DPlot

While some may use the grain size template for showing particle size distributions, we work in a difeerent region, or wider region is more accurate. We have for many decades used arithmatic probability on the "Y" axis to represent cummulative percent passing each screen size or micron size. The "X" axis is linear as to the major and minor lines, but the values are made up of Number of microns roughly correlated to the original screen series (U.S. or Tyler). The screen openings, and series of screens were by the "Root of 2" as the major divisions and the "4th Root of 2" as the minor divisions. The root of two series screens , the micron sizes were not linear, but were plotted on a linear scaled graphed X axis. The "Y" axis as I said earlier was Probability. Between the root of 2, were minor tick marks of the 4th root of 2. I bought your product for the probability and flexibility. I have come close, but cannot quite get it. Any suggestions? I can provide examples. We can go from 12,500 microns down to less than 1 micron (Not all on the same graph).

I can provide examples.

Please do, when you get a chance.

I am attaching two sheets. One is a sheet showing a couple of standard series screens for particle size analysis. In the United States people vary between the U.S. Standard Series and the Tyler series. They often are mixed and matched together. Both use what is essentially a ever decreasing screen openings. The largest being marked in inches, then switching to microns. Once the openings are in microns, the openings decrease based on two factors. The first, which corresponds to the major grid lines of the ( linear) x-axis, is the micron scale decreases by the division of the square root of 2, or 1.414, approx. The half way point between the major divisions, are the alternate micron openings which decrease by (2) to the 0.25 Power, or the quarter root of two which is 1.1892. So, by looking at the attached graph, you can see that it takes a naturally ground natural material, and makes it give you straight lined section(s), with one break or sometimes two (Bi-model distribution). Plotting this way tells someone if the testing was done accurately and if the screens might have a small hole in them and they have lost calibration. You can see the difficulty of having a linear plot on the x-axis, but the values are far from linear. Can you help?

I have gotten this far and cannot figure out how to send the graph and such? Be on the lookout, I will find something.

Probable1

The forums will only allow you to link to files that are somewhere else on the web. In other words you can't have a simple attachment as you can with e-mail. If this is a problem you are welcome to attach your examples to an e-mail and send to support@dplot.com

Nevermind

Just received your e-mail, thanks.

Thanks for the files. This appears to me to be simply a log base 2 scale (rather than log 10). I can certainly add this, but rather than restricting it to grain size plots I'll likely add a global option for log2 scales. So you could have

log2 X, linear Y
log2 X, log10 Y
log2 X, probability Y
log2 X, log2 Y
linear X, log2 Y
log10 X, log2 Y
probability X, log2 Y

and "Grain Size Distribution" plots would have an option to use log2 X rather than log10.

Unfortunately (for this change) I've promised myself to set aside all new feature requests until I have several longstanding to-do items out of the way, so it may be a while before this gets done.

You can kinda sorta fake one of these plots now by applying a transform to your X values (Edit>Operate on X with X=10^(log(x)/log(2)) ) and then replacing all of the powers of 10 on the X axis with the corresponding powers of 2 using Text>X Labels and using 'None' for the number formatting on the X axis. And yes that's a pain... forget I mentioned it unless you're in a rush to produce a plot.

I will try your suggestion. If I am capable of making it work, it's worth a shot for now. The list of options you envision would be great for use in our business, which is the manufacture and sale of fine grinding and ultra-fine classification equipment and process systems.

I look forward to when you will have time to work on this feature. This is a great program, and will become even greater over time. Thanks for your help and caring about your customers!

Probable1

The jerry rigged example you gave me worked for my purposes wonderfully. I hadn't been aware of the method for assigning labels manually. My data is logarithmic (base 2), so I set that on the linear/log scaling. my problem is that now I have a funky grid. Not really sure why it looks like this, but I'll try to represent it below (horizontal grid lines not depicted)... Why did it render the grid this way. More importantly, is there a method I could use to force Dplot to place the grid lines over the labels I've set manually.
Much thanks,
Arthur

Oh, another question. I read the command syntax and didn't find a way to access the text>x labels information programatically. Did I miss something, or is this a situation where I need to save it as a preference?
Thanks again,
Arthur

The "funky grid" is due to the log10 scaling; I had not thought this through as much as I should have. Sorry, but there's no way to rearrange those as you'd like. You can get rid of the grid lines entirely by setting the grid line width to a negative number, and add your own grid lines with Options>Reference lines, though that just adds another bit of complexity to the workaround.

Labels: somehow I neglected to document this. Use [XLabel(X,"label")]

Awesome, that works perfectly. Is there any way to selectively eliminate just one axis' grid lines, in this case the x axis? Not a huge problem, just wanted to know whether or not i need to use the reference line method for the y axis as well.
Thanks,
Arthur

No, sorry - there's no way to do that. You'll need reference lines for both axes. I take it from all of the above that you're really determined if you're still working on this This will of course all be much better with a new scaling option.

Yeah, it is fairly important. As it turns out, though, the log base 2 scaling wouldn't actually work for my purposes. My x axis contains a series of common camera apertures. The common values for these fstops are sometimes fudged. ie f/5.6 is actually 5.657... A more intelligent logarithmic labeling wouldn't have worked, because f/5.6 would display as f/5.7. Anyways, the custom x label works great, and the y label will work fine with some extra effort. Thanks for the help.

I have been following the most recent discussion with Mr. Echells (If I goofed up your name, I am sorry!!), and I had the same problems with the grid, etc. Now, I would have no objection to using reference lines for the X Axis for my use, But, using reference lines for the Y Axis using the probability scale (99.99, .01). I am afraid that is just too many reference lines for me. I guess I will have to wait patiently until this feature comes around to the top of your "To Do" list. Did I say patiently? Let me amend that. I will be waiting, going CRAZY, but so what else is new. Stay healthy, eat right, so you can get to this project!!!

Probable1