(Teletype) IN cv read latency?


#21

I use an unpowered backpack
devices are:

  • meadowphysics
  • earthsea
  • ansible
  • TXi
  • TXo

Here is a small clip (issue is at ~ 1min 50 sec) https://www.dropbox.com/s/00bv0vewkxq6unx/Teletype%20Crash.mov?dl=0


#22

would you be able to try only leaving TXi and TXo plugged in and see if you still experience the issue?


#23

2.2 will feature optional profiling so that we can answer questions like this better.


#24

it feels more unstable


#25

only with the expanders on the i2c bus i am not able to reproduce it


#26

Bingo.

I have found that with four or more devices on an bus without a powered adapter you will get hangs on reads very consistently and, less frequently, writes as well (especially at high speed). Honestly, three devices is pushing it.

You need one of these, my friend:

or DIY:

It is SMD with some pretty tiny resistors, but a pretty easy DIY project if you are into giving it a try. :slight_smile:


#27

this needs to be added to the docs in big letters somewhere.


#28

@bpcmusic thank you very much, i’ll give it a try .
Regarding SMD, everything bigger than 0402 huge :slight_smile:

Best
Marcus


#29

Ok; for the TXi - here is what I found.

I set up a slightly different script. I was setting a random value on the TXo CV output and reading it at a TXi CV input. I printed the values to positions in the current pattern. I used the first potentiometer on the TXi to adjust the delay. The script looks like this:

I
P.N 0
M 400
TI.PRM.MAP 1 0 200

M
A RAND V 10
D TI.PRM 1
P 0 A
P 2 D
TO.CV.SET 1 A
DEL D: P 1 TI.IN 1

Here is what I learned (on the latest v.017 TXi firmware):

  • If there is no delay command prior to the read - it isn’t reading fast enough to return the appropriate value. (That is, take out the DEL D prior to the last line of the metro script.)

  • If there is a delay command prior to the read, even if the delay is zero (0), it reads the output CV appropriately.

So - the appropriate read delay is somewhere between zero and whatever the delay is before something of delay 0 is executed.

I’ve been running it for 2+ hours and it hasn’t missed a beat, btw.


#30

here is a test build (based on the latest official 2.1 version):

teletype.zip (102.9 KB)
teletype.hex (350.8 KB)

pull request: https://github.com/monome/teletype/pull/134

video demo:

there are 2 sounds, one is driven directly from pressure points, the other one from teletype - this one is delayed by 200ms to make it easier to hear. the script is simple:

CV A IN
DEL 200: TR.P 4

first i test without any delay on the first line, you can hear it still doesn’t always get it right. but adding just 1ms delay seems to be sufficient.


#31

@scanner_darkly thanks for the update, it works very well


#32

Awesome, thanks so much for this!!!
I will do my best to get it fired up tonight.


#33

Nice, any chance to add it to the current grid version ?


#34

sure, i’ll add it to the next beta.


#35

CV.OFF seems to be broken in this build (has no effect).
@scanner_darkly or @koerby - do you see this as well?


#36

that’s weird, that part wasn’t touched at all. i’ll do some testing later today. can you send me the scene?


#37

whoops, serious pilot error.
I was doing CV.OFF 2 1 and not CV.OFF 2 V 2.

…slowly backs out of room…

Sorry for the noise ;D


#38

I do this pretty often!


#39

ha, no worries! :slight_smile: i actually never use CV.OFF and had to check the manual to see what it does…