Wilba

I have one too, in AU.

You have not proven that SIDs work in each SID socket. That was the point of running testtone on each Core with SIDs installed. Bridging pins 8 and 27 only validate the audio buffer is working and that any missing audio is not due to a blown transistor etc. i.e. that a known good SID should work in that module. It does not test if the shift registers are working, and that the Core->SID data connection is working. Run testtone app in each Core/SID module pair WITH KNOWN WORKING SIDS INSTALLED! Then for Core/SID module pairs that do not work with SIDs, run the SID interconnection test app. It is advisable to have an LCD connected to the Core when you run this, so you know exactly which pin is being tested. Since you believe Core 1/SID module 1 are working perfectly, mark the 74HC595 used in that SID module as "good", later you can swap them into other modules to see if they work, AFTER you've done continuity tests on each IC socket, because now I'm suspecting you probably have bad ground connections on the 74HC595 sockets or some other short/bad joint.

OK, this is what does not make sense. 1. If you've uploaded using "feedback from Core" to PIC #0 in Core 2, then MIDI In to Core 2 works. 2. If you've run testtone in Core 2 and hear output from SIDs, then Core->SID is working 3. If you've run MB-SID on PIC #0 in Core 1 and can hear notes from MIDI events on channel #1, then this SHOULD WORK exactly the same if you move PIC #0 into Core 2, since the only difference between Core modules on the PCB is that Core 1 connects to banksticks and DIN/DOUT modules and the other Cores do not. So you need to work out why Core 1 and Core 2(,3,4) are different somehow. Given the above report, one explanation would be that your Core 2,3,4 are not receiving MIDI In... perhaps you think you're uploading when you're not really? Perhaps you have "feedback from Core" turned off in MIOS Studio? Unlikely after this much troubleshooting, but still possible. The other explanation is the MB-SID app is just not happy running on the other Cores. The testtone app runs, but perhaps because it doesn't access banksticks or DIN/DOUT modules, it is not affected by some hardware difference, like a short on the IIC pins (J4) or DIN/DOUT pins (J8/J9). Try to really really really prove that MIDI In works in Core 2. Using MIOS Studio and an LCD connected to Core 2, try sending text to display on the LCD using the debug utilities. Or even just alternately uploading testone and MB-SID apps (i.e. you'll hear testtone and then not, I assume). Do that test and I'll think some more in the meantime. I may have to use the Chewbacca defence.

I haven't actually finished building an OPL3 module with the chips I've sold already... and since Sasha had trouble with the ones I sold him, I may not run another bulk order until somebody confirms these chips actually work!

Just to clarify, please point out how anything I've posted: ... gives the impression of a "package deal"? It was clear from the very beginning that the ICs were a separate bulk order, run by myself, paid for in advance by me with my cash and then selling them at cost. Doug said he had 200 lying around which was conveniently enough to cover the orders from the U.S. You don't see me posting here having a bitch about people who put their names down on a bulk order and then change their mind, or don't pay their PayPal invoices or want me to hold their order until they decide whether they can find the parts for the PCBs they have already. Oops, I just had a bitch about it. Man, I'm such a hypocrite. Please advise how I can be Mr. Nice Guy and help people out, without them thinking I'll make their life so easy that they don't need to do anything for themselves. </rant>

I'm using a combined LED and switch matrix. One 74HC595 is used as a current sink for 8 "columns" of LEDs and switches. One 74HC595 is used as a current source for 8 "rows" of LEDs. One 74HC165 is used to sample the state of 8 switches in one "column". The remaning 5 74HC165 are used to sample the encoder states. Four of these are close to the step encoders (to minimize traces), the fifth is on the far right and connects to the menu encoder. If you use encoders with switches, then the 16 step encoder switches (wired in parallel) and the menu encoder's switch are connected to this 74HC165. When putting switches in a matrix, you need to put a diode in series (in this case, cathodes are connected to the "column"), similarly the LEDs are connected with common cathodes. The diodes prevent false triggering if more than one switch is pressed at a time - without them, the current could be sinking switches on more than one column. DIN/DOUT on the same PCB is good, but quite hard to layout, it takes a long time to find the best places to fit the ICs and then even more time optimizing the traces, thus you may start out with switches and LEDs at certain positions in the matrix and then swap them around. For example, the 16 step switches and 16 step LEDs are connected to the 8 matrix columns in a pattern - each column has two switches and two LEDs, but the left and right sides are in a different order. Similarly, the switches roughly below each pair of step switches are connected to the same column of the matrix, but thereafter the row connections are not always regular. I'm not sure which resistors you say are missing - if you mean for the LEDs, then you only need one resistor per matrix row (i.e. connected to the 74HC595) but I've also used two sometimes, i.e. the 74HC595 connects to two resistors, one leading to some LEDs on the right side. If you mean missing resistors for the DIN, there are 6-pin 10K resistor networks next to each 74HC165. BTW it is the same kind of switch/LED matrix I used on the MB-6582, with the exception that I do not use transistors for the columns, since the MB-6582 has two 74HC595 supplying a 16x8 LED matrix, the SEQ only has an 8x8 LED matrix. So you can see some kind of schematic in the wiki (http://www.midibox.org/dokuwiki/wilba_mb_6582 see "Control Surface Wiring"). Note the JDx connections are (essentially) 1:1 with DIN/DOUT D0..D8 pin outputs. If this is your first time designing PCBs, perhaps it might be easier to do it like the MB-6582 control surface - don't put the DIN/DOUT on the PCB, just do the matrix on the PCB and connect it to external DIN/DOUT modules. For an MB-SEQ (potentially 17" wide!) there will be plenty of room for DIN/DOUT modules in the case, so the matrix will just mean you need less DIN/DOUT modules. I've attached the matrix pin assignments, I assume this is correct, but it's not really that important if it is not - you will need to define your own, but you could use this as a guide - i.e. see how I've arranged switches into columns to sort of match their horizontal location. If you match the GP LED assignments, that would be ideal, because then you can use the same code in MB-SEQ firmware. MB-SEQ matrix pin assignments.zip MB-SEQ matrix pin assignments.zip

It doesn't matter which are stuffed... you might want to connect it to the last bank used for patches (address=6, bank G) as the one at address 7 is used for ensembles. I don't think it's worth it. Swappable banksticks were useful when only one bankstick could be used. Now you can have up to eight, so there's no need, unless you want to quickly move patches between two MB-SIDs, and even this is more conveniently done through the patch editor or even SysEx dump between two MB-SIDs.

Good to hear you got it working! Congratulations!

Make sure you are installing the setup_mb6582.hex, the other firmware will not work with the switch matrix. But I think if the menu encoder works then you probably have done this correctly.

Please describe what you have not soldered or connected. The switches rely on all the 74HC595 installed, the resistors and transistors on the far left installed, and connected to the CS PCB. Also all the diodes need to be installed. If the menu encoder works, then you've obviously connected ground to the CS PCB (far right), as this requires ground. The switches do not require ground because they're part of a matrix (in a sense, each column of switches is "grounded" by the 74HC595 and one transistor at a time). Also, all switches need to be connected because some tracks use two pins of a switch that are internally connected to bridge over other tracks.

"Option C" was a last-minute addition to the base PCB, as a possible workaround - that there IS a place for a 5V regulator and associated capacitors. Unfortunately, the current requirements of using a single supply (i.e. powering the 5V supply with the 9V supply) might be too much for the regulators, especially the 9V regulator, which will have the entire 5V supply load going through it, in addition to the 9V supply to the SIDs and audio output buffers. Let's say as a rough estimate, the 5V supply requirement is 1000mA and the 9V supply requirement is 320mA, for 8 SIDs, 4 PICs etc. and a control surface. The 7805 can handle 1000mA, but if this being supplied by the output of the 7809, then that's 1320mA going through the 7809, which has a rated current limit of 1000mA. So my suggested solution probably is a bad idea. It might be better instead to connect the output of the bridge rectifier (J25:9V-11V) to the 7805 input (J74:2) and then the 7805 output (J74:1) goes to the 5V supply (J4:2). However, then you've got another issue, the 7805 will be regulating 1000mA and dissipating heat from a 4V-6V difference between input and output voltages... not too bad, but that will require a serious heatsink and is hardly the most ideal solution. These are perhaps all worst case numbers, and it might not get that hot... I don't really know because I have not done this before. If you do intend on this plan, then if you can find yourself a 1.5A 9V AC adapter, just connect it to the base PCB through the DIN socket by using a DIN plug... i.e. cut off the existing connector on the adapter and solder it to a new plug, or make an adapter (find a female socket for that adapter). J71 and J72 should be bridged as if for PSU Option A.

toneburst: you are not testing things properly. Each test is supposed to validate what does and does not work. You can't do that with different apps on PICs, and now it's getting hard to follow what does and does not work. So... start again from scratch. Do the following, reporting success or failure. Take out ALL PICs. Take out ALL banksticks. Take out ALL SIDs. Set J11 to 1. Upload MIOS to PIC #0 in Core 1. Set J11 to 2 Upload MIOS to PIC #0 in Core 2. Set J11 to 3 Upload MIOS to PIC #0 in Core 3. Set J11 to 4 Upload MIOS to PIC #0 in Core 4. USING ONLY TWO SIDS MOVED EACH TIME: Use testtone in PIC #0 in Core 1. Use testtone in PIC #0 in Core 2. Use testtone in PIC #0 in Core 3. Use testtone in PIC #0 in Core 4. Set J11 to 1. Upload setup_mb6582.hex to PIC #0 in Core 1. USING ONLY TWO SIDS MOVED EACH TIME: Test MB-SID in PIC #0 in Core 1 (send MIDI notes on channel 1) Test MB-SID in PIC #0 in Core 2 (send MIDI notes on channel 1) Test MB-SID in PIC #0 in Core 3 (send MIDI notes on channel 1) Test MB-SID in PIC #0 in Core 4 (send MIDI notes on channel 1) This should validate that the MB-SID app is happily receiving MIDI in each Core. Set J11 to 2. Upload MIOS to PIC #1 in Core 2. <-- Change Device ID in MIOS Studio! Upload setup_mb6582.hex to PIC #1 in Core 2. <-- Change Device ID in MIOS Studio PUT TWO KNOWN GOOD SIDS INTO SID MODULE 2 Test MB-SID in PIC #1 in Core 2 (send MIDI notes on channel 2) By default, PIC #1 will be configured to receive on channel 2. PUT TWO KNOWN GOOD SIDS INTO SID MODULE 3 Test MB-SID in PIC #1 in Core 3 (send MIDI notes on channel 2) PUT TWO KNOWN GOOD SIDS INTO SID MODULE 4 Test MB-SID in PIC #1 in Core 4 (send MIDI notes on channel 2)

very OT... keep your troubleshooting in one thread please

beaten by years: http://www.midibox.org/dokuwiki/doku.php?id=wilba_mb_6582_base_pcb_construction_guide http://www.midibox.org/dokuwiki/doku.php?id=wilba_mb_6582_control_surface_construction_guide

YES and I regret it now. Do you think I would say "Do not use blue LEDs" for no reason? I was most annoyed when I first connected the CS and realised how overpowering the blue LEDs were... and mine were diffused AND in a tinted blue package. I had to use 3K resistors just to get the brightness down to a tolerable level. In retrospect, I should have just replaced them with green LEDs instead of keep it this way... but I hate desoldering and the CS was already attached and I just wanted to play with it a while. Then a combination of being too busy with SID sales and other things meant changing the LEDs was very low priority. It is the prototype after all, and prototypes are meant to teach you what worked and what didn't. BLUE LEDS DO NOT WORK! DO NOT USE BLUE LEDS! The most stupid thing you can do is use blue LEDs just because you have some already, because then you're basically considering the cost of using new ones as more important than the advice given here, and the time and investment you've already put into your MB-SID.

Do not use blue LEDs.

Higher level of background noise suggests that you're not grounding the inputs (jumper in J3_SIDx, J23_SIDx). Can you please confirm that you've tested that each SID and each SID module works, by using the testtone application in each SID module and swapping SIDs. Uploading an app to each Core (by moving jumper in J11) validates that MIDI In/Out is working for each Core. That would basically leave just the communication between the PICs not working. Look closely at the PCB PDF, it will show you where the CAN bus connects pin #36 of all PICS, and connects to pin #35 of each PIC through a diode. There is also R80 which connects the CAN bus to +5V. Check all pads connected to this. Check the PIC pins are connected and there's no short between the CAN bus and +5V or ground.

Yes. Unless something else happened to change the MIDI channels. Try removing all banksticks and uploading the MB-SID application again to all PICs. This should reset it to a default state. Yeah, slave Cores don't write stuff to the LCD, only the bare minimum "I'm alive and this is what I have installed".

The PICs should have device ID (PIC ID) 00, 01, 02, 03. If you got them from SmashTV they should be like this already. You can put any PIC in any Core and it will work, so you can use one PIC with the testtone application to test the other Core/SID slots. After, you can then upload the MB-SID application to a PIC in each Core, using J11 to switch which Core is connected to MIDI Out. All Cores are connected to a common MIDI In. The default "ensemble" should have each SID engine (Core) configured to MIDI channels 1,2,3,4. Only the first Core makes the startup jingle, because it's actually the sound it makes when it finds a bankstick. Likewise, you could put PIC #0 with the MB-SID application in any Core and send it note events and you should get the default patch playing on the output. You probably already validated that each Core is receiving MIDI In by uploading testtone to each Core rather than moving a PIC around. However, TK just changed the note handling for Super Poly mode so I'm not sure if that change has meant that ALL notes are routed through the master PIC to the slaves via the CAN bus. Assuming you've already done all this (used testtone in every Core/SID combo), then the only thing I can think of is that the CAN bus that runs between them is not working. Check you have R80 installed. It is between U1_CORE1 and U1_CORE2 (first and second PICs) Check also the diodes under each PIC are correctly oriented... cathode end (black mark) should be lower pad.

I just want to add, that my motivation here is to help people build something ASAP, but without the problems of finding parts I used. I could run a bulk order now for just the PCBs, it would be very quick and easy, but then there would be 100+ people all needing to buy the same small quantities from the same places and pay multiple shipping costs, etc. in addition to still needing two kinds of knobs from bulk orders, and the panel/case.

Well... that was the plan. Since I would be doing a huge order from Mouser for at least the bicolor LEDs, resistor networks and capacitors (small axial ceramic 100nF), it sort of makes sense to get the other parts, rather than expect people to search for local suppliers of these parts (or bother SmashTV). However, that would almost triple the amount of time I would spend packing, because IC sockets and ICs need to be packed in tubes, involving a lot of cutting of tubes and repacking them. Each PCB requires 2x 74HC595 and 6x 74HC165, 8x 16-pin IC sockets. If people didn't mind them just tossed into a bag, it would be fine :) I don't mind doing careful repacking of SIDs because they're special... I'm just not that motivated to do the same for such common parts that even I can buy them here in Australia at reasonable prices. So... assume I am not getting this other stuff. If you want to get the other stuff ahead of time, because you're ordering parts for other projects, then it's probably a good idea, I will not be forcing people to buy these parts again when I run the bulk order and change my mind and have all this other stuff for sale. The missed savings will be so small that it won't matter. I've updated this page with quantities: http://www.midibox.org/dokuwiki/doku.php?id=wilba_mb_seq_parts_guide Keep in mind that the panel mounted parts (sockets, switch) are still not decided. I may switch to an additional I/O PCB so we can use PCB-mount sockets, which might work out cheaper and have the added benefit of a PCB-mount USB socket.

The prototypes have been tested. TK's prototype was actually a second prototype with some improvements, and from the demo he just posted, it obviously works well and is approved by the big man himself. He mentioned he really likes these switches and knobs. I think I said before that there is no point doing a bulk order of PCBs until Doug's case design is ready, and he has time to make them... because I think nearly 90% of people are wanting to use his case and get all the required parts from me. Similarly, you can get parts now if you want, but there is not much point because I will be arranging to sell nearly all the required parts too, at cheap prices. The only thing I won't be selling are the 2x40 LCDs, and even Doug has hinted about arranging a bulk order of those in red/black (like my MB-6582). If you have specific parts that you need quantities for, just ask.

I am so liberal with the JB-Weld for the corner screws that I need to widen the mount holes on the PT-10 case, sort of making them countersunk. This is easy to do and will not affect anything because sideways movement is restricted by the recessed edge of the PT-10 case front, and you can add a washer between the case mount and the nut holding the panel onto it.

I have been sufficiently teased, vielen dank.

I third the regulator. No way it's the SID. :) I don't think ICs temporarily go into high current drain mode... it is thus more likely the supply is failing, which means the regulator is most likely the problem. Maybe you're supplying the 7805 with a high voltage like 12V and expecting it to shed 7V at 500mA = 3.5 watts of heat.