Duggle

Let me know if this is not clear.

First of all, quickly, do you have a 1k pullup to 5V on pin3 of the LM324? Does this pin vary from 0.09V to 0.60V as you press the sensor? edit: Also where is Rg2 (10k)?

Yes. Remember the power rails for the LM324, that's 0V and 5V. A plug type breadboard is good to test the circuit before soldering the final hardware. This is just an easy way of getting 50k, because 100k was already in the circuit, just use 2 in parallel for the 50k!

Hi, I wasn't that clear. The circuit you should make is like the one labelled "App Note cct" on the schematic rather than the exact circuit of the other two examples. So the values of R1, Rg1,Rg2, and Rf need to be defined. I neglected to mention that Rf and Rg2 get entered to the spreadsheet (by the designer) and Rg1 and R1 get calculated by the spreadsheet. It makes sense therefore to put a preferred, available value into the cells for Rf and Rg2. We can use a trimpot (using the wiper plus one end) for R1, and closest available value Rg1. I tried by putting 1M for Rf, and 10k for Rg2. This yielded 374k for R1 and 173k for Rg1. In practice, I would make R1 a 1M or 500k trimpot, and Rg1 180k. You can try different values in the spreadsheet for Rf and Rg2 but I think it is best if all values end up less than or equal to 1M.

Thanks for posting the schematic. The positive input of the op amp is at 0V and the negative has a pull-down of 15k to -15V. This means there is only about 1mA flowing through the aftertouch sensor. My suggestion is to use the 1k pull-up to 5V. Based on your measurements from before this will produce a signal of between about 0.05 to 0.7V across the sensor. This the circuit you need plus a spreadsheet to calculate the resistors, plus a couple of worked examples. The purpose of the circuit is to shift and amplify the signal to get the best possible voltage range for the ADC. A single op-amp of an LM324 plus a few resistors and a trimpot is all that is needed. Change the blue values (ViFS,ViZS) which represent the minimum and maximum input voltages to the voltages you measure with the sensor connected between 0V and a 1k resistor tied to 5V. Let me know if anything about this is not clear, I can help.

Rightly or wrongly the MIDIbox platform has always had a very strong DIY ethos. This means that although kits are available, they still need to be assembled by someone. MIOS compatible hardware is pretty much limited to the published designs. Although it's very new, the STM32F4 Discovery board is probably the closest thing yet to pre-built MIOS32 board, although it lacks interfaces one would probably want such as UART MIDI ports etc. If you can bring yourself to solder an LPC17 core board you'll have a platform that is ready to go with a whole range of project options (most will require some additional I/O, again easily assembled from kits).

Can you take a photo (or scan) and post the schematic? Yes, there is a simple circuit and some calculations, but I'd like to see the circuit that you have first, thanks.

Just connect the pullup to 5V where it is conveniently found. Beware though, that the device may get damaged by the (large?) current with this method. I have no data on it. Maybe better to use and op-amp and less current through the device.

A smaller resistor is necessary to increase the range over which it outputs... I am going to suggest that you replace the 1k pullup with an 82R (Ohm) resistor which is connected to 5V rather than 3.3V This will yield an output that ranges from close to 0V to a maximum of slightly under but close to 3.3V at the input of the ADC, A0. The current will be 22mA to 60mA which may be a little on the high side (300mW in the pullup resistor will get hot if you hold the strip down for long) Also I don't know the current/power rating of the device. This problem (if it is a problem) can be overcome by adding an op amp to the circuit with a higher pullup. The calculations are based on 150R with the device not pressed. I can post the calculations if you want.

You'll want the voltage range at A0 to cover approx. 0V to 3.3V to get the best performance (which will be a lot better than what you have now). Before I suggest specifics: is there a current or power or voltage limitation on this device (i.e do you have any data or absolute maximum parameters etc)?

I recommend making sure the analog signal at A0 is right before making any bold decisions. Can you make a voltage measurement with a multimeter on A0 (and 0V)? What are the voltage readings when: not pressed pressed at the top pressed in the middle pressed at the other end This will provide information to determine the correct value of bias resistor (currently 1k) to give the best voltage range for the ADC. I use the Core ADC for my pitchbender and Mod wheel and it works fine.

If you mention broadly what you want to actually do, we may be able to point you in the right direction, with some specifics.

Which "fader surface" are you considering as a starting point for your project?

I've only looked at this thread briefly. Do you have pullup resistors loaded on the DIN PCB? What happens if you ground some of the inputs that are being fired? Do the events cease from these inputs?

That's great to hear. But what was wrong? How did you fix it?

Yes, I think you need to be able to have rows=1, so that there is 100% duty cycle and remove need for high side driver.

I edited my last post. Manually energising row/column "manually" is just a way of investigating what is happening with the drivers, resistor, LED to see where the current is limited. I think your final design will run with rows=4.

Can you check the duty cycle of the matrix with an oscilloscope? With rows=4 it should be 25%. If you don't have a CRO, then don't worry about it. Set up special config that allows you to manually turn on/off individual row and column driver lines. Can you determine the voltage drops through the LEDs, resistors, drivers under static conditions (i.e the drivers are in steady state rather than scanning the matrix). This allows you to determine the LED current when they are 100% on. Also allows to identify what is creating the dimming i.e where the voltage drops are worst. Note that when the matrix is scanning (rows=4) they will be only 25% as bright.

The emitter follower (BC547) will have a voltage drop of >=0.7V and the output of ULN2803 will drop another say 1.0V. This leaves only (5-0.7-1)=3.3V to drive the LEDs. The ULN2803 low side drivers only ever drive 1 LED at a time. If you remove the ULN the pin from the HC595 can still drive the load but with less voltage drop. Have you tried reducing the 100, 150R resistors? Your drawing is labelled with "8x8" but declared with rows=4. If you add another 6 SR's you could drive up to 24RGB LEDs in an unmutiplexed way (rows=1). You'd need more than 1 RGB matrix defined. Is this supported? If so, it will be 4x brighter due to the increased duty cycle.

What is the matrix wiring? If 16 displays means 16RGBs, then have you tried 1x16RGB matrix configuration? I think this should give you a maximum LED output the same as with constant supply. Did you say ULN2803 was more (or less?) bright than the HC541? Did you ground the ULN2803 properly? The ULN2803 has an output low of around 1V which implies that a lower series resistance be used to get higher output.

What is your problem? Is it that the RGB backlights for your LCD modules are not bright enough?

The devices feature thermal shutdown.

I would put a heatsink on the 7805 so that it never goes above 40degC. These devices have a thermal shutdown feature which is what I suspect is happening to you. BTW, I'm not receiving email notifications from active threads. Is it just happening to me?

Does the 7805 run hot? (nearly too hot to keep your finger pressed to it?) Just to confirm: If you disconnect half of the displays, then you don't see this problem?

You'd need to eliminate the possibility of a glitch on the power supply lines. I'm thinking it's something that is common to all the displays. What sort of power supply? Is it a linear regulator (getting hot and tripping, perchance?) Does it happen when all displays are running, but then not when most of them are disconnected? If you have the luxury of a digital CRO, set it up to trigger on the falling edge of the positive rail to capture what is happening.