ADC For Compound Eye on Edison

I can't answer your Edison related questions as I haven't used one but for your ADC question about how to read multiple inputs, then the answer is yes you can depending upon a few factors. You need to consider the maximum sample rate of the ADC and think how fast you need to sample each of the inputs. You also need to factor in an time it takes to "switch" different inputs to the one ADC and allow settling time. So when you factor all this in, if you have enough time to sample at the rate you need then you can do it. Is the ADC you are referring to one built into the Edison or is it an external ADC you are planning to use? If it's external then you can get ADC's with a built in multiplexor to take multiple inputs to a single ADC channel. The problem will be controlling it if you are short on GPIO. Maybe if you put up a quick sketch of how you think your system should fit together it might help us give you more ideas on how to proceed.

With regards to the Edison, there are a few on here who have done quite a bit recently, hopefully one of them will see this thread and jump in.

Best Regards,

Rachael

For sampling, in case you don't know, the Compound Eye is used to basically triangulate objects with IR sensors. I pass the sensor value to the processor, and it's usually used to triangulate the closest or furthest object.

So, I'd need to read all at about the same time, if that would be possible?

I'm still free a few GPIOs, with my GPIO listings here, for what I plan to use this on:
Intel-Powered Mech (DW Juggernaut)

In short, I have a number of GPIOs still free, but for analog, I need:
-To read 4 analog for the Compound Eye
-To read a rheostat

-To possibly switch my Sonic sensors to mid-range SHARPs.

Thanks for the help so far!

vandia wrote:

 

So, I'd need to read all at about the same time, if that would be possible?

I don't know what sample rate you need for the Compound Eye but lets for arguments sake say you need to sample it at 1KHz. Lets assume you have an ADC which has multiple inputs and can switch inputs and sample at a rate of 1MHz, that would mean you'd be able to take the four samples for the Compound Eye, each sample being 1us apart in time and then the Compound Eye's overall sample groups would be at an interval of 1ms. So substitute the above for the actual values you need for the system and the ADC you plan to use and if you get values which give you the "about the same time" that you need then you should be ok.

I've got to do some actual work this morning so I will take a look at your blog post in more detail later. I did have a quick look earlier, it looks interesting

Best Regards,

Rachael

I'm sorry, I'm still catching up on some terminology.
Either that, or because I wasn't to bed yet (4:30am), I thought by "frequency" you meant how quickly it can check/carry, not like baudrate. XD

Maybe.... do ADCs usually read and transfer pretty quickly? If not, for what I plan to use the screen for, I may move my directional sensors to the ADC (only checked once/sec), and plug the eye for it when I'm ready directly into the board (needs to update a little more in realtime, let's say 1/5 a sec. I'd have to change some code, but is the speed to read an ADC much slower externally?

Thanks so much for your input!

Edison and ADC throw some extra issues.

It is a multiplexed arrangement, where an extra chip is involved, as opposed to many of the other uP's.

This may have the effect of slowing it down.

You may have more luck with I2C ADC chips and deal with it that way.

Mark

Thanks! Do you happen to have an estimate on the actual delays between the onboard, GPIO connected, and I2C ADCs?
Or, at the very least, be able to tell me what "stat" I'm looking for that's important between these (besides price, lol).

Do you happen to have an estimate on the actual delays between the onboard, GPIO connected, and I2C ADCs?

Sorry no I don't.

This one suggests a conversion time of 3.2uS but there are some other figures that may override that,

http://www.analog.com/media/en/technical-documentation/data-sheets/AD7291.pdf

TI have one that talks about 1uS

http://www.ti.com/lit/ds/symlink/adc121c021.pdf

The Edison Arduino board uses the ADS7951 converter.

http://www.ti.com/lit/ds/symlink/ads7951-q1.pdf

The sheet suggest 800nS.

Since it is multiplexed, you could assume the data is there, but the Edison then polls for it, so in reality 'live' updating is going to be slower.

balearicdynamics is embarking on an I2C series, and this might be a great topic.

Mark

And.... I probably should know.... but uS means....?

Usally it stay for micro seconds as the micro character from Greek character is not available while writing on the US keyboards.

Enrico

Usally it stay for micro seconds as the micro character from Greek character is not available while writing on the US keyboards.

Enrico

Didn't he reference both, nanoseconds and uS? I thought both were 1/1000th though then? *confused*

vandia wrote:

 

Didn't he reference both, nanoseconds and uS? I thought both were 1/1000th though then? *confused*

ms = millisecond = 1/1000th of a second

us = microsecond = 1/1000000th of a second

ns = nanosecond = 1/1000000000th of a second

ps = picosecond = 1/1000000000000th of a second

The list goes further but its unlikely you'll need to worry about ps (and not even ns or us for this) let alone those beyond....

Best Regards,

Rachael