Wilba

This could be caused by a short on the control surface PCB. It is more likely to be a hardware problem than software, if you installed the software without any errors. Check your solder joints around IC2, which is between the F3 and Down buttons. Check also if there are any shorts between the switch pins and the heatsink. You might want to also try swapping the two 74HC165 ICs in case one of them is faulty. What happens when you are in the top menu and press the Up/Down buttons? You say Up works... what happens when you are in the top menu and showing patch A003 and press Menu? Does this change patch to A002?

You won't need 74HC595 to test LEDs... just use 5V. It's not worth simulating a 1/8th duty cycle matrix either. If the LED looks good with a resistor and 5V, it should look good when in the matrix. The alternative is soldering LEDs first and testing resistor values by touching them on the pads on the PCB. Unfortunately the spacing of the LEDs doesn't match 100mil spacing machine pin strips, so I can't suggest making the resistors "socketable" :hmm:

I forgot how to upload stuff to the wiki... so I'll just upload here for now with some info. There are two versions of each file... one has no stroke width on cut lines, so that Inkscape will show width/height not including the stroke. The "ponoko" versions have 0.003mm blue stroke which is required for ordering from Ponoko. The 0.9 files are the ones I use now, which uses "vector engraving" for the artwork. The 0.7 files use the "raster engraving" which didn't come out as nice, but if you are doing some other method of labels (i.e. printing), then you can choose to use these instead. If you plan to order from Ponoko, let me check your file before you send it. Sometimes strange things happen when ordering with SVG files... such as a group of objects having it's own stroke/fill different to the grouped objects own stroke/fill which can lead to those objects not being engraved or cut. It is probably safest to ungroup everything in the final file before sending it to Ponoko. In these design files I've made allowances for 0.2mm laser kerf - i.e. laser burns away 0.2mm of material, thus to make the LCD window cutout and the window fit exactly, each is offset by 0.1mm to allow for material burned away by the laser. Download Inkscape and look at the files. Other apps can scale SVG files differently and the dimensions will be wrong. Confirm the dimensions by checking in Inkscape. sammichSID case 0.9.zip

If you're already familiar with the concept of a LED matrix with common cathodes in "columns"... then I'll explain it briefly... there are six "columns" of eight LEDs, and the current limiting resistors are on the eight "rows", and the "rows" are arranged so the order is middle to extents... i.e. one "row" drives the six outermost LEDs. So if you wanted to make the outermost LEDs red and the next outermost yellow and the rest green (like a LED bar VU meter) that's easy, you just tweak the resistors. For LED specs, I use Mouser parts 638-204GD and 638-204ID which are the right size and good brightness. Note the reds are high-efficiency meaning they're a great brightness for low current (which is what you get from a 74HC595)... don't be turned off by the low mcd values, these are bright enough even in a matrix at 1/8th duty cycle!. It's easy to tweak resistor values to suit multiple LED colours, but obviously a lot easier to test this on breadboard first before soldering to the PCB.

I could probably run another bulk order for OPL3 chipsets but I'm too busy now. Maybe if someone volunteered to do all the emailing and collecting of names, addresses etc. and sent me an order list spreadsheet, I could collect the payments and post the orders. Assuming similar wholesale prices as last time, I could sell complete chipsets (1x YMF262 + 2x YAC512) for 12.75 AUD plus 2.50 AUD postage to ANYWHERE for up to 10 chipsets. I would need a minimum of 100 chipset orders to achieve this price, although I'd be willing to do it for 80 orders and keep spares on sale after the bulk order.

The 6N138 should be inserted with the notch on the same end as the notch on the PCB. There is only ONE right way to insert ICs. The other way is wrong and more often than not will kill the IC, as most ICs have power supply pins at pin 1 and the diagonally opposite corner pin, so inserting them the wrong way inverts the power. This is bad. I can't tell if you've killed your 6N138 or not... it looks like putting it in the wrong way would not kill it from having swapped the polarity of power to something. It is starting to sound like there was nothing wrong with the PICs before you started to upload MIOS, and that your MIDI connections to the sammichSID is the problem, causing a corrupted upload to the PIC. This probably didn't (can't?) kill the bootloader on the PIC, but can corrupt the MIOS that was already installed, so that the only way to fix it is to upload MIOS in the first two seconds after powering on the sammichSID. So if you had some MIDI problem when first uploading MIOS, that's still a problem when trying to upload it again. I think detecting the bootloader is only done by receiving the bootloader's upload request message, and not by sending a message to the PIC and the PIC responding. So it's quite likely that you have a problem with the PIC receiving MIDI from the PC... PC starts uploading but PIC doesn't receive the messages correctly, sends back error messages or no acknowledgement messages etc. So the problem could be at the PC side (MIDI Interface, drivers, software) or on the sammichSID side (6N138 optocoupler faulty, bad soldering joints, missing JMI shunt, etc.) OK so... you now have two PICs with corrupted MIOS uploads. Until you can establish that MIDI is working and that uploading should work, then there's not much you can do. It could also be that MIOS Studio 2 isn't working well on your PC. You need to test that MIDI loopback through the optocoupler is working. That means taking out the PIC and shorting the Tx and Rx pins of the PIC IC socket, so all received MIDI on MIDI In socket gets echoed back to MIDI Out. Do that for both sammichSIDs with optocoupler installed correctly. But first, do what TK suggested before... FOLLOW THE TROUBLESHOOTING GUIDE FROM THE START however, I'm also interested in the result of TEST MIDI2 as this will show you if there is a problem with your MIDI setup or just the PIC. Also do the tests with MIOS Studio 1 (the Java one) and MIOS Studio 2, in case you see different results. Since you already got one sammichSID working with MIOS Studio 1, one would think that you should continue with MIOS Studio 1 for now, at least for the MIDI troubleshooting.

So does the second kit work perfectly? Or do you have two sammichSIDs which have the same uploading/booting problems?

tl;dr :rolleyes:

EPIC WIN You win a free sammichSID keyring. Email me your address.

Ohhh! Trollbait! Om nom nom... I should correct you there... no one was aggressive, ignorant or childish towards you, a total MIDIbox newbie coming along with grandiose ideas of "commercializing" the non-commercial DIY MIDIbox platform before you ever took a soldering iron to PCB and built one yourself. If anyone was ignorant and childish it was you... first for being ignorant of the wishes of the community before you even posted, and childish for not accepting the polite denial of permission to start making money off something you neither contributed to nor was even involved with for longer than it took you to write your post. Frankly, I don't think anyone cares what you've been doing for the past four years, but I will highlight something interesting for everyone else reading this thread, especially newbies who might think that permission to do big things with MIDIbox is never granted. Around four years ago, I started designing MB-6582, which led to being given permission to sell MB-6582 PCBs, then I sold about 4000 6582A SIDs, a few hundred OPL3 chipsets, a couple thousand rotary encoders, designed an MB-SEQ control surface, sold a few of those PCBs, designed sammichSID, sold 200 of those (and still going). I built MIDIboxes for me first, and then shared the designs, helped others on the forum. I contributed to the community first, and then sold things which would primarily benefit the existing community. Then when I presented the sammichSID idea to TK, one that comes dangerously close to being "commercial", I was given permission, because I had an established track record of doing things for the community. That's just one example of how it's done, and that it can be done. In contrast, all you did was start a thread called "Intent to sell midiboxes" with the idea of... selling MIDIboxes. "I'll build one for me first and then sell them pre-built on eBay" is the essential gist of your post. I find it interesting that the people who actually build MIDIboxes thank me and others for all the help we provide, and think this is a great community. The only ones who think we're aggressive, ignorant, childish, elitist, rude, epic superdouches, etc. are new or non-members who don't like being told that selling pre-built MIDIboxes is against the spirit of the community. Maybe you should actually pick up the soldering iron and build something... then you might actually understand where we come from.

Similar to this one, which includes the plastic knob like analog sticks on PlayStation controllers. http://www.sparkfun.com/commerce/product_info.php?products_id=9032

Cool... I was worried I might have convinced you to buy a new multimeter for no reason :whistle:

Yeah that might be the cause of LCD weirdness. It's hard to see the gap on the 2x5 header though. If the LCD becomes intermittent, you can try putting washers between the LCD and the CS PCB to lower it by 0.5mm or so, this might be easier than resoldering the header.

MY guess is the capacitors. Perhaps the electrolytics which are polarized show different voltages. Do you have any ICs installed? They also will have an effect I think. It's probably nothing to worry about. 5K is not a short. 5 Ohms is a short.

IF I was doing it with Ponoko, I'd get clear 3mm acrylic cut out to "snap fit" into the panel cutout, like my sammichSID design. As long as the corners have a good size radius, it's pretty easy to make it snap in (OK, a little fiddly sometimes but I've done it over 100 times without a problem). The laser burns away 0.2mm of material for 3mm acrylic, so to make the cutout and the window the exact same size, you need to make the window bigger than the cutout by exactly 0.2mm all around (including radius of corners!). Just so you're totally clear, let's say you want a 40mm x 40mm cutout in the panel. You make the shape in the design 39.8mm x 39.8mm, because the laser will burn away an extra 0.1mm on the outer edge of the path in the design, making it BIGGER. To make a window fit exactly 40mm x 40mm, you need to make the shape 40.2mm x 40.2mm in the design, as the laser will burn away an extra 0.1mm on the edge, making it SMALLER. The same thing needs to be done on the radius of the corner. The radius of the window needs to be 0.2mm bigger than the radius of the cutout. You can think of it in 0.1mm differences - both the cutout and the window need to be offset by 0.1mm in opposite directions so they'll match the desired real size after cutting.

I don't think you cooked the regulators. A 12V regulated PSU should measure 12V on your multimeter. Maybe +/- 0.1V. 11.12V is way below what I'd expect. That could mean the multimeter is giving you bad measurements. Are you sure you have it set to DC? :) The thing is, you could probably just install SIDs and everything will work fine... the regulators are outputting roughly correct voltages but considering regulators are designed to output exact voltages, the fact that they don't could mean something is wrong. e.g. A short between 5V/ground might cause so much current draw that the regulator outputs sag under the load. However, I still don't think the PIC would be happy with only 4.61V supply, or the LCD. Just a guess. Retry your voltage measurements with the regulated 12V DC PSU and the JBL shunt removed. However, at this stage there might be no point until you truly validate your multimeter is working by comparing with a known good one.

Based on that data, both these PSUs must be regulated 12V output. Therefore you'll get a constant voltage regardless of current draw... so it's safe to crank the brightness up as much as you like, as that current is taken before the regulators/heatsinks. Can you post a photo showing what you mean by "a fingernails width between the male and female connectors"? It's possible your problems are/were intermittent contacts between the LCD/CS and the base PCB. It only takes one data bus contact on the LCD connection to be faulty and the LCD will not receive the correct initialization and stay in "black bars" mode. I advise resoldering the connectors, but let me have a look first to see how yours is different and if that could make a difference regarding reliable contact etc.

I don't see how it could be working if instead of 5V you have 4.16V supplying the PIC. That's too low for it to even start up I think. Assuming your multimeter is working, 9.76V going into the 7809 is too low. Can you test your multimeter against a known good voltage, like your regulated 12V DC supply? And is this 9.76V reading when using the regulated supply too? Because it should be much higher. Possibly the LCD backlight current is interfering. Remove JBL shunt so there's no current being drawn by the backlight (esp. taken out of the "12V" supply before it goes into the 7809). Retest the voltages and let me know if anything changed.

The PCB allows the use of two single-colour LEDs above the Step View, in case you wanted two LEDs to indicate "Step 1-16" and "Step 17-32". But that is optional, you can use just one LED like all the other LEDs. At the time of prototyping, TK preferred a single LED, and I made my panel the same, so it just got ignored and forgotten and maybe no one has ever made a panel with two holes there. Gunars may be confusing Step View with the Beat LED, which on mine and TKs (and a few others I assume) is a 5mm round red LED, but someone sufficiently motivated could use a bicolor LED (common cathode) and connect it to J3... and then maybe beg TK to make the other colour do something (eg. red on every 4th step, green on other steps). You might ask why I didn't just put a 3-pin footprint at the Beat LED position - refer to my original prototype and how it was split into halves right through this LED, as well as the difficulty of combining a 3-pin and 2-pin LED footprint... i.e. too hard. But I left J3 in there anyway. My PCBs tend to have a lot of easter eggs :D

Beer enjoyed *glug* The current rating of a power supply means how much current it can deliver at that voltage. If the power supply is unregulated, then when the current draw is at the maximum, the voltage also will be about the same. i.e. for a 12V 2A supply, the voltage will be 12V or more when the current draw is < 2A. However, this can lead to the voltage being quite high when it goes into the regulators on the PCB, which means the voltage difference between the input and output of the regulator is higher than it needs to be. This voltage difference needs to go somewhere, and it is turned into heat, which is dissipated out of the heatsinks. I know sammichSID has enough heatsinks and ventilation for up to 12V DC going into the 7809 voltage regulator (i.e. 3V difference, times max current draw of 500mA). Anything higher than that should be avoided - it will generate more heat than can be dissipated by the heatsinks (IMHO). I'm no expert, I just experimented until I was happy with how hot the heatsinks became (or rather, how hot they didn't become). I recommend checking the voltage going into the 7809 (i.e. the point labelled "12V" on the PCB) when you have everything connected. That's a bit hard to do with the sammichSID fully assembled, I'd suggest taking it apart and assembling just the PCBs together and measuring on the bottom of the base PCB. This will give you a true indication of how "safe" your 12V 2A supply is... if the voltage is 12V-14V then that's probably OK, but if it's anything higher then you'll need to think about getting another PSU (sorry) or working out why the 12V 1A wasn't good enough... it really doesn't make sense because the 12V 1A should be OK. Compare the two supplies. Also you should see how this voltage changes as you change the LCD brightness.... when the JBL header is set to "12V", the current is taken BEFORE the 7809 regulator so the voltage will be LESS going into the 7809 than if you disconnected the LCD. Similarly, the LCD brightness trimpot will have an effect on the current, and thus also the voltage. SO.... do voltage tests on the "12V" pin, with BOTH PSUs you have, with everything installed, at three different brightness settings (fully anti-clockwise, middle, fully clockwise)... note also which combinations result in fully working LCD/control surface. To give you some helpful insight... I have a red/black LCD that only works (with my 12V AC 500mA supply) if I set the brightness to middle position, any higher and the PIC doesn't boot and the LCD stays in "black bars" mode. I can actually watch it drop out (i.e. PIC reset, go to black bars) as I turn the brightness. I think you could possibly do the same thing with your setup. However, it doesn't make sense why your 12V 1A supply can't deliver the 500mA maximum current draw of a fully loaded sammichSID with a "high-power" (~250mA) LCD backlight. Doing the voltage tests might help diagnose what is really at fault... you might find it's better to use the 1A supply with the brightness at minimum, than the 2A supply (better in terms of the 7809 input voltage being the lowest possible).

Excellent reporting, I wish everyone would describe their troubleshooting like that! I can't be exactly sure, but it sounds like adding the SIDs causes the current draw to be too much for the supply, the 5V supply sags just enough to make the PIC->LCD comms fail on startup, but otherwise the PIC boots OK and the LEDs flash etc. Try inserting just one SID, maybe you have enough supply for the one, and you can actually get control surface + SID working and prove that everything SHOULD work (i.e. the SID isn't strangely interfering with the rest of the circuit). Try also REMOVING the shunt in JBL. This will turn off the LCD backlight completely, removing more current draw from the system. I also suggest you check your plugpack rating... maybe you are using something insufficient. i.e. is it 12V AC 500mA or 12V DC 500mA etc... quote all three values, the voltage, the type (AC/DC) and the max current.

Excellent work... I like how you have used PCB mounted sockets. How did you do the artwork on the frontpanel?

I agree it's annoying.

It sounds like it was OK, then you took the CS board off and installed chips and when it was put together again, you have LCD and switch problems. Or possibly voltage issues, now that you're powering two SIDs. SO I would say, revert back to the way it was first. Take out the SIDs and 24LC512 ICs (I would suggest numbering the 24LC512 ICs with a pencil so you can put them back the same). Also check you have the shunts set correctly for the high-power LCD and turn the brightness knob to minimum. This will help you assess if it's a current draw problem. Upload firmware again (with CS PCB removed!) to be sure the PIC is still working and that you SHOULD be seeing stuff on the LCD. Then you can reassemble and it will probably work again. Then you can see it stop working by increasing the LCD brightness and/or inserting the SIDs, and then prove it's a current supply issue and not anything to do with shorts/breaks/bad joints.

This might not be relevant, but sometimes my MIDI interface is in loopback mode, where it echoes MIDI In to MIDI Out... if a MIDIbox is connected when this happens, the upload request gets sent back to the MIDIbox and it stays in "black bars" mode. At this point I'd have to point the finger at your PC/MIDI interface being the problem. Try MIOS Studio 2 and see if that makes a difference, it could just be the fairly common "bugs" of MIOS Studio 1 (crappy Java MIDI library).