter (indoor) season

just my thoughts...

Bill

"Doug" <[email protected]> wrote in message news:[email protected]...
> Hey all,
>
> I finally am replacing my incredibly tried and true 1993 (or so) vintage Specialized cyclocomputer
> on my main road bike. I wanted cadence, plain and simple. The old Astrale was butt ugly, the new
> one slick. That coupled with a 20% coupon forced me to "upgrade."
>
> I mounted the Astrale 8 and rode a direct comparison ride. Sensors and magnets were on opposite
> sides of the front wheel to eliminate crosstalk. Wheel sized programmed the same on both.
>
> Not to my surprise, the comps didn't slew identically as speed changed. But to my bewilderment,
> the old Specialized slewed faster! Not by a sneeze either. By a good second or fractionally more.
> The comps didn't even agree on the top speed I reached, 21.2 vs. 21.4. That's a 1% difference,
> probably within specs, although there are no written accuracy specs in the Astrale docs. 1% rated
> is a good bet I presume.
>
> The automatic timer mode, which starts and stops the clock as the wheel starts and stops, was much
> better on the old Specialized. The Cateye started instantly, but it overran the stop by a
> consistent 3 seconds. The old Specialized started 2 secs late, but overran by the exact same time
> when the wheel stopped. Hence it's timer was dead on, all according to short stopwatch tests.
>
> Steady state speed readings and distance matched. The timer error on the Astrale messed up the
> average speed calculation on my short ride. How much so on the real 60 mile rides I do on the
> weekends? Based on 40 wheel stops, which probably happens since I hit areas with stoplights and
> signs and take in sights on the grades, that's a 2 minute error in ~5 hrs of riding. That yields
> around a 0.1 mph average speed error.
>
> As an electrical engineer with embedded controller experience, the Astrale start/stop overshoot is
> a trivial matter for which to program. Yes, there are other errors outside the control of Cateye.
> Tire pressure, incorrect tire size, etc. But there really is no excuse for allowing such a
> correctable error to sneak in. A large part of engineering is about controlling that which can be
> controlled, and I think Cateye dropped the ball a bit. Call me a numbers geek, quibbling over 3
> seconds and 0.1 mph errors, but I find these little gadgets fascinating.
>
> Doug

 

[Dr. Dave]

Doug wrote:
> Hey all,
>
> I finally am replacing my incredibly tried and true 1993 (or so) vintage Specialized cyclocomputer
> on my main road bike. I wanted cadence, plain and simple. The old Astrale was butt ugly, the new
> one slick. That coupled with a 20% coupon forced me to "upgrade."
>
> I mounted the Astrale 8 and rode a direct comparison ride. Sensors and magnets were on opposite
> sides of the front wheel to eliminate crosstalk. Wheel sized programmed the same on both.
>
> Not to my surprise, the comps didn't slew identically as speed changed. But to my bewilderment,
> the old Specialized slewed faster! Not by a sneeze either. By a good second or fractionally more.
> The comps didn't even agree on the top speed I reached, 21.2 vs. 21.4. That's a 1% difference,
> probably within specs, although there are no written accuracy specs in the Astrale docs. 1% rated
> is a good bet I presume.
>
> The automatic timer mode, which starts and stops the clock as the wheel starts and stops, was much
> better on the old Specialized. The Cateye started instantly, but it overran the stop by a
> consistent 3 seconds. The old Specialized started 2 secs late, but overran by the exact same time
> when the wheel stopped. Hence it's timer was dead on, all according to short stopwatch tests.
>
> Steady state speed readings and distance matched. The timer error on the Astrale messed up the
> average speed calculation on my short ride. How much so on the real 60 mile rides I do on the
> weekends? Based on 40 wheel stops, which probably happens since I hit areas with stoplights and
> signs and take in sights on the grades, that's a 2 minute error in ~5 hrs of riding. That yields
> around a 0.1 mph average speed error.
>
> As an electrical engineer with embedded controller experience, the Astrale start/stop overshoot is
> a trivial matter for which to program. Yes, there are other errors outside the control of Cateye.
> Tire pressure, incorrect tire size, etc. But there really is no excuse for allowing such a
> correctable error to sneak in. A large part of engineering is about controlling that which can be
> controlled, and I think Cateye dropped the ball a bit. Call me a numbers geek, quibbling over 3
> seconds and 0.1 mph errors, but I find these little gadgets fascinating.
>
> Doug
It seems to me to be a question of having accuracy in either time or distance, but not both. If you
delay the start for two seconds, you miss all of the wheel turns in that period and you do not pick
them back up in the two seconds at the end of the cycle. Thus your distance will be under repeorted
while your time is accurate.

If you start instantly, you catch all of the wheel turns, but then you throw the time off by
whatever interval you decide to wait to stop the computer after you stop receiving impulses from the
wheel sensor.(This delay value will be related to the minimum speed you can measure with the
device). Here your time is over-reported, but your distance is correct.

 

[Doug]

Bill,

>You stated that "the magnets were on opposite sides of the front wheel to eliminate crosstalk."
>
>I guess that to mean that they were 180 degrees apart when it comes to wheel rotation.....and also
>sensors were mounted one per fork blade.

I had one on each side, left and right, to give both comps pulses at almost the exact same time.
Sensors were on both fork blades. "Almost" since the slight offset of the spokes. You are right that
a 180 magnet offset with both sensors on the same fork blade would have led to errors.

>If the magnet just passed the sensor at the time of wheel stop, then it would take almost a full
>rotation of the wheel before it would start again.

Good observation. I thought of that too, so started the rotation with the magnets a few inches from
triggering, behind the sensors in the rotation direction.

>all in all you would think that a 40 start/stop ride, luck of the draw would average them out.

Very good observation. The behavior of the comps may make some sort of averaging out possible,
depending on how they handle wheel spins during their dead zones. Further testing required!

>I love my 2002 Model Astral and the extra second or two for auto start/stop doesn't bother me. It's
>the fact that the older astral has rear wheel sensor

Rear wheel sensor is definitely nice, I didn't know the old one had that. In fact, when I installed
this one, I tried to route the sensor under the top tube to reach the rear. It could possibly be
done with the magnet/sensor at the wheel outside edge (which I don't like), so I settled for a front
mount since I don't have a trainer. Maybe on small bikes it could be a hack solution, or done by
soldering some extra wire in place.

Doug

 

[Doug]

>> The automatic timer mode, which starts and stops the clock as the wheel starts and stops, was
>> much better on the old Specialized. The Cateye started instantly, but it overran the stop by a
>> consistent 3 seconds. The old Specialized started 2 secs late, but overran by the exact same time
>> when the wheel stopped. Hence it's timer was dead on, all according to short stopwatch tests.

>It seems to me to be a question of having accuracy in either time or distance, but not both. If you
>delay the start for two seconds, you miss all of the wheel turns in that period and you do not pick
>them back up in the two seconds at the end of the cycle. Thus your distance will be under repeorted
>while your time is accurate.
>
>If you start instantly, you catch all of the wheel turns, but then you throw the time off by
>whatever interval you decide to wait to stop the computer after you stop receiving impulses from
>the wheel sensor. (This delay value will be related to the minimum speed you can measure with the
>device). Here your time is over-reported, but your distance is correct.

Very interesting observations! I'm curious to know how the delayed start and stop by the Specialized
comp are programmed.

Does it ignore the spins in the 2 secs start delay, or hold them in memory to use if the spins
continue? Does it hold the last recorded speed while overshooting at the end, thereby counting
phantom wheel spins, or go to zero speed and add that to the tally? What does the Cateye do with
wheel spins at the end?

Wow, I missed this whole line of consideration. This changes my error calculations. Holy cow. Yet
more proof that two brains are better than one.

Testing for these by hand is an easy matter, since the comps register distance in 0.01 increments.
I'll run some tests and report back here.

For anyone thinking this is all a crazy obsession with hyper accuracy when that accuracy is
obliterated by other factors on the bike, no, it's more interest in the computers themselves and how
they work, with the byproduct being an interest in their accuracy.

Doug

 

[Ryan Cousineau]

In article <[email protected]>, Doug <[email protected]> wrote:

> Bill,
>
> >You stated that "the magnets were on opposite sides of the front wheel to eliminate crosstalk."
> >
> >I guess that to mean that they were 180 degrees apart when it comes to wheel rotation.....and
> >also sensors were mounted one per fork blade.
>
> I had one on each side, left and right, to give both comps pulses at almost the exact same time.
> Sensors were on both fork blades. "Almost" since the slight offset of the spokes. You are right
> that a 180 magnet offset with both sensors on the same fork blade would have led to errors.
>
> >If the magnet just passed the sensor at the time of wheel stop, then it would take almost a full
> >rotation of the wheel before it would start again.
>
> Good observation. I thought of that too, so started the rotation with the magnets a few inches
> from triggering, behind the sensors in the rotation direction.
>
> >all in all you would think that a 40 start/stop ride, luck of the draw would average them out.
>
> Very good observation. The behavior of the comps may make some sort of averaging out possible,
> depending on how they handle wheel spins during their dead zones. Further testing required!
>
> >I love my 2002 Model Astral and the extra second or two for auto start/stop doesn't bother me.
> >It's the fact that the older astral has rear wheel sensor
>
> Rear wheel sensor is definitely nice, I didn't know the old one had that. In fact, when I
> installed this one, I tried to route the sensor under the top tube to reach the rear. It could
> possibly be done with the magnet/sensor at the wheel outside edge (which I don't like), so I
> settled for a front mount since I don't have a trainer. Maybe on small bikes it could be a hack
> solution, or done by soldering some extra wire in place.

Cateye will sell you a long-wire mount for rear-wheel mounting:

https://www.cateye.com/smallparts01mas.php?cat=com&&products_id=33

Note that they show no parts yet for the series 8 computers.

Note also that when I went to buy an extra mount for my Cateye HRM, I found I could get a Velo 2
(with its identical mounting) for C$5 more. So I did.

--
Ryan Cousineau, [email protected] http://www.sfu.ca/~rcousine President, Fabrizio Mazzoleni Fan Club