latigid on

Apparently you have to set the LXR BPM to 0 to slave to an external clock, it sounds like the LXR is getting confused with two clock sources. http://wiki.sonic-potions.com/index.php/LXR_Owners_Manual#MIDI_clock_sync

Just to clarify; do you think the three rectangular boards (4*OLED; 2*OLED, 1 enc; 6*buttons 2*OLED) are too far away from the encoders? I have the 4*4 illuminated encoder boards already, but my idea was to create another PCB which holds 4*OLEDs mounted at 45 degrees as per the Programma design, which is stacked over the encoder board. So I think that's what you suggest: displays mounted as close as possible to encoders? Sorry if that's confusing, the rectangular boards are useful for different kinds of NG builds or as the menu interface for Programma. I completely agree about the modularity though. You should be able to start with as many (up to 4) or few encoder boards and displays as you need and scale from there. If the arrangement doesn't suit you could also add distance between them or lay them out e.g in a straight line.

You mean for the Programma? That's better explained in their thread but for my variation it's something like this: Potential layout of 8*8 grid with extra displays, top is for displaying 1 or more custom envelope curves, bottom has an "SCS" built with 4*OLED and, 6 buttons, 1 encoder:

I'm working a bit with @Hawkeye and @jojjelito on their awesome MB-Programma project, only we'll try to use my illuminated encoders in place of the sadly obsolete LRE boards. The Programma makes good use of displays mounted at 45 degrees to label each encoder. But to do this it requires quite a mishmash of wiring. I propose a new PCB set to make things easier, not just for Programma but a range of NG builds. I'm still working on the 45 degree board, but I have a fair idea of the others should work. Here's how 4 OLEDDs look. This 7-pin type is very common and quite cheap. Spacing is at 30mm on a 120mm PCB. I thought it would be interesting to combine a sort of SCS, a bit different in layout to @ilmenator's. 10mm switch footprints will also be added. You could split the SCS over two PCBs (using a single ribbon with two connectors on it) or a single PCB if you liked. Once you go over 8 OLEDs, you need to start using DOUTs to generate the #CS (chip select/slave select) lines. It was mentioned that the Programma had a few issues with signal integrity, so I thought it might be an idea to combine a line driver with a shift register chain, and also an optional power regulator circuit. Adding to this, the new F4 Core J10B doesn't use 1:1 pinning with J1 on a DOUT board, so would need a special cable. I found some discussion that a similar circuit using 541 drivers and 1k output resistors could happily drive an array of displays spaced 10 metres in each direction. The resistor also acts as "termination" which attenuates reflections in the data lines. I will also add RC termination on the carrier boards. I didn't buffer the 595 outputs as they have reasonable source current already and #CS is less critical for timing. Apparently the displays run at 22mA each, so a linear Vreg should be okay to drive 32. Standard ribbon wire can take 500mA per per strand, again we should be okay, but I'll put an alternative power option on the carrier board. Two driver boards could be chained for a possible 64 OLEDs (!). If possible I will try to connect the remaining #CS lines so one could drive 8 additional displays, though I think 64 is the limit. @Hawkeye suggests that it's better to use #CS from one source only to avoid problems i.e. directly from J15A or DOUTs. Any questions or comments, please share!

Ah sorry, the text is quite small!

I think for this circuit you'd get quite low range, LM324 is not rail to rail and the maximum output at 5V supply is 3.5V. So less for 3v3 -- 1.8V? MCP600x is more suitable and very cheap. Sorry, I don't have real experience in trigger circuits, but these look quite simple to try out.

Good luck! I know a lot of professional drum triggers use this scheme. E.g. ddrum; they also have a "brain" which must be a simple ADC to MIDI. Bare piezo elements cost almost nothing, although you may need to experiment with the damping material e.g. a piece of foam. Zeners have a fixed breakdown voltage (wide range available), whereas Schottkys have a lower forward voltage than a standard PN junction (4148). The normal idea with MCU pin protection is to use Schottkys as a shunt. The lower Vf is, the more range you have, as the clipping point is the reference voltage (upper limit) - Vf. To protect from overvoltage you connect the anode to the pin and the cathode to 3v3. This ensures nothing greater that 3v3 can pass, it's also normal to include one or two current limiting resistors. To protect from undervoltage you connect the anode to ground and the cathode to the pin. I would suggest using the op amp itself as a limiter, because it's impossible for the output to swing past the power rails. There are protection diodes at work here too, so you need a rail to rail op amp in order to maximise the signal transmission.

Like it! Nice job!

Have you though about a piezo trigger? You may get a lot of feedback with a microphone. For this too you'd need a processing circuit as I think piezo elements can generate a high voltage (low current). This one is very simple (open source): http://musicthing.co.uk/modular/?p=933 It might be quite difficult to separate the rim and body sounds. The Zener (better Schottky) diode would be unnecessary if you used a rail-to-rail op amp powered at 3v3 and your scaling circuit was tailored to give a correct signal voltage.

Do you mean you want to assign a track to an AOUT port? This would be done by setting up the track on the AOUT channel of your choice. You can also route a MIDI input to an AOUT port just like a MIDI-CV converter. Or do you mean something else? It's not so clear from your question.

DIAMEX PIC PROG works fine and is the recommendation of TK. Or, send the PICs to one of many on the forum and we can program them for you :). I had two PICs from SmashTV (since several years) two from Mike (since a few years) and two virgin ones from Mouser. Of the pre-burned ones, one was faulty, even after erasing and re-flashing. So I'm sure the PICs from Mike had the bootloader burned in, but maybe a handling error on my side caused an issue. In any case, I have the Quad IIC working well now!

Okay, I understand what you mean. LED brightness does depend on the current limiting resistor, but it's a non-linear relationship. Brightness is much more dependent on the pulse width (or pulse density). I think the duty cycle in the BLM is 10%, not so easy to calculate brightness. The red LED does have a lower Vf, but also a lower mcd rating than green. In my experience, 220R was a good balance for both red and green LEDs. You might also get some feedback from others here who have mixed red and green LEDs in their builds. If you want, here is a shared OSHpark project for very cheap test boards, you could also check the fitting of your new LEDs. https://www.oshpark.com/shared_projects/I9xfP7XE

If the wiring is very old, then earth might be connected to the neutral line. It's not very safe in any case. Sounds like you need an electrician to look at it rather than internet strangers.

Looks great! 3mm thick? I'd imagine it's stiffer than acrylic?

Check the SCALAR circuit, the networks are bussed for the switch anodes and isolated for current sinks and limiting. There's no real reason to use networks apart from being easier to solder. Blue and green have different mcd values. The 56R/220R choice was to balance the brightness, or did I not understand you properly? The single cable is a way to keep things tidy, but not ideal, I agree with you there. It's also the simplest way to power the BLM from the Quad IIC board, where the optocoupler and pull ups on the MIDI circuit share the same rail. If they were separate, a higher voltage could be sent through the single cable and regulated locally. When this was first designed 5 years ago, TK. used common low Vf, low brightness LEDs, so high current consumption was not an issue. 2A SPSU on my build, I recommend to power the Core or at least the Quad IIC module with a decent supply. Some cases have extra holes for this purpose, but I don't know of any working builds using them as of yet. You could try larger resistors at the expense of overall brightness. See above. For the next run I will modify the miniCore board to take an adafruit "Verter" regulator. This is the simplest option and it works in at least two builds so far (@jbdiver). You are also welcome to connect an external PSU here at the expense of one extra cable to plug in. But with a right-angled DC jack it hopefully isn't too messy. I think the voltage drop is partially due to the regulated source being distant from the load, but also inherent to the circuit. Especially with bright LEDs, we're switching a lot of current quite fast, and it could be a speed limitation of using shift registers in this way. I'd hazard a guess that MOSFETs would be better current sinks, but I'm not an expert at such things. My low-level understanding is that you need special types to use a +5V gate voltage, otherwise the transistor isn't fully turned on. In common use, it would be rare to have even 1/4 of the LEDs active at a time, so while not a perfect design it should be okay for most people.

Because the PSU chassis is plastic, I would say that the earth pin is not necessary. But this is mains voltage and you have to be careful... There's no easy way to remove the pin without hacking apart an IEC connector. Can you try with a linear PSU instead? EDIT: I don't advocate removing the earth connection if the device expects one. As we learned later, the OPs house uses non-earthed main sockets which clearly contribute to the problem.

Looking at the diagram here it's probably the expected behaviour, as +15V is always present. If the polarity was reversed the electrolytics would go pop and nothing would get through the regulators. So the next question will be about the 15VDC PSU, what is the model/do you have a picture? If it has a metal frame and earth pin, do you get continuity between it and the 0V line in the MB-6582? FWIW I did a similar mod on my build, but I use a linear (transformer-based) 12VDC PSU as input. This means that the device is electrically isolated via the transformer.

For me you still haven't explained the problem well. Because this is a modified build, you need to clearly document all changes here. Like drawing all bridges/cuts/substitutions/omissions on the layout file -- it doesn't have to be a work of art, you could use Windows Snipping Tool or a screenshot with a basic drawing program. Please don't expect others to go through all of the disparate threads and collate the info for you. I'm happy to try troubleshooting if you give some more info. And I repeat my "stupid" question above; does the MB-6582 power up only if the DC adaptor is plugged in and switched on? What is the voltage? What kind of PSU is it? How is the other equipment powered? It's important to check your wiring, because there are some complicated powering schemes possible in the MB-6582. Some involve adding two rails together which only works in certain circumstances, and also means "ground" is not always 0V! I assume you're only want to use 9VDC for the SIDs, which means you should come in with ~12VDC and regulate down to 9VDC.

What PSU are you using? The C64 one? Does the board pass the voltage tests like this? Do you mean it powers up with no PSU connected? That's pretty strange if so; perhaps you have a wiring problem in your house? You could have mains voltage where it shouldn't be. Check again through Wilba's layout that you've made the correct mods and upload an annotated image here to confirm.

I'm still undecided on this, an ideal solution would be both! A Voice/channel mode that detailed settings are made for a single channel and a Performance/Multi mode where a set of parameters can be adjusted across multipul channels. An example of Performance mode would be the 4x4 encoders assigned to ENV2, you could control the ADSR of 4 channels without any paging. Right, so now we have three concepts of paging: 8/4/2 Voice mode (TK.'s choice) which is like a traditional MIDI-CV e.g. 4 voice with pitch CV, filter CV, gate and trigger; encoders are arranged in blocks within a voice and menus are used for finer tweaking. Single parameter mode (my choice) where all modulator depths across a single channel are accessible on one page, and specific parameters of modulators are on others. Each modulator is thus tweaked on a channel basis; see the mockup picture: e.g. Channel 1 is selected from the 4*4 button grid the encoders are colour-coded to match modulators represented on the bottom half of the button grid and adjust depth only on Channel 1. Env. 1 is pressed, the encoder field changes to yellow and all stages/other parameters can be tweaked. Unused encoders are unlit. For transpose and tuning, there's the "keyboard" layout of buttons and a dedicated encoder Multi parameter mode (your idea, kind of a hybrid between those above) assigns encoders to specific controls across multiple channels, e.g. ADSR of Channels 1-4. You'd have to make some judicial decisions when more than four useful parameters were needed. We'll have to test the hardware to get a good idea of the most useful workflows, but hopefully it's flexible enough to accommodate everyone's needs.

Agreed about the dimensions, and following your suggestion these PCBs are 100mm tall, 420mm total width which will definitely fit in an 84 HP case. The only problem being that there's no room for the connectors -- as such a 126HP case (25.2"/640mm usable width) could be a better choice if you wanted them front mounted. The LRE board and displays etc. will be difficult to fit into 3U. The Wilba SEQ PCB is 121.92mm tall, which will is fine for 3U but doesn't take Eurorack mounting rails into account. My plan is to buy or make some sort of standard 19" road case and have the SEQ, CV and CV breakout panels mounted and powered off a decent triple PSU. Can you comment on your preferred layout of parameters? I.e. most functions accessible from encoders or more "voice" oriented pages?

6mm switches (4.5/6.5mm lead spacing) are now implemented with a 1.5 1.6mm vertical offset relative to the D6s. Any others? Prog. and Tempo labels reinstated.