Evolv DNA Forum
Sign up Calendar Latest Topics
 
 
 


Reply
  Author   Comment  
SSV

Avatar / Picture

Member
Registered:
Posts: 68
Reply with quote  #1 
Here is a simple comparative analysis between device monitor data logging and an actual oscillogram.

The scope input is smoothed over 5 sweeps, sin x/x interpolation, and bandwidth limited to 20mHZ. 

The function measured is simple voltage.....honestly not too bad for a consumer device.  I wish there was some documentation on device monitors acquisition parameters (especially polling rate and how much smoothing is applied to the curve).

P.S. This is a scripted 5 second "puff", and the plots are relatively time correlated

5 second puff.png    5 second puff Scope.png

James

Avatar / Picture

Administrator
Registered:
Posts: 304
Reply with quote  #2 
No smoothing is applied to the curve. It returns the measurement at the time it receives the command.

As far as polling rate, the less things that are checked, the more samples you'll get for what is checked -- it queries checked items round-robin as fast as it can.
Presently, as an implementation detail (threading-related), it updates a bit faster if the window is smaller.
SSV

Avatar / Picture

Member
Registered:
Posts: 68
Reply with quote  #3 
Quote:
Originally Posted by James
No smoothing is applied to the curve. It returns the measurement at the time it receives the command.

As far as polling rate, the less things that are checked, the more samples you'll get for what is checked -- it queries checked items round-robin as fast as it can.
Presently, as an implementation detail (threading-related), it updates a bit faster if the window is smaller.


Gotcha...

Would be nice to nail down an apples to apples acquisition protocol for the scope....

I only ticked voltage in DM.....would be nice to be able to see statistics for voltage over "last puff" params....

What I am trying to do is develop a relatively time correlated acquisition system, for DM vs external measurement.....we should be able to get relatively close granularity in the measurements, as the Atmel MCU on the dna200 (haven't bothered to put it under the microscope and read the PN) and the WRMXi (scope) are both roughly 12-bits of acquisition data (the LeCroy is running in eRES mode, which is 12 bit effective interpolation).  The MCU is 10 bit a/d? (but i am assuming some interpolation is happening between the serial data stream and DM?)

The most interesting thing I noticed was the difference in ramp-up in the curves....I am going to assume the softer knee in the scope waveform is due to the input capacitance of the scope.....the probe is only 1.8pF.

EDIT : X Y parametric data would be very helpful in the DM graph....especially if it correlated to "industry standard' scope params (plus grid)....I.E. amplitude over time, with zoomable incremental steps.....I know you guys aren't trying to build a "scopeDuino" here, but it might be valuable for targeting inconsistencies in heating (due to load non-linearities).  I've managed to get around the lack of a programmable hysteresis window, by moving TCR split points around a bit....not ideal but it works, when trying to "dance" around any consistent decreasing slopes.....obviously we are going to all run into the very real problem of the enclosure/board/signal path drifting (due to thermo-electric effects)
SSV

Avatar / Picture

Member
Registered:
Posts: 68
Reply with quote  #4 
here is a more detailed analysis of the "knee" in the pre-heat window.....


Preheat power = 100w
Preheat punch = 1 (soft knee)
Preheat Time Limit = 5s


knee1.png
knee2.png 

Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.