weird crosstalk & more... (long post)

[el-dr]

Hi Dmitri,

this is Endre Huszar (http://www.endrehuszar.com/index.php?lang=2, you used one of my performances in the video section of the megadrum site). Although I haven't posted here for quite a while, it doesn't mean that I abandoned megadrum at all. Actually it's the opposite: I spent a long time buiding a 56 input new instrument for finger drumming (which is capable a lot more than a zen drum, I'll soon cover it in a new thread when it gets fully completed) and now decided to make a major upgrade to my setup for a series of edrum-clinics and performances (I'll post some new videos in autumn). I want to make a 56i megadrum (Synthex v3.1) the central piece for triggering, instead of having various trigger to MIDI converters on stage (including my "old" 32 input MD).

I have some questions and a technical problem regarding this upgrade:
Q1: You released a firmware modification by my request that enabled MD to switch between the stored setups by receiving Program Change messages. Will this work the same way for the 56 input Synthex v3.1 PCB kit?

Q2: I read that MD is now capable of positional sensing. I guess this means an increased size of the firmware. How many setups can be stored on a 56 input MD with the latest FW? (Actually, I had no time to try this feature on the 56i kits at all, so it would be important to know this figure for any FW version)

Q3: Does MCT work for the 56 input kits?

And finally the technical problem I have: it could be the same as
http://www.megadrum.info/forums/viewtopic.php?f=3&t=1256&p=15136&hilit=crosstalk#p15136, but I decided to post it here, because I'm not sure if my problem is really the same as his.

The 3.1 Synthex PCB was put in the same 1U rack that I built for the 32 input version: I left the first 24 mono jack inputs on the back panel untouched and the last 8 ones plus the 24 extra inputs were soldered to four DB9 connectors. I made two 5m serial cables, too (same wiring as a normal serial cable with pin5=ground).

When I tried it with a piezo on one input, I got a very nasty crosstalk, the same as if I used the wrong multiplexers (which is not the case, they're HC4851s and work fine on the mono jack inputs).

When I leave out the 5m cable and connect the piezo directly to any one of the DB9 sockets, it works fine.

When I do use the 5m cable, but disconnect all unused wires, so only pins 1 and 5 (gnd) remain connected, it works fine, too.

All 24 jack inputs work fine even with long cables.

Both serial cables are tested and have no soldering errors. All inputs (including the 24 mono jack ones) have a 5k resistor. (This became necessary when I started to feed the trigger inputs with signals from the channel direct outs on my mixer for my acoustic kit.)

I have no idea what can be wrong, thanks for any input to figure it out.

[dmitri]

I have some questions and a technical problem regarding this upgrade:
Q1: You released a firmware modification by my request that enabled MD to switch between the stored setups by receiving Program Change messages. Will this work the same way for the 56 input Synthex v3.1 PCB kit?

Yes, of course.

Q2: I read that MD is now capable of positional sensing. I guess this means an increased size of the firmware. How many setups can be stored on a 56 input MD with the latest FW? (Actually, I had no time to try this feature on the 56i kits at all, so it would be important to know this figure for any FW version)

For 56 inputs it is 2 configs and 6 drum maps.

Q3: Does MCT work for the 56 input kits?

It does.

And finally the technical problem I have: it could be the same as
http://www.megadrum.info/forums/viewtopic.php?f=3&t=1256&p=15136&hilit=crosstalk#p15136, but I decided to post it here, because I'm not sure if my problem is really the same as his.

The 3.1 Synthex PCB was put in the same 1U rack that I built for the 32 input version: I left the first 24 mono jack inputs on the back panel untouched and the last 8 ones plus the 24 extra inputs were soldered to four DB9 connectors. I made two 5m serial cables, too (same wiring as a normal serial cable with pin5=ground).

When I tried it with a piezo on one input, I got a very nasty crosstalk, the same as if I used the wrong multiplexers (which is not the case, they're HC4851s and work fine on the mono jack inputs).

When I leave out the 5m cable and connect the piezo directly to any one of the DB9 sockets, it works fine.

When I do use the 5m cable, but disconnect all unused wires, so only pins 1 and 5 (gnd) remain connected, it works fine, too.

All 24 jack inputs work fine even with long cables.

Both serial cables are tested and have no soldering errors. All inputs (including the 24 mono jack ones) have a 5k resistor. (This became necessary when I started to feed the trigger inputs with signals from the channel direct outs on my mixer for my acoustic kit.)

I have no idea what can be wrong, thanks for any input to figure it out.

I'm afraid that with a long serial cable where all wires are running along each other (and maybe twisted) you are bound to get an electrical interference/crosstalk.
I can suggest to try to make a serial cable where each cable is shielded individually.

[el-dr]

I'm afraid that with a long serial cable where all wires are running along each other (and maybe twisted) you are bound to get an electrical interference/crosstalk.

It seems you're right. In the meantime I made a shorter cable (which has no practical use for me, because I really need the 5m cables to get from the right side of the kit to the racks) and it performs much better, but even this one is not perfect.

I'm afraid that with a long serial cable where all wires are running along each other (and maybe twisted) you are bound to get an electrical interference/crosstalk.
I can suggest to try to make a serial cable where each cable is shielded individually.

Does that mean that pin5 (GND) should be handled the same way (so that all 9 pins should be soldered to one of the individually shielded cores)?

[dmitri]

I'm afraid that with a long serial cable where all wires are running along each other (and maybe twisted) you are bound to get an electrical interference/crosstalk.
I can suggest to try to make a serial cable where each cable is shielded individually.

Does that mean that pin5 (GND) should be handled the same way (so that all 9 pins should be soldered to one of the individually shielded cores)?

Sorry, read it a few time and still didn't get what exactly you meant. Each signal wire should be shielded and all shields should be soldered to the ground pin.
Also, since you join grounds at the remote end the ground connection should have as low resistance as possible otherwise some of the signal current from one piezo will flow into all other signal wires with piezos.

[el-dr]

Sorry, read it a few time and still didn't get what exactly you meant.

You still managed to answer, thanks:)

[el-dr]

I built 3 multicore cables for the 24 pads where every pair of the cores was individually screened. The + and - of the sensors got soldered to the conductors and the screen was only soldered to the ground on the MD end. They work fine, but there's one thing that I noticed. The crosstalk still appears for the first few hits on power up (maybe for the first 30 secs) and then it goes. It looks as if something had to be discharged on power up to reach the normal operational environment. This is no problem if I can be sure it won't come back during a performance. Are there any known reasons for this and should I be worried about it coming back unexpectedly while playing?

[dmitri]

I'm guessing that this crosstalk is to not connected/not configured inputs. In this case by default those inputs have HighLevel set to 64. Even a small electrical crosstalk into those inputs will result in these small signals being detected as strong hits thus exceeding the crosstalk suppression level. And if HiLvlAuto on those inputs is set to Yes (which is default), after a few hits HighLevel will get updated to higher values. This in turn will cause crosstalk signals being detected as weaker and weaker. And in turn again these crosstalk signals will stop exceeding the crosstalk suppression level after a while.

[el-dr]

I'm guessing that this crosstalk is to not connected/not configured inputs.

Unfortunately not. I set all HiLevels to 1023 (or 1022) and HiLvlAuto to Off. Anyway, it only happens on power up and never appeared during prolonged use, so I can live with it.

[dmitri]

Btw, you can completely disable unused inputs by setting Note number to 0.