brendan mccloskey

this is a general question; someone may be able to point me to alternate sources of info.

What are the relative merits and demerits of serial-usb sensor data when compared with OSC? I've a vague idea of the resolution and sampling frequency of my own Arduino-based DMI, and I'm considering the arguments for and against translating into the OSC protocol. I am eminently unfamiliar with OSC. Any pointers would be gratefully received - I'm just looking for a comparison of their relative sampling/baud rates and resolutions etc.

And then there's x-OSC, x-io to think about too.

Also, it occurs to me that sampling freq/baud rate are largely academic; is it not ultimately dependent upon how fast the serial object is polled (qmetro speed)? How low can qmetro go?

Hi
this is a general question; someone may be able to point me to alternate sources of info.

the-last-line-of-my-thesis

Maybe I'm not relevant but beside speed, the physical layer is important.

OSC can be transmitted over many different physical layers. You can use wifi, ethernet cables, etc.

I often have my computer 50m away from the arduino (or whatever sensor) and this would be quite difficult (although possible) over USB. You can also easily use many arduinos with one computer.

But note that you don't have to use OSC to send datas over Ethernet.

Maybe I'm not relevant but beside speed, the physical layer is important.
OSC can be transmitted over many different physical layers. You can use wifi, ethernet cables, etc.
I often have my computer 50m away from the arduino (or whatever sensor) and this would be quite difficult (although possible) over USB. You can also easily use many arduinos with one computer.

But note that you don't have to use OSC to send datas over Ethernet.


thanks for responding. I agree that the physical layer is equally important. At the moment, OSC is an option only based on the wireless compatibility, I'd like to ultimately lose the USB cable. But I would also like to justify it based on resolution and sampling frequency benefits - if there are any.

Hi Patrick
thanks for responding. I agree that the physical layer is equally important. At the moment, OSC is an option only based on the wireless compatibility, I'd like to ultimately lose the USB cable. But I would also like to justify it based on resolution and sampling frequency benefits - if there are any.

I'd be very interested in your results/insights of others.

Isn't this comparrison kind of inaccurate? Shouldn't we say serial vs. UDP for example instead of serial vs. OSC? (I might misunderstand something here..)

Anyway.. I found serial communication always a bit frustrating since it has the potential to crash the computer easily if one doesn't pay attention to what one's doing. Also it seems to drain more performance from the CPU. I've done a lot of experimenting with a raspberry pi and pd and serial communication seems to be much more of a problem for that little PC than UDP.

Hi!
I'd be very interested in your results/insights of others.
Just to add my 2 cents:
Isn't this comparrison kind of inaccurate? Shouldn't we say serial vs. UDP for example instead of serial vs. OSC? (I might misunderstand something here..)
Anyway.. I found serial communication always a bit frustrating since it has the potential to crash the computer easily if one doesn't pay attention to what one's doing. Also it seems to drain more performance from the CPU. I've done a lot of experimenting with a raspberry pi and pd and serial communication seems to be much more of a problem for that little PC than UDP.
cheers!


Yes, I may be being inaccurate, in that serial to usb is a bus, where OSC is a protocol. I think?

What I meant was, in general, which method of sensor-data acquisition has the optimal baud rate, resolution, sampling rate? In the context of Arduino, there is an OSC library is there not? Is this "better" than serial-usb?

Based on my personal experience I completely abandon the USB/serial connection for all my arduino based controllers. Besides speed and possible distance I find the udp/OSC connection much more reliable. (I.e. in case of an accidentally unplugged cable, just plug it back and everything continues to work.)

The real difference in speed I can only guess, I never measured it. But taking into account that the max baud rate on an Arduino is 115200 and the ethernet interface of an ArduinoEthernet has a speed of 100Mb/s I would say it is secure to assume that ethernet will outperform the serial connection by far.

115200 baud = 921600 Bit/s are approximately 0.9 Mb/s

The real speed of course will depend on many more factors (speed of the Arduino board sensor reading, overhead of the the OSC protocol and others)

... and yes there is an Arduino OSC Library by CNMAT that works like a charm :)

Assuming 8-bit bytes:
Serial:
115200 baud = 921600 Bit/s are approximately 0.9 Mb/s

The real speed of course will depend on many more factors (speed of the Arduino board sensor reading, overhead of the the OSC protocol and others) 

If it comes to the resolution I'd say the limiting factor is not the data connection but the sensor accuracy and most important the Arduino AD converter wich is 10 bit (max 1024 steps).

Jan, you are an absolute diamond, this is exactly what I need; I anticipate there may be additional comments, and thanks also to Patrick and woyteg.

:) Thanks Brendan. What exactly are you working on? Will there be chance to see something in the future?

and just for completeness: other limitations with ethernet are introduced by the other hardware components in your network. So max speed is determined by the slowest device in your network/data chain. though in reality you won't find many (if at all) hubs, switches routers that still operate on 10Mb/s.

I have a matrix of DIY FSRs, embedded with LEDs in several silicone pads. It's designed to maximise independent creativity amongst digital musicians with cerebral palsy, and is based on an Arduino-Max platform.

Nothing new here (Linnstrument, SLABS, Soundplane et al), but the underlying control-performance layers are quite sexy, with some novel touches to expose explicitly expressive control modalities.

Anyway, I'm planning to iterate the design over the next few months/years and I want to maximise the sensitivity and resolution of the sensor matrix, so the numbers you quote will allow me to identify "better" platfroms and protocols, so your help here is invaluable. I was recently encouraged by someone at a conference to translate the platform/protocol into OSC (it was/is becoming the lingua franca in this field).

This is one of the remaining questions in the conclusion chapter of my PhD thesis.

Hi Jan
I have a matrix of DIY FSRs, embedded with LEDs in several silicone pads. It's designed to maximise independent creativity amongst digital musicians with cerebral palsy, and is based on an Arduino-Max platform.

the last line of my thesis . . . .