| T O P I C R E V I E W |
| Alchemy2 |
Posted - 07/28/2014 : 13:07:17
Just a query re: calibration...
How does the unit handle the calibration points? If one uses the default values, the response is linear to the maximum value of the cal. 3 point.
But, if the calibration is somewhat non-linear for some reason, how does the unit handle this? A linear combination? power curve?
THanks! |
| 7 L A T E S T R E P L I E S (Newest First) |
| ZLM |
Posted - 08/05/2014 : 18:41:47
Thank you for your input. We will definitely look into this calculation, and see if we can do anything with it.
Add a x4 on zoom is feasible, but custom factor for zoom may not make much sense due to the limited screen size. It is useful on a bigger screen size. |
| Alchemy2 |
Posted - 08/03/2014 : 03:02:39
While you're at it, if you're updating the firmware, can you add also on zoom for the graph a 4x as well?
(0.1, 0.2, 0.5, 1, 2, 4, Auto) choices would be ideal. Or, allow us to select a custom factor instead:
(0.1, 0.2, N, 1, 2, Auto ; N= some custom factor for zoom)
Thanks! |
| Alchemy2 |
Posted - 08/03/2014 : 02:53:43
Here's a way to do this!
Algorithm is known as a weighted mean:
x values shown below are CPM.
y is the dose rate (x/z) in uSv/h
z is the CPM per uSv/h (200 is the unit default, on all 3 cal. points)
The treatment below can be used to work both for linear calibrations, and a smoothed non-linear treatment, as well as retaining values for the three calibration points. Data points for calibration are
x.y values, with z(n) = x(n)/y(n)
z = ((x-x0)/(x1-x0))*z1 + ((x1-x)/(x1-x0))*z0 ; from 0 to Cal 1
(with x0 = 0 CPM, z0 = z1)
z = ((x-x1)/(x2-x1))*z2 + ((x2-x)/(x2-x1))*z1 ; from Cal 1 to Cal 2
z = ((x-x2)/(x3-x2))*z3 + ((x3-x)/(x3-x2))*z2 ; from Cal 2 to Cal 3
z = ((x-x3)/(x4-x3))*z4 + ((x4-x)/(x4-x3))*z3 ; above Cal 3
(with x4 at the max of the instrument, 65535 CPM max; Z4=Z3 - this is optional but may be left out if easier to omit, as behavior close to max counting capability of instrument may be harder to predict - assume linear beyond Cal 3 data point.)
This ensures that at each data point, the value of the calibration is retained. This above weighted-mean treatment of data points ensures proper mean and linear measures are retained too.
|
| Alchemy2 |
Posted - 08/02/2014 : 12:47:36
Technically, there are 5 points!
Point zero can use the same z value as point 1, and then calculate the values the same way to get the y dose, and past point 3 to the max counts, the value for point 3 (z3) may as well be used for z4. If one is closer to the max counts, one would need to calibrate in that region anyway to obtain any reasonable estimates of dose.
|
| ZLM |
Posted - 07/30/2014 : 11:54:08
The current version firmware uses lower value first. If the CPM reading reaches second point, then change the conversion to second vaue.
It is a good idea to give an hybrid reading for dose when the CPM close to second point but still in first range. Please let me know if you have any idea to calculate the hybrid reading, so that we can try to implement it. If this hybrid calculation works, then we can have a non-linear calibration capability. Which is much better. |
| Alchemy2 |
Posted - 07/29/2014 : 16:18:46
I understand that these points are roughly linear in the range that they are set for, but does the software/firmware then interpolate the cal points between the set ranges? Example, let's say that point 1 is set for 200CPM = 1.0 uSv/h, and 500 CPM = 3.0uSv/h for point 2. If I am reading 400CPM what happens? Does it take the 200 value and says ~2uSv/h, or does it average the two set point values in some way to give a hybrid reading for dose?
Thanks. |
| ZLM |
Posted - 07/28/2014 : 20:15:34
The unti has three points calibration. That means it may ba able to set three different linear combination value. |
|
|
