Analog Discovery DAC

[gus789]

Hello - looking at the specs for the Analog Discovery kit I could not find out as to whether it has analog output capability other than the arbitrary function generator. I am looking at analog out for the purpose of sourcing stable voltages as part of a feedback control loop, i.e. read value from analog input, compare with setpoint, adjust output accordingly - could this be accomplished by simply programming a constant voltage in the arbitrary function generator? Thanks!     

[tggzzz]

... - could this be accomplished by simply programming a constant voltage in the arbitrary function generator? Thanks!     

That is the mechanism used to calibrate the analogue inputs.

[gus789]

Thanks - any thoughts as to how fast on can program the arbitrary function gen ( as that could limit the feedback loop bandwidth )? I am trying to decide between the Analog Discovery and the NI USB-6001 for this application which will be ultimately run with matlab. The Analog Discovery seems to be the better bang for the buck in terms of specs and ultimately more usable beyond this particular application - although the NI would more than suffice for the application in question.   

[f5r5e5d]

going thru USB driver layers on PC makes sample rate and latency rather unpredictable

I often had problems researching NI USB DAQ interfaces for control apps - they simply can't/won't tell you what speed you can close the loop at


closing the loop is better done with a uC and bare metal programing - could be < 100 lines of code - and you can hard code control loop sample rate in timer interrupts

lots of <$100 options today with Arduino, PIC, ARM dev kits and project boards for chips having built in ADC and DAC or PWM output

of course if you want to learn some uC RTC environment there are free options there too today

BeagleBoard or RaspberryPi could run open source Matlab workalikes Octave/SciLab in Linux with a RT kernel but probably need added DAQ shield/cape/add on boards

[gus789]

Thanks, I have considered using an mbed board for this, but the 12-bit resolution on the adc is not enough and would require an external adc and probably a dac too, although the STM uC on the Nucleo board I have seems to have a 16-bit dac (not sure how many channels - I need at least two at 14 bit and would like another two with at least 10 bit resolution). Then there is the complexity of programming the thing and interfacing it with matlab, so that is why the USB NI and Digilent modules are much more attractive to me. Raspberry Pi and Beagleboard have other complications IMO. I have just ordered the NI module and will have to try it out and see how fast it goes. If I can get the loop to ~1kHz bandwidth I am gold.  If it doesn't work well I will  steer clear of usb daq boards in the future for such applications. Thanks for the input!

[tggzzz]

Thanks - any thoughts as to how fast on can program the arbitrary function gen ( as that could limit the feedback loop bandwidth )?

No idea, but an API has been produced so you can program it from the PC over the USB.

If control loop speed is important, then you might benefit from looking at the Red Pitaya. That has such control loops as a prime use-case. The control can be done either in a on-board dual-core A9 ARM or even directly in hardware: dust off your VHDL/Verilog expertise