Is there any Arduino clone available with a TCXO driven MCU? I would like to have a more accurate internal clock compared to ceramic oscillators / crystals in the current Arduino reference designs?
Yep, I like the AVR capture hardware, it can time all sorts of sensors and would be accurate if the MCU had a crystal. Anyway, I've started to make my own boards now and have found I like the 1284P more than the venerable 328P.
One option is the pricey Ruggeduino-SE
but I like the 1284P so perhaps Crossroads has a board...
http://www.crossroadsfencing.com/BobuinoRev17/
The ATmega1284P is my favourite: Another Arduino clone with 1284P - Microcontrollers - Arduino Forum. The board has the pins for 32k crystal and also on header.
On Ebay chinese clones for less 10 $ exist with an Xtal .
Going all the way from the ceramic resonator is a TXCO is quite a jump. Would a higher-accuracy crystal be sufficient?
Looks like a "poor" resonator is 0.3% (3000ppm), a good crystal is 10ppm, and a TCXO is 1ppm. (with the crystal, you might have to pay more than usual attention to the load capacitor values to actually achieve that accuracy...)
The ceramic resonator I have [MURATA,CSTCE16M0V53] is not very good, it is +/-0.5% Tolerance and +/-0.3% Stability. So it could be 0.8% off from 16MHz at the edge of its temperature range. The part also has a better grade .15% grade, but that is not much of an improvement.
I thought the TCXO was a 10ppm crystal (like what I use to measure pulse timing from a capacitive humidity, liquid flow, coulomb counter or capacitive level sensor) so I did not look it up. After doing so I found a 2.5ppm device, so I need to remember this.
This TCXO takes 3.3V so it is not really a good fit for the older 5V AVR's (like the Mega328 and Mega1284), but is a good fit for ARM, XMega and many others. It also has an HCMOS output... so that probably means a fuse needs to be set to use an external clock.
Also, if the timing is important I would make sure to use a trustable source for the crystal or TCXO. Too much stuff on Ebay is just crap that failed the test and then was not properly destroyed.
I am wondering whether the ChipKit Max32 is a valid option. At first the hardware looks promising, however I do not comprehend how they implemented the software part since I have no knowledge re PICs. Therefore, will attachinterrupt() work as usual and does the PIC use the 16MHz crystal as reference for ticks, i.e. micros()?
I do not comprehend how [ChipKit] implemented the software part
There is an arduino-like IDE for the ChipKits called "mpide" that implements most of the same functionality and libraries as the Arduino IDE (though it seems to be somewhat stuck at a version slightly earlier than the Arduino 1.0?)
http://chipkit.net/started/learn-basics/mpide-quick-start-guide/
micros() should work fine. I'm not sure about attachInterrupt() (and it might depend.) The exact nature of pin-driven interrupts tends to be subtly different between vendors.
I've also been looking for an arduino with an more accurate crystal too. For the uno r3, I've found my favorite clone that I like even much better than the original:
it has a real quarz crystal, male AND female pins and space for 2 buttons. I know it's "just" a quarz crystal, but it's much more accurate than the ceramic resonator on the original r3. That's enough for me.
There is a Mega clone with real crystal from geeetech (red board) , but the few reviews I've found say that this board is faulty. So I won't try that one, but there is a version of the Mega clone that has prepaired spot to solder on a crystal yourself, however this also involves desoldering the existing smd resonator and caps.