New build V4.1: MIDI ports and irritation

Building Zynthian
1

So, I’ve put together a nice little 8GB Pi 4 Zynthian - it’s all gone smoothly, except…

One MIDI cable appears to have reversed wires on the plug.

I’ve now removed it from the equation (it’s on Thru with the LED connected and no devices) but nothing is making MIDI work and any attempt to play the demo MIDI test file (which doesn’t play via the MIDI out) sounds like total warped gibberish with sort of recognisable bits of the tune) is unlistenable.

I’ve burned the image twice and checked my wiring etc. but this seems like such a bizarre failure. The screen is working and when I plugged in a USB MIDI keyboard that worked fine for controlling the synths - it’s just the internal MIDI ports that aren’t working. Faulty board? How does MIDI clock work/timing, given the bizarre playback from a clean install on the MIDI test?

Plugged a MIDI signal tester into the MIDI out and I just get a solid red light, too - stands to reason as that’s all the activity LED is equivalent to. Something’s clearly not right, but as you can’t get the ribbon ‘wrong’ and have a working machine essentially as far as I can tell - what’s the issue?

This is more irritating than if it just didn’t work. To actually use it I’ll need it to be a class-compliant MIDI device, not host. But I think part of it is this duff MIDI cable.

Related: Should the MIDI out LED light as soon as MIDI device is connected to it? MIDI in flickers with I use a MIDI device (not if I use the miswired-looking cable). Either there’s something wrong with the Zynaptik module, or the reversed-wire MIDI cable has damaged that module, or I’m doing something wrong with the software, but it’s proving very difficult to find an actual ‘same situation’ answer on here or in the wiki :confused:

Aside from this pretty impressed with what I have put together and seen working.

2

There are pitch bend messages all over the MIDI log when playing the test MIDI file, which corresponds with how awful it sounds.

There is no MIDI controller connected to it and this is a clean config. Even when there is a MIDI controller connected on the MIDI port I get nothing in the MIDI log; I have removed both out and through; MIDI in LED is flashing with input from controller keyboard when tested.

CH#01 NOTE_ON 61, Vel: 74
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 65, Vel: 74
CH#01 PITCHWHEEL 0
CH#01 NOTE_OFF 63, Vel: 64
CH#01 NOTE_OFF 65, Vel: 64
CH#01 NOTE_OFF 78, Vel: 64
CH#01 NOTE_OFF 58, Vel: 64
CH#01 NOTE_OFF 61, Vel: 64
CH#01 NOTE_OFF 39, Vel: 64
CH#01 NOTE_ON 44, Vel: 97
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 51, Vel: 78
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 62, Vel: 103
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 82, Vel: 107
CH#01 PITCHWHEEL 0
CH#01 NOTE_OFF 44, Vel: 64
CH#01 NOTE_OFF 51, Vel: 64
CH#01 NOTE_OFF 62, Vel: 64
CH#01 NOTE_ON 39, Vel: 102
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 40, Vel: 61
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 59, Vel: 86
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 58, Vel: 72
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 61, Vel: 71
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 65, Vel: 70
CH#01 PITCHWHEEL 0
CH#01 NOTE_OFF 58, Vel: 64
CH#01 NOTE_OFF 61, Vel: 64
CH#01 NOTE_OFF 65, Vel: 64
CH#01 NOTE_OFF 40, Vel: 64
CH#01 NOTE_OFF 59, Vel: 64
CH#01 NOTE_ON 51, Vel: 80
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 58, Vel: 83
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 61, Vel: 80
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 65, Vel: 79
CH#01 PITCHWHEEL 0
CH#01 NOTE_OFF 51, Vel: 64
CH#01 NOTE_OFF 58, Vel: 64
CH#01 NOTE_OFF 61, Vel: 64
CH#01 NOTE_OFF 65, Vel: 64
CH#01 NOTE_ON 62, Vel: 57
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 58, Vel: 71
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 61, Vel: 71
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 65, Vel: 71
CH#01 NOTE_ON 65, Vel: 71
CH#01 NOTE_OFF 65, Vel: 64
CH#01 NOTE_OFF 62, Vel: 64
CH#01 NOTE_OFF 61, Vel: 64
CH#01 NOTE_OFF 58, Vel: 64
CH#01 NOTE_ON 51, Vel: 81
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 34, Vel: 102
CH#01 PITCHWHEEL 0
CH#01 NOTE_OFF 39, Vel: 64
CH#01 NOTE_OFF 51, Vel: 64
CH#01 NOTE_OFF 82, Vel: 64
CH#01 NOTE_ON 40, Vel: 40
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 59, Vel: 83
CH#01 PITCHWHEEL 0
CH#01 NOTE_OFF 34, Vel: 64
CH#01 NOTE_ON 35, Vel: 44
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 62, Vel: 95
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 65, Vel: 58
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 61, Vel: 61
CH#01 PITCHWHEEL 0
CH#01 NOTE_ON 58, Vel: 53
CH#01 PITCHWHEEL 0
3

Unplugged the Zynaptic board. MIDI test file plays back fine.

FFS.

Going to assume that the MIDI connector with the red/black wires the opposite way to the others has, in fact, done something unfortunate to the board. Given UPS managed to get the Zynth to the local sorting office then sent it right back home and delayed it, I don’t hold out much hope for this working while I had time to get to know it before work got manic again.

IMG_1132
IMG_11324032×3024 2.03 MB
IMG_1131
IMG_11313024×4032 1.64 MB

4

I see black and red swapped on the middle din connector which is wrong.
Also for better contact on the leds its good to tin the led leads a little.

5

Hi @Richard

Try loading this snapshot:

002-FluidR3 GM.zss (41.5 KB)

before playing the MIDI demo. It will setup your zynthian as a General MIDI device with the Fluid SoundFont. It will be included on the next update.

Zynthian can work as a MIDI Host (the 4 USB type A ports in the side) and also as a MIDI Class Compliant Device, using the USB type C connector in the back, that is normally used for powering Zynthian. By connecting the Zynthian to your computer, your computer must provide the power for the zynthian, or use a Y cable for injecting extra power.

The zynthian MIDI activity LEDs use the MIDI’s “current loop” for flashing . It’s un-ortodox, but it works nicely with all the MIDI devices i have tested with and has the advantage of detecting the MIDI connection at the lowest level.

If you don’t like this design, you can bypass the LEDs by closing the JP1 & JP2 solder-jumpers on the Zynaptik PCB surface, obtaining a “more standard” MIDI circuit.

Regarding the MIDI connector with crossed wires, sorry by the mistake. I’m not sure of the result of connecting the MIDI Thru reversed like it seems on your photo, i never tried, but if you suspect the zynaptik module is broken, i would send you another one, of course.
If you didn’t fix the crossed wires of the MIDI-THRU and don’t like to solder, i could send you a new one, of course.

Regards,

6

I’d originally put the reversed one on MIDI in, I think - I’d need to check my build pictures. Yes - I did - but I swapped it over to out before testing because of a picture in the wiki. Argh.

There was nothing in the wiki about the connectors being different, and when I looked at the pictures they were inconsistent as well, so figured the logical ‘difference’ might be that the one for out was different in some way - there’s probably some documentation that makes it clear in the repository, but I was looking at Wiki/build guides when assembling.

V4 viewed from ‘inside’ - MIDI out is different.

V4.1 viewed from ‘inside’ - MIDI out is different.


V4.1 viewed across - all three match.

The MIDI test played back fine when first assembled, but that was before I hooked up a controller - when hooked up, it initially didn’t show any activity on the MIDI in… so I then looked more closely at it and checked the V4.1 and V4 guides and realised that there may be a problem - but at that stage, I think the damage was done.

With the Zynaptik board disconnected, the Zynthian runs fine playing the MIDI Test with MIDI channels I’ve assigned so I have to assume that the reversed MIDI in wires fried the board (as I only swapped them over to other ports after seeing the odd behaviour, like the MIDI LED being on the instant I plugged a device into the out port.

I like the MIDI activity LEDs and have no problem at all with how it’s implemented! Just frustrated that I had to wait four more days for the kit because of UPS (I’ve never had a perfect delivery from them - I must have upset the gods of UPS once), then have a component I didn’t realise wasn’t right which wiped out a chunk of the Zynthian-ness of it.

I haven’t soldered anything more complex than a car wiring loom in 5 years, eyesight’s too messed up to focus and no depth perception so I tend to miss my target… :slight_smile: I would very much appreciate a correct MIDI socket and a working board! Apart from this it’s gone together exactly as it should - from assembly to SD-card build to playing a multi-instrument setup from the Keystep Pro.

1 Like
7

This is totally normal. The PB messages are added by the zynthian’s MIDI router and they are used for tuning the engines globally. By default, at 440Hz, a PB 0 is sent after each note-on. Try moving you pitch wheel while playing notes on your keyboard, you will note that this “tunning” PB messages also tracks the external pitch-wheel. You can change the global tuning from the webconf’s MIDI options to fully see how it works.

Until now, nobody reported problems with this. but @riban is not 100% convinced of it. And i start to think that we should remove the spurious PB 0 messages after note-on when zynthian is tuned to 440 Hz, what it’s most of times.

Regards,

1 Like
8

Ups! This photo is wrong. MIDI-OUT is reversed. The 3 MIDI connectors must be identically connected, with black at left and red at right. You found a “bug” in the wiki’s documentation. I will replace the photo ASAP.

Thanks!!

1 Like
9

Yep, when it played normally they’re there - but at that stage I was just wondering WTH was causing it to sound so weird and that seemed logical even though there’s no PB ‘value’ as such. I think whatever has gone wrong on the board was just ‘really screwing everything MIDI up’; I made a video of it but it’s a faff to share.

Have just been playing with layering and transposing synth engines - once figured out the UI is really quite good (short/bold/long press timing without nudging the encoder is still defeating me, but then my case isn’t screwed shut yet). I’m running it off a big QC USB-C charger I had lying around and will sort a USB-C splitter to use it hooked up to the main MIDI/USB setup; counted that I have 19 synths (including small ones, excluding the Zynthian - it’s cheating to count Pocket Operators, Monotron and NTS-1/Modal Craftkits in that total?) at the moment so no wonder wiring it all is proving to be a bigger task than expected; I’ll probably pop this one on the desk next to my Keylab and iMac for easy fiddling with complex snapshots.

The output on this is SO hot, too. No gain needed on the Korg desk at all.

You see why even after noticing the reversed wires, I was a bit unsure if actually that was how it was supposed to be? I also checked the small connector hadn’t got black/red reversed, as I think there’s a mention of one of the components having been done that way at some point.

10

Yesss! The new balanced outputs are hot, jeje! :+1:

1 Like
11

I just improved this:

Simply update and test.

Regards,

2 Likes
12

It’s definitely different…

CH#01 NOTE_ON 65, Vel: 56

CH#01 PITCHWHEEL -8192

CH#02 NOTE_ON 70, Vel: 56

CH#02 PITCHWHEEL -8192

CH#01 NOTE_ON 58, Vel: 58

CH#01 PITCHWHEEL -8192

CH#02 NOTE_ON 63, Vel: 58

CH#02 PITCHWHEEL -8192

CH#10 NOTE_ON 35, Vel: 44

CH#10 PITCHWHEEL -8192

CH#10 NOTE_ON 63, Vel: 91

CH#10 PITCHWHEEL -8192

CH#02 NOTE_OFF 51, Vel: 64

CH#01 NOTE_ON 80, Vel: 117

CH#01 PITCHWHEEL -8192

CH#02 NOTE_ON 85, Vel: 117

CH#02 PITCHWHEEL -8192

CH#01 NOTE_OFF 61, Vel: 64
1 Like
13

I’m getting the same behaviour. I haven’t been able to check the MIDI logs, but I imagine they would look something like this.

14

Hy,

I have the same problem, all notes are played 2 half tones down. See although Pitchbend Error

Best regards, Andy

15

I was just trying to work out what what going on with the PB messages too - I thought I was adding spurious messages over the serial somehow (I have my own MIDI interface and was driving it from a microcontroller), so I guess I’m kind of relieved to read it was meant to be like this!

I do wonder though, does anyone know what this does to a MIDI running status stream? From a few quick experiments sending a byte stream, it does seem to get broken up, at least when reported in the log anyway.

If that is true, then does that mean that a Running status note on for three notes - i.e. 7 bytes - becomes three individual note on and pitchbend message pairs, so I guess it is now 15 bytes…?

Kevin

16

The PB problems were solved some time ago. If you have updated your zynthian, it should work OK without “strange” PB messages.

But …

  • If you fine tuned your zynthian to a different value than 440 Hz, then you will see extra PB messages that are generated internally for changing the tune. These extra PB messages are for internal use only and they are not sent to MIDI output ports.

  • If you have a full-featured zynaptik circuit and have configured some of the 4 analog inputs as MIDI Pitch Bend, then you will see lots of PB messages coming from these analog inputs.

Regards,