The Stouffer step wedge I received today (January 23, 2023).

dkonigs

Subscriber
Joined
Sep 17, 2009
Messages
355
Location
Mountain View, CA
Format
Multi Format
This is certainly an interesting idea to try, though there are a few problems with its practicality:
  • The relationship between PWM duty cycle and LED brightness may not be exactly 1:1, so I'd still need a way to measure/calibrate that part. (Which kinda already requires a calibrated sensor.)
  • The LED brightness is absolutely not a constant, and varies with on-time and power level in a way that looks like an exponential decay function. I've put a lot of work into minimizing the effect of this, by improving heat dissipation around the LED itself (which makes the curve shallower), and by making the LED on-time a constant for each measurement cycle. However, this also means that I need at least some cooldown period between measurements to keep things consistent.
  • Trying to measure by reading a target, then reading an unobstructed light and varying its brightness to match, would take a lot of on/off/cooldown/on/off cycles especially given the resolution I'd need to match numbers so they'd still hold up on a logarithmic scale.
In other words, it may be worth attempting, but has serious limits to its practicality.
 

Nicholas Lindan

Advertiser
Advertiser
Joined
Sep 2, 2006
Messages
4,214
Location
Cleveland, Ohio
Format
Multi Format
dkonigs said:
The LED brightness is absolutely not a constant...
Click to expand...
Yup.

There are a few approaches:

  1. Measure the LED output by measuring a blank (either air or a reference reflective surface) and use this measurement as the LED output. The LED is kept on for as little time as possible - this means the LED is at the same temperature (or close enough to) for the blank measurement and the sample measurement;
  2. Measure the LED temperature and make an open loop first order correction for ambient temperature, again keeping measurement time as short as possible.
  3. Use closed loop control of the LED brightness. The LED is kept on so the control loop can function. A benefit is that the LED becomes an aid to positioning the sample.

A drawback of the first method is when the system doesn't have mechanical control of the sample. The instrument can, of course, use the operator for mechanical control with messages of "Lower the measuring head with no sample", "Insert the white reference plaque", "Insert the sample."
 

albada

Subscriber
Joined
Apr 10, 2008
Messages
2,171
Location
Escondido, C
Format
35mm RF
I used a simple solution: Keep the LED cool by running little current through it.
I'm driving a typical through-hole indicator-LED at 5 volts through a 6k ohm resistor in series, so the current is under 1 mA. The DA meter sees no evidence of sag at any PWM duty-cycle from 100% down to 1/16 or 1/32. And the DA meter has a resolution of 0.01 stops (not logD), which is plenty sensitive to detect LED-sag.

Mark
 

MattKing

Moderator
Moderator
Joined
Apr 24, 2005
Messages
51,937
Location
Delta, BC Canada
Format
Medium Format
Thread title updated - because any reference to "today", "soon", "now" can get confusing over time.
 
OP
OP

Chuck_P

Subscriber
Joined
Feb 2, 2004
Messages
2,369
Location
Kentucky
Format
4x5 Format
In the interest of accuracy, I realized that my PD Cal 2 from my post #71 was incorrectly calibrated.......the Stouffer step #20 density is 2.88 where I had incorrectly input a density of 2.84..........so, after re-calibrating the PD to that density and reading each step of the Stouffer wedge again the below graph is the result............it's an exact match, basically.

 
You must log in or register to reply here.
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…