I can make a board with 4 or 8 PIC on (I think I prefere one with 4 so you can stack them), but it will have alot of wire on it. I will then put then side by side.
The reason to make a board with just 1 PIC was the easy connection between then, you just use a cable with 20 wire and 8 IDC connector (like IDE HD cable) to connect them together without soldering. Only the end going to the core have to be solder.
Look at this:
Let me hear what you think. (keep or make one with 4 PIC)
I will change the MIDI connector so its the same as the core modul.
Hmmm.. the long cables could decrease the max. data transfer rate due to additional capacities, so that I would prefer a 4-slave board. Ok, this could be connected with an IDE cable to the core module (nice idea!)
Nice project, I’ll definitly build one sometime coz at the moment I only have 1in/1out on my card and as Thorsten seems not to stop building new Controllers ;D ;D ;D and stuff this is THE perfect solution for an interface to my PC.
MMM: really astonishing work in that short time and your idea to connect the boards via IDE cable is really great!
IMHO i would prefer the 4-slaves-on-a-board solution. I have some real troubles with some of my older equipment using this ´double-decker´technique. First, its more or less hard to change components, second, the hard wiring tends to produce bad soldering points after a while (ok, some years maybe… ;)). Another point is, you will need much space in z-direction for this construction.
Well I think we (you :P) should go for the 4-slaves-on-a-board solution as I suppose most of us would like to build a 1U 19’’ unit and boards with ´double-decker´technique would most probably not fit in there (1U = 4,4cm).
If you only need 4 slaves the mount 2 SIL 6x100K in R21 and R22.
If you need 8 slaves, use 2 board in ´double-decker´technique (easy fits in a 19" rack 1U case), on board A, connect CON9 to the core, on board B mount 2 SIL 6x100K in COM9 place (there are some extra pad for it).
Make a connection between the 2 boards using R21 and R22 place with 13 solid core wire.
A better explanation and a componentlist will follow later.
Also the picture is scale down.
Kent
PS. Thorsten can I send you a pdf of it, that you can put on your site?
I wished my danish skills were better, but the only words I still remember from my last holidays in denmark are “smörrebröd” and “tuborg beer”
We can publish this circuit officially when it has been verified. I cannot guarantee that it really works before I’ve tested the whole system (however, I don’t expect unsolvable problems…).
Which dimensions did you use for the board? Does it fit on a common euro size PCB (160/100 mm)?
Christoffer noticed that I used a 10 MHz crystal instead of 20 MHz. This was an error in the schematic — 20 MHz are required for the maximum performance.
Sorry, the dimensions does not fit on a common euro size PCB, its 177x77mm.
I may be able to squeeze it to 160mm, but the tracks will be to close for “amateur” to make it.
Btw, your danish is something like my skills in german language ;D
As there are no hardware select to each slave, I suspect that each slave will have a unique id and the core will have to check all slave if one request service, or ?
Something like select a slave by software and the selected one will only be on the databus.
(Just thinking a little about how you will make the software).
please try it, because it will be hard to get a cheap non-euro size PCB. :-/ Maybe it’s better to put CON9 in the middle and to arrange two slaves at the left, two at the right side.
The slaves itself are selected over the data bus when A0=1. Every slave has one dedicated select bit.
For the slave ID setup I plan to provide a seperate MIOS application which initiates a special password sequence to change the ID. Only one slave should be connected to the bus at this time. The slave will save the ID in EEPROM so that this procedure is only required once.
Best Regards, Thorsten.
P.S.: although my plans are clear, I don’t will find the time to realize this application before May. Other projects have higher priority
Making a 2-layer PCB will be mostly impossible to many users. I would not know how to do this with affordable equipment, and the costs for doing this externally (at some place specialized in making PCB boards) will be comparably high… I even think we lose the chance of Mike offering the board if it is 2-layer.
Making a 2-layer PCB will be mostly impossible to many users. I would not know how to do this with affordable equipment,
Making a 2-layer board is possible with your existing equipment. First, drill a couple of holes through the board so the top and bottom negatives (or whatever method you are using) can be aligned by putting a pin or piece of wire through the holes. Then expose/transfer the pattern and etch. When assembling, solder both top and bottom of component leads and, since the holes are not plated through, use a piece of wire or cut-off piece of resistor lead through the vias and solder to the tracks.
Now three or four layer boards are another story. ;D ;D ;D
