SEQv4+ - LeMec - LED-Ghosting...

e8abac78cdf65c56

Error, and Questions:

i noticed when removeing the 808 transistors, and cleaning up the solder flux, the ghosting is less… is there any trick (a cap? a resistor?) to avoid that? … the rest please look the video.

17703227d9536da1

Did you also deflux the ICs (74HC595) on the back, also the 818 transistors? Maybe you have some leakage there?

Exactly what parts are used for the transistors etc.? All LEDs from the midiphy kit? All transistors correct?

What is your power supply? Is the voltage stable?

Do the superflux LEDs work well?

What values are in for RJs/SJs?

17703227d9536da1

Also check that all diodes and LEDs have correct polarity.

e8abac78cdf65c56

hei

863-BC808-25LT1G
863-BC817-25LT1G (which one is for the 4th and 8th LEDs?)
Yes 595 defluxed…
SJ: 102 1K OHM (soldered from PIN1 to Middle)
RJ: 47 Ohm
LEDs from the Midiphy Kit.
PSU: Lenovo T440P Laptop, but i also tested a 2A wall charger…

Superflux work well, but if i connect the common (at this point unsoldered pin) they also ghosts (of course)
4148 are poled correct

optically the board all legs and ics are clean, but i cant say whats under the SMD-Parts:
— in order to not desolder all parts (and give them flux again, so all is messed up again), it makes sense to only desolder and clean the components which are Ghosting:

Parts for the 4th and 8th MEC-LED (to get the first board working) - which Transistors (a shematic would be nice)

Which commonly-to-get Cleaning Fluid is good for “under-smd-body-low-impedance-fluid-WASHING - without remove the part”?

17703227d9536da1

What flux are you using anyway? I’ll make sure to avoid it! 🙂
I’ve heard of people using water or a mixture with isopropylalcohol (IPA) and an ultrasonic bath but usually before soldering THT parts where the liquid can get stuck.

Does the DIN side work well? If the DIN is constantly triggered/latched then it might give the same appearance. You should check the button behaviour in MIOS Studio.

Here is a schematic of the source and sink driver:

A0-7 are the anode columns (R1G1R2G2…R4G4 and then R1G1…R4G4 again for the second half of the board) and connect to the RJ A/B. The SJs connect after the RJ resistors (labelled with ½/3), so always to the red superflux LED/RJX A and then to the MEC LED. Sink rows C0-C7 are in blocks of 4 with the encoder switches attached to the first and fifth blocks.

My guess is that you have a short circuit on your BC808 transistors that doesn’t allow them to turn off.

Best of luck,
Andy

e8abac78cdf65c56

@flux: it is in my “wire solder”:

SN60, PB40, - FLUX: CAJ 2% (or OAJ 2% — not so clear, because prin is old…) “www.aimsolder.com made in canada”

@ Washing: i got me a medical grade Isopropanol - and took it 100% poor without H2O - and gave the 2 PCBs several long bathes, and brushes, i shaked it - so the stucked cornes and holes can get rid of it… its very clean now … my cleanest pcb so far… BUT without luck.

@Din Part: no problems with double triggers, or fantom Button presses, very clean in mios-studio.
— what i notice is when i press for example button 1, then button 5 begin to ghost…. that is for all buttons on booth boards.
— on the first board the 4th and 8th button is ghosting brighter then all others - and they do it always.

I checked specially the parts for the 4th and 8th button where the ghosting is extrem — the diodes and resistors are ok, the transistors i cant say at the moment - because want to avoid to desolder them again - only for testing.

the rest of the pcbs are working great, specially the Transport-Board (play record…) no ghosting of course - because of no Matrix there?!

i tested the board with Midibox NG & the SEQ_R, and the SEQ_L script, and with the orginal SeqV4 & Seq_R config file, on booth firmware i have the same problem.

i video about all that - is at the moment uploading to youtube (of course as “not listed” 😉 )

e8abac78cdf65c56

in german… weils leichter geht…

17703227d9536da1

if you can get the isopropanol (IPA) out of the switches, maybe it is okay to give it a good wash… maybe check the manufacturer page or ask at APEM. At least IPA will evaporate better than water. IPA is practically a “non-polar solvent” (or at least less polar than water), whereas the ionic compounds that would cause shorts would be better dissolved by water. If you used water or a mixture with IPA, then a final wash with IPA would be advisable for better evaporation. (After seeing the rest, I am not sure if it is really the issue.)
The lemec_L board I would consider to be okay. As the LEDs have slight variations in forward voltage Vf, also the tolerance of the RJ/SJ comes into play, some slight ghosting might occur there. With the MEC caps on the switches it shouldn’t be too bad.
When you uploaded _NG, the lemec_R seems okay.
JA board seems fine. It is a simpler matrix without transistors

Just check that you are using the latest scripts as there was a bug for a while in the _R.NGC, also that your SEQ HWCFG is for the right-hand, but I guess it is okay. If in the last video you were using the SEQ firmware to test rather than NG last time, then that explains a bit more why one LED stayed “on”.

IMO it seems okay and you probably won’t notice it with caps/switches on. It is a limitation of this sort of LED driving.
If you like, you could play a bit by swapping out LEDs. or easier: slightly increasing the SJ resistance for the columns that ghost a bit more. Ghosting is expected for the column pairs ¼, ⅖ etc.

Another option is to slightly increase the forward voltage of the problematic LEDs but there is no easy way to bodge this, maybe at the SJ but you would have to do it with THT components probably.

e8abac78cdf65c56

like you you wrote: the ghosting is not noticable in the finished device