Ant+ Fundamentals Question
 

I am now dealing with my first Ant+ device (Bontrager Cycling Computer with Cadence/Speed sensors). This was installed by my LBS and for a couple weeks worked just fine.

Today neither cadence or speed were there. The Bontrager instructions are 'not very rich', so I tried to go through pairing again. As best as I can tell this requires stepping through all the setup stuff (enter time, date, wheel size, and so on - not hard). The last step should take you to pairing. Instead it just said 'done' - and it was done (and no cadence or speed).

I finally gave up and just started my ride. Roughly 3 miles out (call it 10 minutes) speed and cadence just appeared. And it was rock solid for the next 67 miles. From what I can tell cadence and speed are now gone again.

I have never dealt with Ant+ so I don't really know exactly what to expect here. So some questions.

1) I assume that once 'you are paired' that no repairing is normally required. Is that right? Is losing pairing common? Note that it happens on two sensors simultaneously.

2) How does an Ant+ sensor 'know' that it is pairing? Does it have the ability to 'listen' to a signal from the computer so start broadcasting? Does it always send signals that the computer can sense and use for pairing purposes?

3) Does this sound like a common Ant+ problem? How do folks 'typically pair' a new sensor? I'm suspicious of my computer instructions because I have encountered at least one other scenario (Start New Ride) that absolutely does not match the instructions.

dave

Which model of bontrager computer?

On the node 2.1 (i think all the nodes work the same), hold the lower right button to connect to the sensors and hold the two right buttons to pair the sensors (you only need to pair with the first set-up or a new battery, maybe).

Sensors transmit the data with a code that uniquely identifies the sensor. That way, the head unit can ignore sensors it hasn't been paired with (like sensors on the bikes of the people you are riding with). "Pairing" is the process by which the head unit determines which sensors it should listen to. You pair rarely.

The sensors are only transmitters (they don't get information from the computer; they don't care whether there's a computer listening at all).

When you are using the computer normally, it has to "connect" to the sensors. The nodes display the connected sensors at the top left.

My node works well but it sometimes drops the connection before riding (I get double dashes instead of the speed). Holding the lower right button fixes that. Note that you can stop the connection process by pressing a button once the indicators for the sensors you have stop flashing.

It seems that your issue is a connection issue (not pairing).

Try forcing a connect before you start riding. Spin the crank or wheel (either is enough) to wake-up the sensor and then put the head unit into connect mode. The crank/wheel does not have to be moving while connecting.

Mine is a Trip 300. And to correct what I said earlier, there is a second way to initiate pairing. Just press/hold the two side buttons (5 sec).

Apparently my speed/cadence is back this morning (appeared to be gone last night). I also notice that the manual indicates that I should have a couple of icons in the upper/left portion of the screen (showing speed/cadence sensors). Those do not appear. Just not sure what to think at this point.

Thanks.

dave

I suspect that computer works like the nodes.

Your problem is connecting (not pairing). I added more info to my earlier post.

It's normal that the sensor icons disappear. They disappear when the device goes to sleep.

At the start of a new ride, the computer has to reconnect to the sensors. (Pairing is connecting and registering; connecting is just connecting.)

Thanks - that is very helpful.

dave

Just one more question for anyone. I spent about 15 minutes Googling through Ant+ information. It sounds to me like an Ant+ sensor in a typical bike scenario (speed, cadence, etc) is unidirectional (sensor cannot receive) and therefore it is always sending information. Power drain becomes manageable because Ant+ power levels are quite low and information is sent in very short bursts (zero output power most of the time regardless).

Is there something else going on here?

Thanks.

dave

As I said earlier, the sensor ony transmit (they don't recieve). There is nothing useful they can receive.

The sensors have two modes/states: a sleeping mode and an active mode. The sleeping mode transmits nothing. In the sleeping mode, the sensor checks for movement in a way that conserves power. It's possible that the active state is doing funky things to reduce power demands.

Keep in mind that the basic idea of ANT is that the sensors provide very basic data in a way that uses minimal power and that it's the head unit that does anything complicated.

I'm not sure what else you think could be going on.

In my case the Bontrager manual is pretty explicit. You can pair or connect ANYTIME, initiated just on the computer side of things. So something else is going on. There is a reasonable chance that it is simply a manual error, but that is what caused me to speculate other scenarios (such as sleep mode transmitting on a much reduced duty cycle).

dave

You cannot, without first waking up the sensor. You can connect anytime the sensor is awake. Pressing the setup button on the sensor or the passing of a wheel or cadence magnet will wake the sensor.

You are correct and the manual (and me) is most definitely wrong.

TWENTY steps (count'em - 20) into setup is 'press the whatever button' to start pairing. There is absolutely zero reference to having to wake up the sensors nor is there any reference anywhere in the manual to this concept (outside of the very last page which references connecting). Note that EVERY ONE of the 19 previous steps can be done without 'sentient sensors'.

One of two things has happened here (in retrospect)

1) The manual is badly thought out

2) My LBS didn't deliver to me sensor documentation that would have clarified this. Even so, IMHO, the complete lack of reference to such documentation (if it exists) in the computer docs is still a significant 'error'. But see my post at http://www.bikeforums.net/electronic...-question.html to see why I might have a bias here.

dave

It's fairly easy to surmise that the sensors aren't that "sentient" (it would consume much more power than necessary and isn't really a useful feature). No wireless bicycle sensor (ANT+ and others) work that way.

That's one benefit of resources like this forum being available.

I wonder if any documentation makes it clear.

Good luck with the computer!

Making that assumption (consume more power than necessary) assumes a knowledge of the power cost of scenarios such as

- low duty cycle bursts of 2.4GHz broadcasts
- doing a relatively constant monitoring of input (or not)

that is not obvious to me. Maybe it is obvious to you. While I would hardly say that 'low duty cycle/continuous broadcasts' is huge useful, it would allow folks to both pair and connect their sensors and computers without having to deal with (for example) spinning their wheels and cranks to check things out. And I would point out that there IS a duty cycle that makes continuous broadcasting of data irrelevant (pick your definition of irrelevant - unless it is aboslutely zero). The question is the relative power consumptions and marketing trade-offs derived from that.

Plus it would save Bontrager the effort to fix their documentation :-)

dave

Assuming that these sorts of things strongly favor minimizing power consumption doesn't require detailed understanding at all. Especially, considering the source of power and that the transmitters spend a majority of time off duty.

You have no idea if things where really working unless you are getting data (if the wheels/cranks are spinning).

The sensors often get bumped and move too far from the magnets (this is a common issue).

So, off-duty transmissions are very close to pointless even without considering power and complexity issues.

You only (basically) have to pair once. Pk

I think some head units do that.

Just an interesting tidbit snipped from ANT - Wireless connectivity - WikID, the Industrial Design Engineering wiki that is related to parts this conversation.

dave

• Micro-watt power consumption (Approx. 4 years of battery life on a 2032 coin cell when sending a message every 2 seconds, 24 hours per day)

For cadence/speed/heartrate transmitters, the life time, for normal usage, is much less than that.

The ANT+ stuff works well anyway.

It seems BTLE was inspired by it.

IIUC ANT is the basic protocol that allows for a wide range of network possibilities, while ANT+ is a restriction on that which makes for easier interoperability by requiring each device to implement a "profile".

But the basic setup is for the sensors to operate as masters (that transmit) and the head unit/ computer to operate as a slave (that receive). I guess there could be bi-directional comms using ack messages for example but I don't know if that is required in any of the profiles. The sensor has to be placed into "pairing" mode by eg a button push which causes the sensor to transmit an ID. My Garmin slave also needs to be placed into "search" mode which I guess is a receive mode that looks for the pairing transmissions. Once the slave accepts the pairing info it should remain paired. At that point the slave should be listening and when a master (sensor) xmits it should save data if it is paired or ignore it otherwise. At least on my Garmin head unit, it seems to recognize a paired speed/cadence sensor as soon as the magnets pass, so I assume on a single xmit message from the sensor. I haven't checked range, but at least a couple of meters I get reception. The Garmin also shows the saved ID of the paired sensor in the status pages.

scott s.
.

This wouldn't be a good design for sensors generally.

Multiple head units can read the same sensor data (I do it and so do some tandem riders).

Acknowledging the head units would require the sensor to keep track of more than one head unit. And, if an acknowledgement failed to occur, it's not clear what the sensor could do that was useful (retransmit the data? for how long?). It would just make the sensor more complicated for no good reason (for a situation that can easily tolerate a bit of data loss).

Having to put the head unit into a "search" mode is important for reliabilty/sanity. Otherwise, your unit could pair with sensors that are not your own (such as the case when somebody decides to pair their sensors when you happen to near-by).

Pairing is some thing that needs to be done rarely (it's OK to make it a bit harder to do).

The Garmins start listening for sensors when they are turned on. They have to recognize (basically) all transmissions from paired sensors. So, it shouldn't be surprising that one pass/transmission works.

Having a 'button to push' on the sensor to make it send 'additional pairing info' would not be surprising. But do any Ant+ sensors work that way? I have only seen one (Bontrager Speed/Cadence pair of sensors) and there is no button. So I assume that, much like the Netgear router in my house, anytime the sensor is actively sending data it is also sending the required pairing data.

dave

"Pairing" is a bit of a confusing term.

The sensors send out packets of data. Each packet includes a number that uniquely identifies the sensor.

The sensor doesn't care what is listening to it or that it's being listend to at all.

That is, there's nothing special that the sensor needs to do when paring.

It's the head unit that does all the work.

Basically, "pairing" is the process by which the head unit knows which sensors (by their identification number) to pay attention to.

The only type of sensor that would need to have a button is one where the communication was two-way. (The fact that the bicycle sensors don't have a button is an indication that the communication is one way.)

Note that Bluetooth devices are often two way. For two-way communication, both devices need to be put into a "pairing" mode so that both can determine which other devices to pay attention to.

The ANT devices are not usually two-way but, since two-way is possible, the same "pairing" term is used.

So in general terms (for the typical Ant+ implementation that a cyclist would run into)

1) When the sensor is alive it is sending both ID (and other if it is required) information that allows the head unit to distinguish it from the other devices on your bike, devices on other bikes near you, noise/crap that exists in the 2.4 GHz range, etc

2) A head unit has a 'list' of devices that it should pay attention to. In a noisy or possibly crowded situation it might well be getting all kinds of 'input' and it is ignoring this unless it can see that it is coming from 'somebody on the list'

3) Ant+ dramatically narrows the range of data for a given/valid connection (part of the device profile for a given sensor)

4) "Pairing" is the process that puts a device 'on the list' (see #2 )

5) Anytime an Ant+ sensor is active (for the typical implementation of Ant+) it is sending the data required to authenticate itself to the head in the system

6) The 'special steps' that happen when pairing happen at the head. In this case it is critical that ONLY valid devices be in range of the head. Once you put the head in pairing mode, unless the Ant+ implementation in question has additional steps for pairing, anything 'in range' at that time is now paired.

Is that how things typically work?

Thanks.

dave

Yup (basically). Well-described (by you) too.

The important thing that you might be missing is that the data is being sent in packets.

For networks (where you have multiple devices talking to other devices on the same channel), information is sent as "packets". These packets are marked with something that identifies the sender and (often) the receiver.

This is very analogous to letters mailed in envelopes that have two addresses (the sent-to and sent-from addresses).

For the typical ANT+ sensor, the packets being sent are being broadcasted (where there is no specific receiver). For something broadcasted, the packet isn't marked with a receiver.

This broadcasting is analogous to radio stations where the station sends stuff out whether or not anybody is listening but you have to tune-in (pair) to a particular frequency to listen to the data.

The paring process looks at the send-from addresses in the packet (ignoring the contents of the envelope).

ANT and Bluetooth are light networks (sort of like what the internet/wifi uses but not as heavy-duty or robust).

"Handshaking" is a synonym for pairing (usually for two-way communication).

Makes sense. My direct experience is with wired sensors and polling is an option there.

scott s.
.

I was looking at ANT Message Protocol and Usage Rev 5.1. But looking a little more closely, it seems there are a couple of implementation options. One involves a "pairing" bit set in the Channel ID of the master (sensor). That requires some action by the sensor to enter the mode. But I guess it is also possible to simply have the slave (head unit) search without following the pairing bit. That way the master doesn't need to do anything. I guess the idea is that "pairing mode" is intended to reduce false (unintended) pairings in some environments. It would help to have the ANT+ docs but I think you have to be a member of ANT+ consortium to access them (didn't look for them uploaded elsewhere).

I have a Garmin GSC-10 sensor and that has a "reset" button.

FWIW the ChannelID data format:

http://bikeforums.net/attachment.php...hmentid=416178

scott s.
.

It's not really a wired/wireless thing at all.

Polling is an option (theoretically) here too. "Not polling" is a design choice.

The head unit having a "pairing mode" is a requirement. The sensor having such a mode is not a requirement (whether or not there's a benefit).

Being required to put the sensor into a pairing mode would keep the head unit from pairing with sensors on nearby bikes (not that unusual a situation).

Two sided pairing is required to make a single channel for two-way communication (common in Bluetooth).

You reset the Bontrager by putting in the battery backwards for 2 seconds.

The reset button on the Garmin GSC-10 appears only to be used when setting up the sensor with the magnets (it isn't used when pairing).

The way the Bontrager works, it doesn't need a reset button for set-up.