Eigenharp glitch with ribbons?

I finally have my Alpha working with the eigenapi from @thetechnobear , sending OSC messages, etc, working rather beautifully.

Unfortunately, I’ve run into an intermittent problem where I get an erroneous value for my finger position on the ribbon strips. In this example, I have values around 170 - 220 but notice two places where I get 1 and 4 (see the bold items below).

Has anybody seen this issue before?

strip 14320000 5723560077 183 1
strip 14320000 5723575131 172 1
strip 14320000 5723583072 1 1
strip 14320000 5723583092 2048 0
strip 14320000 5723731108 174 1
strip 14320000 5723766107 186 1
strip 14320000 5723795106 197 1
strip 14320000 5723863956 209 1
strip 14320000 5723913111 220 1
strip 14320000 5724040106 4 1
strip 14320000 5724041066 2048 0
strip 14320000 5724481120 211 1
strip 14320000 5725226576 223 1

I haven’t noticed this myself, but considering what I’m doing with the strips I probably just haven’t noticed it.

It looks like the next value is always a non-active (“finger removed”) value in the example above. Could it be that the values get a bit weird just as the finger is lifted?

yeah, Im not sure what the values are exactly, as its been changed from my test…?! e.g. Im assuming the last number is active status… but Id also like to see strip number.

Im with @Kai it looks suspiciously like the finger is then ‘lifted’ afterwards, and Id not be surprised if when there is light pressure then the value becomes a bit spurious.

note: Im not saying you actively lifted the finger, rather just reduced pressure a bit, to where the harp decided it had been lifted… given flow of numbers this seems likely.

Im guessing you have looked at same number via eigend , and not see the same?

if this is the case, it could be at a higher level they use something like a low pass filter to smooth out the results.
(they obviously post process the numbers… as they have both absolute and relative modes on the ribbons)

Im off to continuucon, so dont have time to take a look until next week.
but if you can let me know a few things I can take a quick look at the EigenD code then…

a) does EigenD show the same behaviour?
b) can you show me the data when you do a ‘proper’ lift off
c) is this only happening in a particular part of the ribbon?
d) is it happening on both ribbons?

as a side note:
I will say, Ive never found the Eigenharp ribbons particularly ‘good’, often needing quite a bit of pressure.
of course, I know you’ve used them quite a bit, so are aware of thier limitations, but my point is… given these limitations Id, frankly, not be surprised if the underlying data is a bit ‘unreliable’ and so subject to heavy filtering.

(on the alpha in particular the ribbons and surface even look a bit ‘uneven’)

I have to press quite firmly to get any touch detection at all. Not what we are used from capacitive touch surfaces these days :slight_smile:
Are it always the same two physical positions where you get low readings when repeating the experiment?

It’s probably an excuse to dig the Eigenharp out of the case where it’s been “resting” for a while. As with the above suggestions, is the location repeatable? The other thing i recall from the strips is that if the pressure point (and I found it most reliable to use the first joint of my index finger) strays too far from the centre of the strip it can register as a release.

1 Like

I’m considering an approach where I will just look at whether the latest value is higher or lower than the previous value and I’ll just move 1 “unit” in the appropriate direction. That way, an erroneous event will only have a tiny impact. It means I’m stuck with “relative” approach but for what I need, that’s going to be good enough.