On Sun, 14 May 2000, Saul Goode wrote: > i have the new o.s. and the knobs still fluctuate, which really sucks. in > the old o.s. they didn't go all the way to 127, but they didn't jump. in the > new o.s. they do go to 127 but switch around to 125, 126 and 127 which is > kinda worse, cuz if you programming, your values may jump. this is because > the dctrl knobs are all the way up, so make sure that they're not up all the > way. Ouch - doesn't sound good. I knew there was a risk with the improving the end value accuracy - but in my tests I did not get the fluctuation. It can differ on some units - and depending on how the problem exactly appears there might be something we can do about it. Is all knobs affected, or is it mainly one of them? > but i have a lot of faith in elektron, and i'm sure all these > short-comings will be addressed. It is very hard to make the direct controllers 100% stable for all units. We use the A/D-converters inside the HC11 microcontroller to decode them, and the values I get from these are not stable at all. From an 8-bit value they vary as much as 8 steps up and down, and to get them to the stability they are now involves a lot of massaging and filtering. We've tried many different approaches to make it better, but haven't succeeded. The problem is that the fluctuation is not purely quick ones around a stable middle value, but also sometimes a bit slower. When we've tried to filter these out the responce have been slower which is not good as well. Right now I'm using a middle value algorithm together with a "pithole" around the last accepted value. Ideas to improve the algorithm are welcome! :) //Daniel
Message
Re: [elektron] Realtime Controllers?
2000-05-15 by Daniel Hansson
Attachments
- No local attachments were found for this message.