Wilba

Maybe you didn't put a jumper in J11 to select which Core is connected to the MIDI Out port. Thus you're not receiving the upload request from the Core. Note also you'll need to change the device ID in MIOS Studio to target the upload to each Core (the PIC with that device ID already set). If you get PICs from SmashTV, they will already have ID 0,1,2,3 set and printed on the label.

Are you crazy? Desoldering braid/wick is awesome!

To be honest, I wouldn't bother with heatsinks on 8580R5 or 6582A, they just don't get hot, only warm. (Well, mine don't get hot enough, perhaps others do). 6581 on the other hand are hot little buggers and do get hot... In one model of C64 (a beige one I expected to contain an 8580R5)... a big piece of folded sheet metal surrounded a few of the ICs including a 6581R4AR and made contact with some conductive silicon goo. Perhaps the very high failure rate of 6581 SIDs when salvaged from the brown breadbox C64s is partly due to them running hot with no heatsinks. I do like these heatsinks, though... I originally planned to stick on some nice copper finned ones meant for PC memory but went off the idea of (semi-)permanent gluing onto rare 6582A (that was before I found the stash of 4000+). So removable heatsinks are cool. It can't hurt and can only help prolong the life of your SIDs, but certainly isn't required, much like the 40mm fan is not required on the MB-6582.

I'm only running a small bulk order for people willing to make/buy their own frontpanel and case. "Small" for me is only 36 PCBs and less than US$1000 in parts... ;) This small bulk order is closed, it was announced months ago and people have had plenty of time to join. Once Doug is able to make panels/cases as per the original design, I'll run another bulk order for more PCBs and parts.

[img height=345]http://media.ebaumsworld.com/mediaFiles/picture/404456/80456372.jpg

I think Darlington transistor arrays are faster than relays, and the DOUT module even supports these ICs directly, the output pins become high current sinks instead, so the negative pins of all your mechanical stuff is connected to the DOUT pins and the positive pins connected to whatever supply you want. Didn't you already do stuff like this with old hard disks?

Cutting tracks is easy and in this case pretty safe... as they are pretty fat tracks and there's no other tracks nearby you could accidentally cut. I typically use a craft knife to make a deep scratch, then use a small flat screwdriver to scratch deeper until it cuts through the copper. You can test with a multimeter to make sure it's disconnected and also make sure it's not shorting with nearby copper (in this case the ground plane). Duplicating the backlight circuit elsewhere could be done... you'd have to disconnect the B+/B- pins on the LCD (typically labelled A/K at pins 15/16) and use separate wires from the LCD so it's not connected at J15_CORE1. I just realised though, you don't need to break the connection at the LCD to test it's current draw, if you're doing the fix on the PCB, you can test current through the new wire where it's connected to the different supply.

In case you are interested, you can supply the LCD backlight with a different supply to the +5V... i.e. if you have spare current on your +9V supply, you can use that instead of using the +5V supply. Most LCD backlights draw about 280mA current so it can help a lot if your +5V supply is sagging. Note that each SID draws 100mA on the +5V supply, so there's 800mA just powering 8 SIDs! If you're interested, all you need to do is cut the tracks around the B+ pin so that it is disconnected from +5V, then connect that pin to the new supply, be that +9V or +12V or even pre-regulated supply, it doesn't matter what the exact voltage is as long as it's sharing the same ground. On the MB-6582 PCB, the +5V supply passes through this pin to the components on the bottom of the board, so you also need to reconnect +5V between the P2_CORE1 pot and P1 pot. The backlight circuit basically is a current limited sinking transistor that sinks a constant current through the B- pin, constant for a given voltage that is. The transistor is amplifying the current at it's base, so you need to keep the +5V supply through P1 and through R4 so it's the same current going into the transistor base, and thus the same amount of current being sinked by the transistor's collector. In my tests with low-power LCD backlights which only need 25mA, switching from +5V to +12V on the B+ pin changed the current from 25mA to 30mA (I have increased R4 to achieve this lower maximum of 25mA)... I haven't tested this with high-power LCD backlights yet... it might be relative to current draw so perhaps it jumps from say 250mA to 300mA. UPDATE: with regulated 12V connected to B+, and a 50K (not 10K) brightness pot set to max, current was up around 600mA! and drops to 240mA at around 1/3 the pot range... so if you try this, be sure to test actual current draw and use a 50K trimpot at P1. Also, replacing P1 with a 50K pot can give you more range in brightness, same maximum but a dimmer minimum. If you're interested in tinkering with this idea, I suggest you also add a way of testing actual current draw of the LCD backlight, by breaking the connection where the LCD cable connects to B+ on the LCD, so you can optionally connect your multimeter between the cable wire and the LCD pad and test current. Note that some multimeters have a 200mA mode and a 10A mode, so make sure you don't blow a fuse by testing with a mode that can't handle that limit (assume the backlight circuit might be drawing 500mA to be sure). If you prefer to wait, I'll post some results of my tests with high-power backlights, so I could tell you what works, what's safe, and maybe how to change R4 to set an exact upper-limit on the current draw. Note to other MB-6582 builders - you typically don't need to be doing this if you have a C64 PSU that can happily deliver 1.5A of current on the regulated +5V supply... the "brick" has a massive heatsink and switching the backlight supply to one regulated within the case is only going to increase load on the regulator within the case, leading to (maybe) another 1W of heat that needs to be dissipated somehow, so no advantage really.

Thanks for the beer :)

This is the problem... you may have followed instructions for when you are NOT using the parts provided... i.e. where I wrote "If you are choosing not to use the PCB mounted sockets and switch on the base PCB" etc. Take out the bridges between S1 pins and everything should work.

The red is probably the backlight showing through the LCD. Make sure when it's on and working, you turn the contrast pot... most LCDs look best with the pot turned fully one way (I think it's to ground).... if it's at the other end you won't see anything and think it's not working. One SID not producing tone is a good sign that there's just some bad connections on that SID socket, or something wrong with the audio buffer above that socket. It would seem the PIC connections to the shift registers below both sockets, and the shift registers themselves, are all good.

I said, I don't keep track of the list... YOU add it when you want to give yourself a serial number!

Very unlikely you blew the 9V SIDs unless you had them connected to 12V, and you would not have inserted a SID without checking the voltage on pin 28, right?? :)

Quite possible... it was written before MB-SID V2 and stereo SID support. However, I am very sure the testtone app outputs on both channels.

First make sure the PIC has the firmware correctly uploaded, and perhaps check this PIC in another working Core/SID combination. Then check it is actually running happily in Core 1 (by connecting LCD, looking at MIDI Out log, etc.) Then you can try the interconnection test app (found in the same place as the testtone app). Someone else recently was doing this and updated the wiki, check for recent posts. If the interconnection tests all pass, you can try checking the audio output buffer separately (although I doubt both buffers will be at fault causing no sound on both channels)... you use a resistor to bridge two pins of an unstuffed SID socket... the details should be in the testtone .asm file. One of the control lines to the SIDs is pulsed at 1Khz and you connect that to the SID output pin on the socket, which produces a tone out of the jacks. Since you say you uploaded to all the PICs, it's probably most likely one of the common control lines are not working, or the shift registers underneath (they're both plugged in correctly? getting power correctly?) You'll know for sure when you run the interconnection tests.

Stupid bloody Aussie Post bastards, they're not supposed to leave it on your doorstep.

Serial numbers are self-assigned, I don't have to see proof or anything... the wiki page just makes them unique numbers I guess... http://www.midibox.org/dokuwiki/doku.php?id=wilba_mb_6582_builders_list

I guess I'm too busy to make the SVG files right now, sorry. What little spare time I have at the moment is spent running bulk orders.

Refer to the SID module schematic: http://www.ucapps.de/mbhp/mbhp_sid_v3.pdf It labels which pins on the SID are A0-A4 etc.

You need to know what their maximum forward current and forward voltage drop are, and choose a resistor appropriately. http://www.kpsec.freeuk.com/components/led.htm

Oooh! Canadian beer! Thanks!

the "little lighted" LEDs are because you're using superbright LEDs and not using a big enough resistor. This happens on my MB-6582 when they are too bright... it is because the LEDs are in a matrix and when it switches "columns" of the LED matrix, the LEDs in the column that was just "on" has some lingering current sinking (perhaps the transistor doesn't switch fast enough?) It is not a problem with standard LEDs or superbright LEDs when using a big enough resistor to make them "normal" brightness (i.e. 3K-5K) Why are you burning "device_id_0*.hex"? Did you burn the bootloader yourself with ID=0 and need to change IDs on them? You should not need to do any changing of device IDs if you got PICs from SmashTV. Some info: ALL the PICs are connected to the SAME "MIDI In". J11 only controls which PIC uses the "MIDI Out". So if you put PICs with the same device ID in all sockets, and upload to that device ID, they all will receive the upload and install it, but only one of them will be "sending feedback". Once they all have different device IDs, it's no problem, because then they all will receive the upload but only the one with the matching device ID will install it (i.e. the MIOS Studio setting targets which PIC). So you need to set them to different device IDs... put them each in Core 1 (without any PICs in Core 2,3,4)... give them IDs 0,1,2,3 and then put them in Core 1,2,3,4. Then you can upload the MB-SID app to each one just by changing the device ID in MIOS Studio.

The GOOD NEWS is... your optocoupler is working and your MIDI sockets are working. OK some more questions (just to find out what it might be): Did you get the PIC from SmashTV with MIOS already installed? Did you get the base PCB parts from SmashTV? Perhaps (unlikely) the crystals are not 10MHz so the PIC will boot but it is not transmitting on the correct baud rate. Let's try testing if anything is being sent from the PIC. Insert PIC into Core 1, set J11 to 1, NO jumper in J11_CORE1 (Tx/Rx). Take out MIDI cables. Put a LED in the MIDI Out socket like this: In this case, the cathode (short lead) goes into the pin closest to the power socket/switch, next to the middle pin... the anode (long lead) goes in the other pin next to the middle pin. Make sure the leads are parallel and go at least 5mm or so into the socket (ensure good connection with the socket). Power on. The LED should flash on for maybe 1/10th of a second. Since you have MIOS installed, it should only flash once. If only the bootloader was installed, it should flash every 2 seconds. Repeat this test with ALL the PICs you have, installed into Core 1 (only use one PIC at a time in Core 1). You can also switch from LED to MIDI cable and check again if you get any MIDI In from any of the PICs in Core 1. I'll help more after you report back with results...

*yawn*

So what colour LEDs are you using elsewhere? It is hard to tell from the photo. 10K resistors will work for superbright LEDs, but only the very brightest ones. The problem is, different colour LEDs have different voltage drops, so using the same resistor will mean different current for LEDs of different colour/type. So did you fix the problem with the LEDs that do not turn on?