Encoder

[heikoh]

Guten Tag,

mich würde interessieren, ob jemand Erfahrung mit Rotary Encodern (z.Bsp von Alps: ALPS STEC12E07 Encoder...) im Zusammenspiel mit Fischertechnik Rastachsen oder Metallachsen hat.

Gibt es Encoder die besonders gut zu Fischertechnik passen (Durchmesser)? Die Spannung von 5V sind kein Problem, da ich die Auswertung über einen Arduino mache. Lässt sich ein herkömmlicher Fischertechnik Motor damit vernünftig ansteuern, oder gibt es gravierende Unterschiede zwischen FT Encoder Motoren und "normalen" Motoren (abgesehen von den integrierten Encodern)?

Viele Grüße

Heiko

[H.A.R.R.Y.]

Hallo heikoh,

zu den Encodern wäre meine Überlegung mir welche mit 4mm Achsdurchmesser und mindestens 20mm Achslänge anzuschaffen. Alles andere benötigt Adapter, die extra Geld kosten oder Bauraum beanspruchen. Wenn so ein Encoder eine Zentralbefestigung à la Potentiometer hat, läßt er sich recht gut mit einem Poti-Flachstein 32116 befestigen - das erpart auch hier weiteren Aufwand.

Die Motoren selbst sind einfache Gleichstrommotoren mit "Bürsten"-Kommutator. Da ist es herzlich egal, ob Du einen uralten mot1 oder einen modernen Encoder-Motor nimmst. Klar gibt es ein paar Feinheiten von wegen Leistung, Wirkungsgrad und so weiter, die grundsätzliche Wirkungsweise und die Ansteuerungsmöglichkeiten (=Motortreiber) sind jedoch gleich. Paß aber bitte mit den Betriebsspannungen auf, die alten Motoren eignen sich nicht für 9V.

Grüße
H.A.R.R.Y.

[winijenh]

I've been struggling with encoders. Tried everything from printing encoder wheels myself on transparent sheet, demolishing old computer mice to get encoder wheels from there, etc. I've glued them to Rastritzels (31082) and Hülsen mit Scheibe (35981). Results not very satisfactory.
At some point I decided to get some real sensors. I use the AMT10X series by CUI (http://www.cui.com/catalog/components/e ... al/modular). Benefits: very reliable, insensitive to electromagnetic noise, fit to 4mm, 5mm, 7mm and everything else round that fischertechnik has. They can also be set to a very wide range of pulses per revolution to meet any axle rotation speed. You want as many pulses per rev as the TX counter inputs can handle (which is not more than 600 per second for pulses with a 50% duty cycle, slightly higher for shorter duty cycles). Drawback: they are expensive (EUR21 each), and they are not small, they need a specific connector/cable. They take 5V and the outputs can go directly into the TX inputs. I have not tried running them from the 5V source in the TX extension slots, as that doesn't seem to be able to deliver any current to speak of. I've glued them to a Lagerblock 5 (38252).

Now the real pain: the sensors provide nice quadrature outputs, but the TX can't handle these. So I used the LS7183 by LSI (http://www.lsicsi.com/encoders.htm) IC to convert into two clock signals: a clock up and a clock down. Feed both into the TX counter, and use the software to subtract them, and you get a very accurate position reading. If you use an Arduino, you should be able to deal with the quadrature output directly in the software.

You can see the sensors in action here: http://www.ftcommunity.de/categories.php?cat_id=2833

Hopefully the TXT will be true open source so I can modify the counter inputs to deal with quadrature inputs properly...

Regards,
Willem Evert Nijenhuis.

[vleeuwen]

Thanks Willem Evert,
Nice sensors. Bad luck that Fischertechnik never implemented the black box (need two fast C counters) for a Rotary or Quadrature encoder on the TX-C.
This suggestion was already done 3 years ago.
This kind of encoder could be very helpful for resolving position problem with the robots and Industrial automation models.
Creating these sensors is not the biggest problem.

[Peterholland]

Hallo Willem Evert, Carel + andere FT-ers,

Interessant und lieferbar bei Knobloch ist auch Artikelnummer 75150 :
https://knobloch-gmbh.de/de/quadratur-e ... riebemotor
Quadratur Encoder-Motor 9V (Getriebemotor) Motorabgangswelle für Rastsystem, eingebauter Quadratur-Encoder 0/9V Ausgang (max 10mA) mit 90° zueinander versetzem Taktsignal, mit 4pol. Steckverbindung für Impulsausgang (benötigt 1*119785) zum Anschluss an Auswerteelektronik; Abgangsdrehzahl unter Last ca. 175 U/min. Einbaumaße ohne Welle: 30 x 30 x 60mm;
Dies ist eine Sonderversion von uns aufgrund der großen Nachfrage: Der normale 9V Encodermotor (135484) bekommt eine andere Elektronikplatine mit Quadraturausgang - Aktuell wird diese Technik in RoboPro noch nicht unterstützt (Stand 03.2014)

Schau auch: https://knobloch-gmbh.de/de/catalogsear ... ?q=encoder

Gruss,

Peter
Poederoyen NL

[vleeuwen]

Thanks Peter, very nice.
I know that there was a prototype but now it is available. I love it.
Useful in combination with a I2C application, because there exist a need for the firmware.

[H.A.R.R.Y.]

Hi,

generally I prefer using I²C devies for such tasks like an encoder interface. The benefit is that such a small device unloads the main processing unit (e.g. TX) from the time consuming tasks of asynchronously evaluating fast encoder signals or whatever the application needs. The TX just needs to ask the interface via I²C for the latest news whenever the TX likes to do this. But be aware that I²C traffic also consumes some time.

How about some skilled ft:c members to develop such gadgets? Is there some interest?

Regards
H.A.R.R.Y.

[winijenh]

I've searched for ICs that take quad encoder signals and deliver a position over I2C, but they don't seem to exist. If anybody finds one let me know. The cheapest solution is likely to build something with an Arduino that takes quad encoder signals and acts as I2C slave to the TX. But for the purists amongst us that is not acceptable...

Willem Evert.

[Ad2]

Or program it into an attiny which costs less than €2.

[vleeuwen]

Why not, Jan Willem?
A sensor with preprocessing the data in a microcontroller into meaningful data for the TX-C is the intention of the I2C interface of the I2C.
I don't see the problem for even the purists.

However low speed of the I2C master in the online mode is regrettable.
So it is difficult to use it in high speed implementation.

[winijenh]

@Ad2 Have you ever tried that? Are the attiny's (with internal clock) fast enough to handle I2C at 400kHz and maintain the counters at a reasonable speed (let's say 1kHz)?

Willem Evert.

[H.A.R.R.Y.]

Are the attiny's (with internal clock) fast enough to handle I2C at 400kHz and maintain the counters at a reasonable speed (let's say 1kHz)?

It depends on the exact model of ATtiny and how you write the code. So a general answer is not possible.
From experience I expect ATtiny2313 or similar to be able to handle one encoder up to 1kHz (maybe higher) and work as an I²C slave with up to 400kHz.
Using C-coding should work for most applications. In special cases (FLASH/RAM footprint, IRQ response, ...) you might want to consider manually optimized assembler code. Implementing the I²C compliant slave behaviour is the tricky part with the ATtinys I already know; they just feature USI not TWI. Internal clocking usually is above 1MHz with these constraints, a typical value is 8MHz but this also depends on what the finally choosen ATtiny offers.

Regards
H.A.R.R.Y.

[Ad2]

I have used the USI in an attiny2313 to connect to the TX via I2C, using internal clock. I have not tried a quadrature encoder but I don't expect problems. For a single channel you can probably use an 8 pin attiny85.