www.megadrum.info

Hi Dmitri. All kit assembled, general settings done.
70 % satisfaction.

Drums set up first:

DIY snare: 13" roland style (head piezo/rim piezo) plus head edge piezo (i used that with aquawicket edrum monitor to get positional sensing to drive Addictive Drums, details above...) I disconnected the 3rd piezo using the snare with MD, of course.

DIY toms: edge head trigger (like 2box). Very good results, but some inconsistent mid/high level strokes triggering (very rare), no rim piezo (not yet )

Cymbal: millenium 1 zone, with DIY choke switch a huge better response with MD then with old PM16

Ride: Quadrapad (my DIY project...) very good feel ( ) same accuracy MD/PM16+edrum mon

HAT: Yamaha p65s modified piezo/piezo (for acquawicket edrum mon, but now i'm reconverting to yamaha def. piezo /sw/sw). nowadays with pz/pz i have good pleyability shank/tip rolls etc. (same quality with PM16).

Hat Pedal: FD8

KICK: DIY wood/metal pad with real head glued on spongy rubber

EXTRA: SPD-S and his trigger inputs....

THE TROUBLE:

SNARE PREVIOUS SETUP CONNECTIONS:
Snare cnter head (cone)--------SPDStrigger 1 input----MIDI---FASTRACK PRO MIDI----- EDRUM MON 3zone channel 1
Snare edge head---------------SPDS trigger 2 input----MIDI---FASTRACK PRO MIDI----- EDRUM MON 3zone channel 2
Snare rim----------------------PM16 trig input---------MIDI---FASTRACK PRO MIDI----- EDRUM MON 3zone channel 3

I hope it's clear by now...

Spds trigg input setup is 2 single input (no interaction between) Type: PD120, sens: 9 thresh: 0 retr: 12 minscan 1.2
curve: exp1 for both inputs

Now:

The snare connected to MD gives me less satisfaction then my previous setup. Now i have real rimshot and nice cross sticks(previous i did hit only the rim to get rim shot, so no sticks) and works well. But now i have a consistent hot spot (tipical in also in Roland pads) and less sensibility to edge rolls. Also velocity level"consistency" is not so good.
My snare have a high level of 350, not very hot, positional works quite well, but i don't reach the same results than before.
I tried different MD parameter tweaking, but what i achieve is or good pressrolls but double strong strokes, or viceversa.
What trick they use, terrible Nippos?

May i try to connect my edge piezo in parallel with the centre piezo? But i'll miss positional sensing, don't I?
Your thinking?

P.S.: Sorry for the long post, i thought it was required to understand the trouble.

daniele99 wrote:The snare connected to MD gives me less satisfaction then my previous setup. Now i have real rimshot and nice cross sticks(previous i did hit only the rim to get rim shot, so no sticks) and works well. But now i have a consistent hot spot (tipical in also in Roland pads) and less sensibility to edge rolls. Also velocity level"consistency" is not so good.
My snare have a high level of 350, not very hot, positional works quite well, but i don't reach the same results than before.
I tried different MD parameter tweaking, but what i achieve is or good pressrolls but double strong strokes, or viceversa.

You need to show the actual settings you tried and what results you got with each.
The level inconsistency maybe either due to suboptimal settings or due to very large levels difference going from the "hot spot" to the edge - when hitting with the same force different places on the pad the signals coming from the piezo are noticeably different and it causes the perceived level inconsistency.

Also a few things to consider:

1. Curves may play big role in reducing inconsistency and it is important to know here that same curves names on different module makes are actually different so you may want to try using different curves on MegaDrum to find one which suites better your taste and your pad. Custom MegaDrum curves may be of help here as well.
2. If you get 350 in HighLevel, to have more dynamic range to apply a curve to, it is better to raise sensitivity by increasing Gain to get higher HighLevel.
3. Mesh type pads produce longer wave lengths and so that not to miss a peak of a signal it may be necessary to raise MinScan.

Now i'm at work, i don't remember the ultimate exact values. Tonight i'm going to write them down.
But i remember these:

AltrigSupp: no (yes only give me 20% of notes played, every dyn lev/time used)
All gains low: yes
High lev: auto gives 350 (tried 450/280)
Thresh: 7/8
Retrigg mask: 12
min scan 20/35
dyn lev: now 7 (tried from 0 to 15)
dyn time: now 28 (tried 0/60 )
curve: log2 or even log3 (to fight against "consistency") or even custom 1 (log2 with more slope from 0 to 170 and compressed above it)

in this mode i have resonably linear sigle mid/strong strokes, quite fluent center rolls but I miss edge rolls.
L-R-L-R octave notes with same energy and strictly on center give velocity like: 72-58-45-95... not very good.
Suggestions?

Time to go home. Tomorrow i'll post more accurate values.
I'm thinkin' about using two dual input for snare in MD involving the 3d piezo and bypassing positional sensing. But i'm wondering how to do it without missing rimshots and without having xtalk issues.
Thank you by now Dmitri!
see you

Here we are. Tonight better results, more fluent center rolls, less false triggering, but still not like with Roland converter.
Parameters:

curve: custom 1 (like previous mail)
comp lvl: 0
lvl shift: 0
xtalk: 2
xtalk group: 0
thresh: 10
gain 7 (your advice )
hilvl: 263 (auto detect 300)
retrigger: 16
dyn lvl: 10
dyn time: 24
minscan: 27

Now i' curious. I want to go in deep and to know how the algorithm picks up velocity value.
Correct me if i'm wrong:
MD samples the wave (only half positive or negative) with 11 bit resolution. Minscan sets the min time to wait before reading the peak value from thresh to 1024, due to the rising time of the pad (mesh slower, gummy faster), ok?
I'm imagining that MD doesn't read a sort of RMS value (not more like a VU meter but like a peak meter), am i rigfht?
Then MCU choses the higher level reading of the first half of the wave and interpolate a 9bit value according to the curve settings? Or it choses the last peak value at the end of minscan?
If i'm not clear, i can draw...

daniele99 wrote:gain 7 (your advice )
hilvl: 263 (auto detect 300)

I guess it from the detected level - you still have AllGains Low set to Yes. In this case Individual Gain levels are ignored. To have a wider range to apply curves to, you would want to get HighLevel above 500.

retrigger: 16
dyn lvl: 10
dyn time: 24

Rettriger 16 looks bad. It is better to get the best balance between fast rolls and false triggering suppression using DynLevel and DynTime with low Retrigger value. Especially so when using AltFalsTrSupp.

minscan: 27

I don't know what signal your mesh snare produces, but the mesh pads I used (Pintech 10", Roland-125X, Drumtec 14") all produced much better result with longer MinScan between 30 and 40.

Now i' curious. I want to go in deep and to know how the algorithm picks up velocity value.
Correct me if i'm wrong:
MD samples the wave (only half positive or negative) with 11 bit resolution. Minscan sets the min time to wait before reading the peak value from thresh to 1024, due to the rising time of the pad (mesh slower, gummy faster), ok?
I'm imagining that MD doesn't read a sort of RMS value (not more like a VU meter but like a peak meter), am i rigfht?
Then MCU choses the higher level reading of the first half of the wave and interpolate a 9bit value according to the curve settings? Or it choses the last peak value at the end of minscan?
If i'm not clear, i can draw...

MegaDrum measure peak values. It starts measuring at around the Threshold value for at least MinScan period and it will than use the peak value within this period as the signal level. If the peak level is reached later than MinScan period (and for mesh pads the peak will move from one side of the half wave to another depending on where it is hit) MegaDrum may miss the peak value.

dmitri wrote:I guess it from the detected level - you still have AllGains Low set to Yes. In this case Individual Gain levels are ignored. To have a wider range to apply curves to, you would want to get HighLevel above 500.

Yes, as i wrote in a previous post "All gains low" set to yes. Uuuh! Now i understand why changing indiv gains changes nothing, especially high level detection with auto lvl engaged!!
So if i have some hot pads and some others cool, i'll never be able to fine tune all the kit.
Time to go to hundreds of voltage dividers.

dmitri wrote:Rettriger 16 looks bad. It is better to get the best balance between fast rolls and false triggering suppression using DynLevel and DynTime with low Retrigger value. Especially so when using AltFalsTrSupp.

As i wrote in that post, altFalsTrSupp give me worst results, so is set to "no". I tried with less Retrigger, but i achieve only a lot of false triggers, with every Dyn parameters.
The thing that push me down is that with roland inputs snare triggering was still perfect with these parameters:

daniele99 wrote:Spds trigg input setup is 2 single input (no interaction between) Type: PD120, sens: 9 thresh: 0 retr: 12 minscan 1.2
curve: exp1 for both inputs

So my pad is not so bad!!
I'm also trying to make an hybrid MD-SPDS-Edrum monitor, but my MD doesn't digest midithru thru USB. It frequently go crazy shooting random midi messages...or may be Windows MIDI doesn't handle well virtual midi sw (Yoke or Loopbe), so i need to restart all the gears!

daniele99 wrote:

dmitri wrote:I guess it from the detected level - you still have AllGains Low set to Yes. In this case Individual Gain levels are ignored. To have a wider range to apply curves to, you would want to get HighLevel above 500.

Yes, as i wrote in a previous post "All gains low" set to yes. Uuuh! Now i understand why changing indiv gains changes nothing, especially high level detection with auto lvl engaged!!
So if i have some hot pads and some others cool, i'll never be able to fine tune all the kit.
Time to go to hundreds of voltage dividers.

Are you guessing about hot/cool pads here or you have already tested all your pads with different Gain levels? AllGains Low is only a last resort option.

dmitri wrote:Rettriger 16 looks bad. It is better to get the best balance between fast rolls and false triggering suppression using DynLevel and DynTime with low Retrigger value. Especially so when using AltFalsTrSupp.

As i wrote in that post, altFalsTrSupp give me worst results, so is set to "no". I tried with less Retrigger, but i achieve only a lot of false triggers, with every Dyn parameters.

I just tested AltFalseTrSupp and it looks like I introduced a bug in one of the latest firmware versions - it indeed misbehaves. Thanks for reporting it and I will look into it.

The thing that push me down is that with roland inputs snare triggering was still perfect with these parameters:

daniele99 wrote:Spds trigg input setup is 2 single input (no interaction between) Type: PD120, sens: 9 thresh: 0 retr: 12 minscan 1.2
curve: exp1 for both inputs

So my pad is not so bad!!

I wasn't saying it's bad. As I said before, settings names between different modules makes may appear the same but the same values on those settings are not interpreted the same internally - minscan 1.2 in Roland may actually correspond to MinScan 30 in MegaDrum if at all they do the same thing.

I'm also trying to make an hybrid MD-SPDS-Edrum monitor, but my MD doesn't digest midithru thru USB. It frequently go crazy shooting random midi messages...or may be Windows MIDI doesn't handle well virtual midi sw (Yoke or Loopbe), so i need to restart all the gears!

Make sure you have not introduced a MIDI loop in your setup

dmitri wrote:Are you guessing about hot/cool pads here or you have already tested all your pads with different Gain levels? AllGains Low is only a last resort option.

Yes, the first testing day. Now i'm afraid that if i uncheck this option, i will be forced to re calibrate hat pedal. I noticed they are correlated.

dmitri wrote:I just tested AltFalseTrSupp and it looks like I introduced a bug in one of the latest firmware versions - it indeed misbehaves. Thanks for reporting it and I will look into it.

Pleased to be usefull!
Let me know when are you going to repair FW, i think that with that algo working, i would stop the troubles. I hope.

dmitri wrote:I wasn't saying it's bad. As I said before, settings names between different modules makes may appear the same but the same values on those settings are not interpreted the same internally - minscan 1.2 in Roland may actually correspond to MinScan 30 in MegaDrum if at all they do the same thing.

I know, i know. But again, still perfect: no false trig, high sensibility, consistency and low latency (sob!) But even no real rimshots and PS...(sob sob!)

About MIDI, i,m sure i have no loops, the flow is very simple:

SPDS-----MD midi in------MD usb midi----Edrum monitor-----MIDI Vcable-----cubase Addictive (track midi in=the Vcable used). I work with MIDI from my commodore64 in the 80's and i builded trees of expander with more intricated connections...
When the bad things happen, they affect some MD parameters so i have to reload the config 1.
Is it possible to disable sysex communication if needed?