rejon [rejon!~rejon@li382-141.members.linode.com] has joined #milkymist
<wolfspraul>
what about removing the 3 buttons to make place for more USB connectors?
<wolfspraul>
(in rc4)
<wpwrak>
you should ask this when adam is here. i think you'll see the shine of great fireworks from the general direction of taiwan ;-)
<wpwrak>
or does roh pre-fab them ?
<wpwrak>
that is, assemble
<wolfspraul>
yes
<wolfspraul>
roh does it
<wolfspraul>
if we need a button for some very low-level features like enabling rescue mode, we could also go down to just 1 button
<wpwrak>
then the fireworks will come from berlin. even easier to see ;-)
<wolfspraul>
that would also create space to go to 4 USB connectors
<wpwrak>
we could perhaps use IR to select rescue mode. that's relatively light-weight.
<wpwrak>
but one button for maintenance tasks sounds acceptable, too
<wpwrak>
that could also be standby/reset/etc.
<wolfspraul>
standby is removed and I think that's good
<wpwrak>
i think there could be a place for a "real" standby. but the old logic wasn't so great.
<roh>
i think it should keep 1 or 2 physical buttons. ir i do not use anyhow
<roh>
but to be fair, from my pov we should rather make hubs work than waste endless hours on redesigns of mechanics (even if that means work for me ;)
<roh>
after all.. lots of keyboards etc include hubs already which means, they do not work at all atm
<wolfspraul>
if more usb would only be about bypassing the hub problem, I would agree
<wpwrak>
hubs are a separate issue :)
<wolfspraul>
but a hub that is needed only to connect essential devices, that's a lot of clutter
<wpwrak>
bbiab
<roh>
wolfspraul: thats my impression, please correct me when i'm wrong there
<wolfspraul>
right now we have keyboard + mouse, if even 1 usb-midi devices is added it's too much already
<roh>
wolfspraul: hubs ARE normal and mandatory in usb. the only ones not supporting them are we atm.
<wolfspraul>
of course we should support them
<wolfspraul>
but that's unrelated to having more connectors on m1
<roh>
not from my pov.
<wolfspraul>
would we need to include a hub in the box?
<roh>
no.
<wolfspraul>
I think the unit should work for the intended use case
<roh>
i would rather use a keyboard which has a hub included and connect the mouse there
<wolfspraul>
keyboard+mouse+usb-midi = 3
<wolfspraul>
later on we might support usb storage or wifi dongles, or other usb devices
<roh>
the 'cable layout' around a mm is messy anyhow atm. connectors on all sides means it eats around the space of 20-30cm x 20cm, even if its much smaller
<roh>
wolfspraul: are you still in .de?
<wolfspraul>
yes
<wolfspraul>
how was the congress btw?
<roh>
maybe we should make some paper models of different case-variants and connector placements and play it through the different usecases
<wolfspraul>
sounds like a lot of work :-)
<roh>
wolfspraul: nice.. wasnt there much, but we had quite some interresting guests in the raumfahrtagentur
<roh>
wolfspraul: i dont think so.. rather like a afternoon with some drinks and a bit of glue on fingers
<wolfspraul>
ok but how could that lead to something we implement later?
<roh>
lasercutters can do paper too.. really fast :)
<wolfspraul>
the product feels very focused to me, sure you can argue about connector position for a long time, but why?
<wolfspraul>
the idea was just to remove 2 buttons and add 2 USB connectors instead
<roh>
wolfspraul: i work best when i can imagine well how stuff looks/works in different usecases. so if we maybe find a variant useable in the vertical or so. or some placement of connectors on 'less sides'
<roh>
maybe not even for v4, but some later revision
<wolfspraul>
vertical - you mean make the m1 stand on the side?
<roh>
maybe? not sure yet. i just see the tables of dj's and vj's being _stuffed_ always.. so a device eating a lot of horizontal space isnt likely to be placed there without some cleanup
<roh>
maybe even make rackmount possible. i mean.. what would you do to 'install' it in a club setup?
<wpwrak>
back
<wolfspraul>
good point about vertical, but I think that takes us far away from the initial "how about removing some buttons to make place for USB?" question
<roh>
wolfspraul: true.
<wolfspraul>
if you ask so openly, sure, the whole shape of m1 can be questioned
<roh>
what i am thinking is that the corner with the buttons is the only one left which does NOT have connectors atm. so placing some there would make it even worse...
<wpwrak>
(connector positions) maybe a topic for M2 ?
<wolfspraul>
my perspective is mostly manufacturability, yield and cost
<wolfspraul>
for example the buttons are painful, all three
<wolfspraul>
could we have some sort of proximity sensor that works through the acrylic?
<wolfspraul>
no hole, no mechanical that can wear out, no buttons to assemble
<wolfspraul>
I'd LOVE THAT, even for rc4
<roh>
not really precise that stuff.. and gets confused in piles of cables
<roh>
so i dont think that its a good choice for that device
<wpwrak>
roh: i don't see M1 as a rack-mounted device from the ownership perspective. also, we may consider adding a touch screen in M2.
<wolfspraul>
or a conductive stripe we glue around the top and into the case
<wolfspraul>
there must be some way to get rid of the mechanical push-button
<wolfspraul>
in a cheap and easy way, not big redesign, touch-controller etc.
<roh>
wpwrak: huh? id rather remove all buttons and leds besides one and make it 19" to be fair.
<wpwrak>
(manufacturing buttons) i think they should be milled. it can't be *that* hard. i get quite acceptable results with wood. and that lacks structural integrity and is generally "fuzzy"
<roh>
i'd rather not try that
<wolfspraul>
if we have 1 button, and only for rescue cases, it might be enough to make a small hole in the acrylic, and tell people to use a thin pen or so for pressing
<roh>
wpwrak: milling is a possibilty, but also a lot of work (and needs difficult to get material (weird thickness acryllic))
<wolfspraul>
then no more acrylic button either, just the push-button inside
<roh>
wolfspraul: thats sucks. hw needs a reset/powerbutton.
<lekernel>
i'm for the capacitive touch sensor through the acrylic
<roh>
its bad style not adding one
<roh>
lekernel: you obviously never played around with capsense.
<lekernel>
actually I did, with the small avnet spartan-3a board
<wolfspraul>
lekernel: do you have a particular module in mind? url?
<roh>
i wasted days on implementing capsense and its annoyingly tricky to get even ONE button reliable. and it depends majorly on component and spacing tolerances.
<wpwrak>
roh: you could mill also the button surface. it'll change the appearance, but you can probably make this look intentional.
<roh>
wpwrak: still i need thicker acryllic for that.
<wolfspraul>
lekernel: how do you feel about removing 2 buttons to make place for 2 more USB connectors?
<roh>
4.5mm we currently use
<wpwrak>
roh: anything >= 5 mm will do
<lekernel>
why 4 USB connectors? 3 are enough, no?
<wpwrak>
roh: 4.5 mm ... maybe. it'll be "sunk" but may work okay
<roh>
wolfspraul: >4.5 looks weird because fo sticking out too much
<wpwrak>
my wooden buttons are about 4.75 mm
<lekernel>
and those three buttons are actually used in the software
<wpwrak>
roh: (> 4.5 mm) you mill of what you don't like :)
<wpwrak>
s/of/off/
<roh>
.oO(we wouldnt have this discussion at all if we had usb-hub support)
<wpwrak>
lekernel: 4 dual usb take up about as much space as 3 single ones. and they're probably more robust, mechanically
<wolfspraul>
dinner, bbiab
<wpwrak>
roh: you still don't want to have to carry around an extra usb hub :)
<roh>
-> shower and lab
<wpwrak>
lekernel: for USD 500 screensaver mode ;-))
<roh>
wpwrak: they are built into keyboards. nobody i know besides lekernel uses that wobbly rubber thing we ship out.
<lekernel>
roh: btw what's going on at the agentur those days?
<roh>
lekernel: diverse stuff... mostly we are happy about a bit of quiet time after the rush of xmas and congress
<lekernel>
roh: ...and before xmas?
<roh>
work, infrastructure improvements etc
<roh>
there is always something to do
<lekernel>
sounds very fuzzy
<roh>
it is. its not like a company where you can assign tasks and stuff gets done
<roh>
there is an audiolab now. (we got a wall built etc)
<roh>
2 rooms, one for preproduction, one for live stuff (radio)
<lekernel>
you're starting a radio?
<roh>
we somehow inherited one.. or rather the bootlab brougt it with them.
<lekernel>
oh, so the bootlab really exists? :-)
<lekernel>
that's the thing rumored to be at tacheles, right?
<wpwrak>
lekernel: they're probably still busy building the subterranean biohazard level 6 lab ;-)
<roh>
bootlab is a eV which now resides inside raumfahrtagentur, and they do some hours a week on 88.4 fm
<roh>
bootlab was at tacheles once, and before raumfahrtagentur, they had their own small room in stattbad
<lekernel>
and which none of the dozen or so tacheles people I asked about has heard of :-)
<wpwrak>
roh: (usb hub) don't count on every keyboard one will use there to have a hub
<roh>
hrhr... i woudlnt say tacheles is 'managed' somehow
<roh>
wpwrak: i dont. but i know that people have such stuff anyhow and its integrated into a lot of devices. even the kinect has one inside separating 3 devices
<roh>
i bet there are even usb-midi devices consisting of more than one usb device and a 'hub', even when there are no additional external sockets
<wpwrak>
roh: i'm not arguing that we shouldn't support usb hub. just that it doesn't replace the usefulness of having more built-in USB ports
<roh>
wpwrak: true. but you can see how reworking the whole socket setup could make sense, right?
<wpwrak>
you mean away from the "connectors everywhere" approach ? of course
<roh>
i would rather have 2 usb sockets and buttons on front and ALL sockets on the back if ever possible. (and of course 2x usb on the back also)
<wpwrak>
but that sounds like an M2 task
<roh>
maybe.
<wpwrak>
massive redesign
<roh>
true
<wpwrak>
you'd probably also want to move to a more solid case. and panel-mounted connectors. etc. quite a bit of work
<wpwrak>
and i think a touch screen, 7" or maybe even a bit larger, would be part of the equation :)
<roh>
i know that m1 get where it is now though lots of small iterations over time.. maybe its time to think about it from 'the outside' and redo some (mechanical) details before continuing
<roh>
touchscreens fail massively in the creative areas.
<roh>
reasons are the same they worked so bad for openmoko, but worse since they are bad in timing and precision
<roh>
also they are expensive and would drive price up not down, even when only a small percent of users needs em.
<roh>
wpwrak: lemur is an 'input device'.. you cannot compete with that
<roh>
lekernel: not sure.. you mean on the door?
<wpwrak>
roh: i mean the touch screen as a input device. well, and a bit of status display
<lekernel>
yes
<roh>
lekernel: maybe 10-15E? but if you are there earlier i could get you in touch with the guy doing vjing for the builing
<wpwrak>
roh: the use case would be: 1) all the controls you don't assign to external devices, 2) navigation (image/patch selection)
<wpwrak>
roh: no virtual keyboard or such nonsense. well, at least not for any relevant functionality.
<roh>
wpwrak: you see.. i'd rather use vga only for beamer output, no ui whatsoever, put that on the web.
<wpwrak>
roh: vga for the visual output. built-in screen for control. right now, we have everything time-divided. the best we can do with the current approach would be an overlay. but that wouldn't be available in performance mode, only when developing.
<roh>
mmmmh.
<roh>
i still think built-in screens are too expensive for us.
<roh>
why not let others make that investment and the users 'use a tablet' ?
<wpwrak>
roh: seems that you can get retail iphone screen assemblies for ~USD 100.
<wpwrak>
using a tablet would be an option, yes
<roh>
wpwrak: thats not a screen but a toy. expensive toy
<roh>
and you still would need another 'gpu' for it. and pins for driving it
<roh>
with a tablet you could go onto the floor and 'work' from there
<roh>
thats why i seperate 'user interface' from 'the box with all the cables' in my thinkning.. it makes things so much more flexible
<wpwrak>
advantage: hw off our back. no need to write sw infrastructure. drawbacks: mechanically less tidy. connectivity tricky. need to develop for android and/or the democratic people's republic of apple
<lekernel>
wpwrak: just use a web interface
<roh>
wpwrak: the webinterface would be some xmlrpc or json control interface, the rest is html, css and 'images'
<lekernel>
you can do tons with javascript and ajax those days
<wpwrak>
lekernel: that would probably take away the controls
<roh>
ack. drive the xmlrpc (which then also native apps for droid and iOS could use)
<roh>
i'd only leave a legacy ui in the mm for set-up-times before the event
<roh>
lekernel: btw.. i really need to get you in touch with the guy from stattbad. he is really interrested into talking with you about mapping
<roh>
thats something people waste the cpus of multiple macbooks with atm. just for doing some transformations and clipping before pushing stuff to the beamers
<lekernel>
ok ...
<roh>
wpwrak: btw.. lemur seems to have removed its display and is avail now as iOS app
<roh>
not sure if they sell hw still or its only a piece of sw now
<wpwrak>
roh: as i said, the tablets killed them. too hard to compete with mass-produced hardware.
<roh>
true. i hope we learn from that ;)
<lekernel>
roh: so what do you propose?
<roh>
not building input devices for now. do the rendering/backend part properly if we want to sell hw.
<roh>
put input interfaces where artists like them to be/make it flexible
<roh>
let them use whatever midi/osc/dmx stuff
<roh>
need to run, bbl
<wpwrak>
for now, we don't have much of a choice anyway
<wolfspraul>
I agree with roh but I also think that advocates more USB ports
<wolfspraul>
(will read back the entire thread in a bit)
<wpwrak>
hard to tell. if you want to make an external tablet part of the equation, you need to solve a few integration issues. not sure how well they work with just USB. or whether they can even run with USB power.
<lekernel>
imo web interface with modern javascript/ajax is the way to go
<wpwrak>
they come from a world where even tethering is a forbidden fruit. so my expectations are low.
<lekernel>
almost no android/apple specific development to do
<lekernel>
can't they connect to any wireless access point?
<wpwrak>
yes, they have wireless. then you need to add that to milkymist. plus all the configuration issues.
<roh>
wpwrak: routers dont cost real money. 20-25E for something which can ap mode, briding from ethernet. don't build it yourself.
<wpwrak>
and i have my doubts about web interfaces. the more demanding they get, the more compatibility issues you find. too many moving parts.
<wpwrak>
roh: so you'd have a stack of PCBs ? Mx plus an OEM router. is there a long-term stable wireless router PCBA on the market ?
<roh>
wpwrak: i wouldnt integrate the router
<wpwrak>
kewl. yet another small box to carry around.
<roh>
maybe package one if customers really ask for it.
<wpwrak>
see, that's the problem with using off the shelf tablets. you save the manufacturing work but integration explodes into your face.
<roh>
wpwrak: people do anyhow.
<wpwrak>
(do anyhow) i don't see many people carrying wireless routers around
<roh>
no, but shitloads of cables, input devices, etc. and a notebook/tablet regardless of what you add on features.
<wpwrak>
you're describing a scenario where the milkymist doesn't add value
<wpwrak>
milkymist offers integration. if you assume that people are happy with an IT/multimedia device zoo and a killer laptop on top of it, then they probably don't need milkymist
<roh>
less devices isnt everything. it already would mean 'no usb-midi/usb-dmx' interface if you can make the mm do that from ethernet
<wpwrak>
nothing attractive speaks ethernet
<wpwrak>
do you even have ethernet cables at the dj/vj desk in your clubs ?
<roh>
yes. at an increasing number
<wpwrak>
how many out of how many ? :)
<wpwrak>
and how much time does your average visiting VJ spend setting up their equipment ?
<roh>
i dont have numbers. and _a lot_ of time.
<roh>
any reduction there is your product-gain
<roh>
also the part of 'realiabilty
<wpwrak>
yup. it all derives from simplicity
<wpwrak>
integration increases simplicity (for the user)
<roh>
the latter is what i see as the strongest point. control-sw on notebooks fails. setups on notebooks fail. if you can reduce the notebook part to 'a working webbrowser' you increase the productivity from the start a lot.
<roh>
it makes the 'control device' much more replaceable with ad-hoc equipment
<wpwrak>
then you get "plugin the wrong version" ;-))
<roh>
in case of a notebook you need... one ethernetcable. in case of tablet you need.. an accesspoint.. which is as big as a pack of cigarettes nowadys (you know fonera?)
<roh>
wpwrak: dont make anything depend on plugins then. just use what works in chrome, ff and it will work.
<wpwrak>
remember that you still need interactive graphics. and fast user input.
<roh>
wpwrak: the user interface is where setup happens, not rapid timing critical stuff. thats what dmx/midi/osc is for
<roh>
a web/external interface would even improve that because you COULD work on stuff, maybe even another patch for the next set while one is rendering
<wpwrak>
you'd want to combine the control functionality - reduce the number of devices
<wpwrak>
so you may have one MIDI controller you really like
<wpwrak>
(or none, if you're cheap or feel too avantgarde for anything but a touch screen)
<roh>
and depending on the interface to that webinterface (imagine xmlrpc or so) you can also connect client-running native apps for stuff like android and such (but i would let 3rd party/userbase do the development there)
<wpwrak>
and then you do the rest with the surface you already have for administrative tasks
<wpwrak>
(3rd part apps) your complexity increases again
<roh>
often you only need to 'switch to the next patch' and for that even a webinterface is fast enough (one click)
<wpwrak>
correct
<roh>
if you need more, you have a midi/whatever 'realtime' surface anyhow
<wpwrak>
but you'll also have a bunch of "minor" controls with no good place to go. they'll still have to be responsive, though
<wpwrak>
i don't see the current scarcity of controls a permanent feature
<roh>
true. thats why modern webinterfaces are using so much JS... moves the 'app' to the client side
<roh>
reduces response time, reduces needed ip packet traffic
<roh>
i really dont see an issue there. i can even imagine a responsive, interactive editor for mapping setups that way
<roh>
maybe i should write that proposal down, with some graphics
<roh>
point is: it would not need any hw changes, so we should discuss it seperated from the hw stuff
<wpwrak>
i think adding an android tablet for prototyping would be a good idea
<wpwrak>
see how it goes
<wpwrak>
find out if you can avoid wireless and do everything (power and comm) with usb
<roh>
wpwrak: we can do that later (sw issues only)
<roh>
just make it work via ip
<roh>
android has a mode to do ip via usb afaik
<wpwrak>
IP is good. there are also things for OSC, so you have a not to terribly hard start (as far as controls are concerned)
<larsc>
haptic feedback with gentle electro shocks through the button surface
<kristianpaul>
nah, for example a fast blinking led if rescue mode entered
<wpwrak>
beware of metallic buttons caps with wires on them
<kristianpaul>
je
<wpwrak>
it should probably just say "rescue mode" :)
<wpwrak>
well, that could of course go wrong. considering that vga may have troubles as well. (could be the reason why we go to rescue in the first place)
<kristianpaul>
say, yes !
<larsc>
hehe. when you wrote "_say_ 'rescue mode'" it was thinking of speech synthesis
<kristianpaul>
we need a buzzer ;)
wpwrak [wpwrak!~werner@94-163-231-201.fibertel.com.ar] has joined #milkymist
<wpwrak>
some pcs had/have voice output for bios messages. rather unsettling when you boot at 3 am and your stereo shouts at you in a distorted female robot voice "SYSTEM BOOTING"
<larsc>
:)
pablojavier [pablojavier!~pablo@186.19.202.221] has joined #milkymist
pablojavier [pablojavier!~pablo@186.19.202.221] has quit [#milkymist]
<wolfspraul>
at an MOQ of 60pc, I got the following prices from Comtech Shanghai (official Xilinx distributor in China):
<wolfspraul>
XC6SLX45-2FGG484C 44.31 USD
<wolfspraul>
XC6SLX45-N3FGG484C 51.27 USD
<wolfspraul>
XC6SLX45-3FGG484C 56.34 USD
<wpwrak>
-N3 is 15% more expensive than -2
<wolfspraul>
XC6SLX75-2FGG484C 81.02 USD
<wolfspraul>
rest even more
<wolfspraul>
and -N3 is quite rare
<wolfspraul>
I think we should stay with -2 for the time being, we have so many other good opportunities to spend our energy than on trying to upgrade to -N3 or -3. also it's clear that slx75 is too expensive, especially given that we simply don't need it right now.
<wolfspraul>
the SLX45-2FGG484C could well be less, it's a high volume chip and they like to negotiate a little. we paid 39 USD up to now...
<wolfspraul>
but the -N3 at 51.27 is definitely un-negotiable, it has to be ordered just for us
<wolfspraul>
so the difference from -2 to -N3 is more like 10 USD or more, plus being rare, plus creating one-time and ongoing support effort (to support both -2 and -N3 variants)
<wolfspraul>
just sharing data points, no action take on this right now anyway
<wolfspraul>
taken
<wpwrak>
ah. "special order" is a no go :)
<wolfspraul>
sure, sooner or later we will be back at the -2, which is the most common variant of SLX45
<wolfspraul>
also in that FGG484 packaging (C=commercial, also the more common one)
<wpwrak>
(back to -2) yeah. or suffer :)
<wpwrak>
likely, both. because you can be sure that something or someone really needs those extra 20-30% speed.
<wpwrak>
(needs) or, rather, "will have developed a need"
<wolfspraul>
well you know how it goes. "the best" is attractive, even if the comparison is relatively random.
<wolfspraul>
when you go to your dentist, you want him to have "the best" (currently available) x-ray machine
<wolfspraul>
the idea that somehow you are not worthy of it and your treatment is being done on 2-year old equipment feels wrong to people, even though it's so random and they didn't say a word 2 years ago either :-)
<wpwrak>
i'd actually prefer her to have a machine she knows to use :)
<wolfspraul>
yeah but you know what I mean. few people have truly absolute criteria in mind, most go with the flow.
<wolfspraul>
our entire economy is run on that, it's what finances innovation
<wpwrak>
x-rays haven't changed a lot in the last decades. the imaging has a little. but whether it's on film or on a TFT, duh :)
<wolfspraul>
so yes, the -3 has a premium
<wolfspraul>
the -N3 is a special case, only few customers use it according to Comtech Shanghai
<wpwrak>
(go with the flow) yeah, in general we assume that newer things are better. so without better information, that's what we tend to prefer.
<wolfspraul>
sure
<wpwrak>
-N3 seems to be a hack
<wolfspraul>
imagine a car salesman trying an "the car you buy today was *BY FAR* the best 5 years ago" argument :-)
<wpwrak>
"we couldn't get the memory controller to work at that speed, so we just sell the critter for a little less"
<wolfspraul>
it would make people's minds hurt, the few that actually cared to think about that argument
<wolfspraul>
most will still just want "the best", "the best of today" :-)
<wolfspraul>
yes Xilinx indeed had problems with the memory controller, I think I've heard that a long time ago when they had many months delay in moving from spartan-6 sampling to production
<wpwrak>
oh, i think comparison should be with choices of all ages, including the brand-new latest thingy. you just can't assume that the latest model is the best.
<wpwrak>
heh :)
<wolfspraul>
so the yield may just come out like this and -N3 is a chance to sell the chips at a few dollars more than what they could otherwise get selling them as -2 variants
<wpwrak>
did you also inquire about that arctix[sp?] ?
<wolfspraul>
not yet
<wpwrak>
artix
<wolfspraul>
but I know for sure it will be many months out until it's available
<wolfspraul>
and we are a tiny customer, and calling sales reps that have never heard of us before
<wolfspraul>
not a good idea to ruin their patience first thing
<wpwrak>
(-N3) IF they'd work as -2 ...
<wolfspraul>
I could ask about artix-7, and I know I won't get a sample even for at least 6 months, maybe more like 12
<wpwrak>
okay. way out then :)
<wolfspraul>
I will ask when I make the first actual purchase, even if it's just a few hundred USD
<wpwrak>
sebastien seemed to be restless about it, so i thought it was much closer
<wolfspraul>
I would think the first artix-7 samples are floating around right now, and as usual they go to the larges customers first, etc.
<wolfspraul>
no need to be in that battle right now
<wpwrak>
yeah. asking when ready to buy is what i prefer to do as well
<wpwrak>
let them do the testing ;-)
<wolfspraul>
sure
<wolfspraul>
we can get process, tools, testing and so on in shape on our end
<wolfspraul>
that will all reduce time overhead *on our side*
<wolfspraul>
not just others are slow...
<wolfspraul>
it was an embarassing experience in the end to force Samsung out of 6 precious early high-end chip samples, and then wasting months and months on a totally overloaded and incompetent (for the task) engineering team
<wolfspraul>
what a waste
<wolfspraul>
not again
<wolfspraul>
there's a reason the early samples go to the largest and most trusted customers, let it continue that way
<wolfspraul>
like you said, it's even better for us if larger and more competent teams get them earlier