Circuitbenders Forum
Circuitbenders Forum => Circuitbending discussion => Synths & Samplers => Topic started by: Circuitbenders on April 13, 2010, 03:05:21 AM
-
Now heres a very strange one. I finally got around to cracking open my Akai S20 and located the RAM chips (which are a pair of HM514260CJ's), and with the help of the datasheet got some excellent stuff going on by shorting the address and I/O pins together as per usual.
Looking at the thing i concluded that theres no way to mount anything on the case so i cut a hole in the back, mounted a 25 pin socket and then wired all the useful pins to the socket for an external patchbay. Unfortunately as soon as i plugged in the connection cable the s20 suddenly started acting like every single one of the pins was connected together, even though there was nothing on the end of the cable. I've tried a few different cables and can only conclude that the individual wires are acting like aerials and somehow interacting with each other inside the cable.
What makes this even stranger is that a lot of the RAM chip pins, specifically the I/O pins, have no effect when shorted together with a wire longer than 5 or 6 inches, but with a wire of about 2 or 3 inches there is a minor effect, and when shorted with a half inch piece of bent jumper wire there is a full on effect. It seems like the resistance of the length of wire is actually cutting the level of the effect somehow!
Has anyone got any explaination for what is going on here as it seems completely inexplicable to me. It actually looks like i'm going to have to give up on the external patchbay idea and just mount a few switches as close to the RAM chips as possible just to get them to work! :-\
-
hm yes, irritating problem.. i think i had something simmilar with another device..
didn't you had something like that with an sr16 before ?
maybe you can make use of a 25 pole switch just before the connector.. like the ones in printer switch boxes.
-
I think the SR16 problem i had was with the cables picking up noise from other parts of the circuit. In this case its not so much noise as long unconnected cables acting like they are connected, and the length of connected cables apparently creating too much resistance for the connection to work.
I'm going to have to come up with something more cunning here :-\
-
thats too bad. i dont know much about this kind of thing but i do remember seeing in a lot of old things from the eighties had capacitors going from each io pin to ground before going to the joystick port. maybe it would help to try adding .001uf caps to each of your 25 io's? couldnt hurt to try.
-
Well, yeah it *could* hurt to try, since the pins are switching very, very quickly. You may find that stray capacitance in the big long dangly leads is causing all sorts of weirdness. Adding more capacitance won't help.
As an aside, I once had an Amiga in to repair that had a corrupt display. The memory had been upgraded, but whoever had upgraded it had used the wrong speed of memory - 150ns instead of 100ns (or similar). This meant that single pixels would randomly smear into blocks three or four pixels wide, because the chip switched far too slowly.
-
Well heres some picture of the finished thing, i'll start another thred when i've taken some samples with some more details of the mods.
(http://www.circuitbenders.co.uk/IMAGES/S20/S20main.jpg)
(http://www.circuitbenders.co.uk/IMAGES/S20/S20switch.jpg)
-
hey
i'm presuming you managed to find out what was wrong with it and why the length of the wire was messing it up?, would be interested to know as i have a couple of these floating about waiting to be victimised.
cheers
ian
-
I assume the issue was something to do with what gorden was saying there i.e. capacitance in the cabling. For some reason you can use longer lengths of wire if you use wire with bare copper conductors as opposed to tinned copper conductors.
What i ended up doing was using 12 SPDT centre off switches with the centre pin connected together between the switches as shown in the diagram below. When you switch a switch up or down you connect one of two bend points to the blue 'bend bus'. This way you can connect up to 24 different points together in thousands of different combinations without having to use a patchbay.
(http://www.circuitbenders.co.uk/IMAGES/S20/bendbus.jpg)
-
excellent thanks Paul, good work.
-
indeed excellend.. i like a lot your bentbus interface :)
-
Thats:
BEND BUS™ ©2010 circuitbenders.co.uk (patent applied for) ;D
Sorry.................. for a moment i thought i was s-cat there ::)
-
Theres a mp3 of the S20 grinding up a breakbeat and a vocal here:
http://soundcloud.com/circuitbenders-co-uk/circuitbent-akai-s20-demo (http://soundcloud.com/circuitbenders-co-uk/circuitbent-akai-s20-demo)
That vocal sample was on the disk i got with it when i bought it. I'm guessing it used to belong to some kind of theatre or music venue.
-
Feckin hell, if i ever needed reminding why i got into circuitbending :D
http://soundcloud.com/circuitbenders-co-uk/circuit-bent-akai-s20-demo-mk2 (http://soundcloud.com/circuitbenders-co-uk/circuit-bent-akai-s20-demo-mk2)
-
I'm really coming to love this little sampler-even before bending-it's so simple. Did you end up just using the chips near the SIMM for your bends or did you have any other points?
BTW, as I mentioned on Facebook, X3 is the pitch mod crystal and it's AWESOME.
-
Yeah, the two big (for surface mount anyway) RAM chips next to where the extra SIMM slots in. Soldering to the chip pins isn't much fun if you have shaking hands ;)
The datasheet can be found here http://pdf1.alldatasheet.com/datasheet-pdf/view/118617/HITACHI/HM514260C.html (http://pdf1.alldatasheet.com/datasheet-pdf/view/118617/HITACHI/HM514260C.html)
Any chance of a demo of the clock mod if you've done it?
-
Y'know, you could desolder the chips off a scrap SIMM and make a pluggable bend controller...
-
I might end up JUST doing the clock mod. I hate making leads off of SMD stuff. I'll definitely put it up when finished (ugly pictures and all). I've been putting LTC1799's in almost everything I own these days, I'll have to put together a demo of my otherwise unbent pieces: Electribes EA-1, ER-1, EM-1 and now this sampler. I love that it's tempo independent for midi.
Amazing how low you can clock it down. I had the clock crystal down to the Kilohertz range and it was still sampling.
I don't know why more old school samplers didn't do that: Sample at a relatively low rate and then use pitch shifting to effectively increase your sampling time with a quality trade off. You get a lot more time and for grit, it's pretty useful. I know that a lot of manufacturers were trying to get as high quality samples as possible, but it's still interesting.
-
If you listen to a lot of early Public Enemy and similar stuff. you can actually hear the aliasing whine on samples. Listen to the main hook of "By The Time I Get to Arizona", for instance.
Now, what they did there to squeeze longer and longer samples into the cramped memory of the samplers of the day was, they played the record in at too high a speed, or played it in off tape at twice the speed. Then when it was pitched down they had it at the right pitch albeit with reduced quality.
-
I've finished the S20 mod and finally recorded a demo of what it can do:
Akai S20 modded with an ltc1799 pitch mod (http://www.youtube.com/watch?v=Q22_oxeIDlo#ws)
-
:othat's impressing !
-
Thanks! I normally hate thread jacking, but I had forgotten that youtube links automatically add the video in the forum software.
Gordon, you're suggesting to completely desolder all of the memory + related chips off of a simm and just connect bend points to the equivalent address points?
-
Yes, just wave a heat gun at it and the chipsshould just fall off. You can then solder to the pads, or make a SIMM with jumpers for "preset" bends ;-)