[argyllcms] Re: spotread Question

  • From: "Brian W. Gortney, II" <bwg@xxxxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 10 Aug 2017 10:29:30 -0400

Thanks very much, Graeme.  That's exactly what I was looking for.

The only thing that fit - that I could come up with, given the patterns I was seeing - was that it was temperature-related, but I hadn't taken it further than that and likely couldn't have without a fair amount of digging.  What you're suggesting - tungsten deposition on the inner surface of the bulb - makes sense to me. I'm going to give -Y l|L a shot shortly; hopefully that helps some, though I suppose it's no big deal if I have to simply discard ~20 readings to heat the bulb before beginning my sampling in earnest. If the bulb life is indeed ~1000 hours, what's ~20 seconds of waste each time I read a large set of patches?

Also, sampling with a regular cadence like you suggest does seem to help considerably (after 'warm up').  I developed an XY-positioning table and spotread interface (using pipes) for my application (I couldn't find anything that I liked, that was accurate/precise enough, that wasn't too expensive, that I could interface with in the way I need to, etc.), so I do have good control over read timing.  I had actually noticed that readings after repositioning (moving to the next patch) were slightly 'off' until after the second read (I'm currently doing 5x oversampling), where they'd start to stabilize again - and I had assumed this was due to bulb 'cooling' during movement - so I'm likely going to add a proper scheduler to trigger the bulb at a fixed interval, and then I can just alter the movement pattern (from zigzag to S-shaped, if that makes sense) and then increase the stepper speed such that movements occur in the 'off' interval.  I think that'll probably get me where I need to be.

Thanks again,


On 8/10/2017 08:19, Graeme Gill wrote:

Brian W. Gortney, II wrote:


As you can see, after ~20 samples the Y values level off and are more-or-less stable
(range-bound), but the difference between the initial sample and stable samples is roughly
0.2%, which is a little more error than I'd like to see for this application, generally.

this is likely the "thermal drift" behavior. See:

So a few questions, I guess:

  * I assume this is likely a heating-related effect, in that the illuminant (bulb)
    outputs more or less light depending on its temperature; is that reasonable from
    others' perspectives?  Have others seen this same effect? Is it possible anything
    else (firmware or software-related) is involved?

The state of the bulb has an influence (i.e. it's past usage pattern). If particularly
bad, then the spotread -YL procedure may help.

  * Assuming that temperature-variable light output is the issue, I assume the only way to
    eliminate these effects is to 'warm up' the bulb by taking multiple readings before
    beginning to sample.

Using a steady cadence can help. i.e. taking samples at regular intervals.

    Graeme Gill.

Other related posts: