Increase Accuracy of Ping Sensor

Howad01

I received a reply to use a math coprocessor compatable with the BS2 to increase the accuracy (decimal and fractions of inch or centimeter) using the ping sensor. Unfortunately I was on the road when I checked my e-mails and left it on the mail server. But when I logged from home it was not there. I would greatly appreciate someone resending this reference.

Thank You,

Howard Epstein
Howard@FCC1.biz

rtowler

We have been experimenting with a PING in our shop here (our first BS project [noparse]:)[/noparse] and we have found that you can get mm resolution *MOST* of the time without any additional hardware, but there seems to be an inherent limitation in the PING itself where at certain distances the TOF (time of flight) varies significantly between 2 discreet values. Most readings are correct, but occasionally they will vary by say 40us which is around 6-7mm (The test setup is as close to ideal as one could get so you can rule out interference). This is very repeatable and is independent of sampling frequency.

It would be nice to know why this is happening and if there are potential fixes/workarounds. If you ignore the errant samples, the PING measurements are very good (we've tested to about 1.3m). For our application these errant samples pose a problem though We understand that we're probably taking PING to places it wasn't intended to go.

So Howard, you may not need the uM-FPU...

-r

Jon Williams

For clarification, the Ping does what it does; using a math coprocessor may help with conversion from timing units to distance measurements -- though I dare say the Ping accuracy is more coarse than numbers inside the Stamp when you use our demo code.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax

rtowler

I should add that we're using a BS2p. The relatively slow sampling rate (probably not the best term, but I mean the frequency at which PULSIN checks the state of the pin) of your BS2 will limit you well before you do any fp calcs.

-r