Inolux uses (in its datasheet template for each of its LEDs) a log-log scale to depict the Relative Luminous Intensity, but it's not depicted in a manner that I've witnessed before.

For example, IN-P32TRRRGB: From the datasheet:

IN-P32TRRRGB - Relative Luminous Intensity - Questionable Grid Spacing

There are 8 increments between each decade rather than the 10 I'm more accustomed to. The increments do not get smaller from left to right toward the end; some actually get larger.

  • Is this typical for depicting log grids?
  • Is there a straightforward, calcuable means to approximate these data points?

I reached out to the manufacturer but have not heard anything yet.

If they were cutting corners, doing a sloppy job on the grid which goes on every LED datasheet seems like the worst way to go about it, so I questioned myself rather than them.

  • 4
    \$\begingroup\$ The scales on that graph are nonsense. Don't trust the supplier. \$\endgroup\$
    – Andy aka
    May 2, 2023 at 9:02
  • 2
    \$\begingroup\$ It makes even less sense once you notice that 10² to 10³ has one less division than the other decades. \$\endgroup\$
    – jpa
    May 2, 2023 at 10:24
  • \$\begingroup\$ "some actually get larger" - I suspect that's because they've misaligned the labels. But it does look like a graph drawn by someone who understands drawing better than graphs. As for the gridline count, it's certainly possible to force a silly number of divisions on log graphs (in LibreOffice, play with the Minor Interval Count) \$\endgroup\$
    – Chris H
    May 2, 2023 at 10:44
  • 2
    \$\begingroup\$ Kelvinhertz does not inspire good confidence either. \$\endgroup\$
    – winny
    May 2, 2023 at 12:57

3 Answers 3


These graphs could be pure fantasy. But we don't know that. Certainly they are not very useful.

If they did such a sloppy job with a simple graph in a datasheet, can you trust the manufacturer with other things?

I would definitely measure these things myself, and use a more trusted LED as a crude calibration source for a home-made photodiode+TIA measuring jig.

  • \$\begingroup\$ Aww. I was afraid that might be the answer. \$\endgroup\$
    – kando
    May 2, 2023 at 1:59
  • 1
    \$\begingroup\$ @kando I'm sorry I don't have better news. \$\endgroup\$ May 2, 2023 at 2:09
  • 4
    \$\begingroup\$ "Right in the heart of silicon valley" - you would expect them to be able to produce a loglog plot. \$\endgroup\$
    – D Duck
    May 2, 2023 at 11:17

All the log scales in that datasheet are messed up. This isn't some weird standard they've employed, it's just lazy, crappy graphing. It's not you, it's them.

The graphs are next to useless as they are, but I suppose you could superimpose that curve onto your own scale, and use that to take readings instead.

  • 2
    \$\begingroup\$ At that point I'm highly suspicious whether the curve is something someone just drew so the graph resembles graphs in the other datasheets, without quite understanding what they are doing. It's all a matter of opinion, but it just doesn't pass the smell test to me. \$\endgroup\$ May 2, 2023 at 14:43

Is there a straightforward, calcuable means to approximate these data points?

The log point corresponding to grid position can be calculated as 10x where x is the position between decades as value between 0 and 1. From the graph the grid lines appear to be:

  1. Position 0.00: log value significand 1.0x
  2. Position 0.25: log value significand 1.7x
  3. Position 0.50: log value significand 3.2x
  4. Position 0.60: log value significand 4.0x
  5. Position 0.70: log value significand 5.0x
  6. Position 0.80: log value significand 6.3x
  7. Position 0.90: log value significand 7.9x

So roughly, you can think of the grid lines as 1, 2, 3, 4, 5, 6, 8.

I agree that the mess of the grid doesn't speak well for the accuracy of the rest of the data.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.