1088XEL Alternative Mother-Board Project

[ivop]

MIDI Keyboard connected to MIDI-IN.

 

Basically its due to the 'OR' function that automatically routes either MIDI-IN or MIDI-OUT to the built-in synth based on activity coming from either source. Since the keyboard even when idle (no keys pressed) still puts out some sort of activity pulse, this can interfere with what the MIDI-OUT is feeding to the built-in synth.

 

Unplugging or turning off my keyboard eliminates the problem during playback from the Atari.

 

If it wasn't for the activity pulse coming from my keyboard when idle, there would be no issue. Do all MIDI input devices do this? Or is this unique to my keyboard?

 

That's the Active Sense you see in midimon and every compliant MIDI device should send this pulse at least every 300ms. I have not seen a device that doesn't do this.

Edited October 30, 2018 by ivop

[+mytek]

That's the Active Sense you see in midimon and every compliant MIDI device should send this pulse at least every 300ms. I have not seen a device that doesn't do this.

 

Thanks for the info. No problem really, since just leaving out one diode (D2) would prevent any feedback of the keyboard into the on board synth for those that either don't like or don't need the feature. I suppose with a properly written application, the MIDI-IN could be passed to the MIDI-OUT to allow the on board synth to echo the keystrokes instead.

[ivop]

 

Thanks for the info. No problem really, since just leaving out one diode (D2) would prevent any feedback of the keyboard into the on board synth for those that either don't like or don't need the feature. I suppose with a properly written application, the MIDI-IN could be passed to the MIDI-OUT to allow the on board synth to echo the keystrokes instead.

 

Yes, that is what I meant earlier by "software" MIDI-THRU. It's also what the Atari ST does with proper software.

[+mytek]

In retrospect, with a bigger PIC as the MIDI controller, I could have done the internal synth enable/disable and echo selection via extra I/O pins on the PIC. Then have the PIC do the MIDI-IN/MIDI-OUT routing to the synth, while monitoring the Active Sense of the MIDI-IN, and filtering it out of the on board synth stream. Easy peasy . Well a bit more involved from a firmware standpoint, but simpler from a hardware point of view. Anyway that's neither here nor there, since I need to call it good enough and go with what I got which looks pretty solid at this point, and should meet all the main objectives.

 

Gotta watch out for feature creep .

[Kyle22]

How about the bigger PIC as a future upgrade MIDI board? Leave it alone for now since you already have boards done.

We always love features to creep in, but only when you have spare time.

 

[+mytek]

How about the bigger PIC as a future upgrade MIDI board? Leave it alone for now since you already have boards done.

We always love features to creep in, but only when you have spare time.

 

Certainly doable down the road, and the updated board could be made to fit in the same footprint and use the same interface header and mounting points. But will I do that... who knows. If I get bored enough perhaps, but that doesn't seem likely .

[+mytek]

Updated schematic for MIDI-XEL board: MIDI-XEL_schema.pdf

(Schematic now includes a BOM)

 

Note: R1 and R2 have been changed to 22K from original 33K specification to better match volume with POKEY audio.

[ivop]

I remember experimenting with different values and I had to be very careful not to overload the audio input. I know the S2 is pretty "hot" so to speak, so I guess if playing, like, 16 notes at once at 127 velocity does not cause any distortion, you'll be OK for pretty much any SIO device that utilizes audio-in.

[Mr Robot]

Bugger! I just ordered all the parts! I hope I have a couple of 22ks in my bits box!

 

Boards are on their way from OSHPark if anyone needs a board? Ill have a couple of spares

[ivop]

Well, 33k won't be that different from 22k. You could also use anything in between, like 24k, 27k or 30k. Just make sure to use the same value for both.

 

Another option might be to use a dual-ganged 20k audio pot plus two 10k fixed resistors and mount the pot somewhere on the housing. That way you can adjust the volume to your own liking. Some MIDI files are not that loud and never exceed a velocity of 100. Others are way louder. And it also depends on which WaveBlaster board you'll be using.

Edited November 22, 2018 by ivop

[+mytek]

Yeah there is quite a dynamic range to contend with, and I have found a couple of MIDI files that do tend to push the envelope such as a few Roxette songs. But if you adjust for them, then everything else is way too low in volume. Having a variable resistor would allow for that, but even then those songs would end up having passages that are very, very low in volume. I wonder if there is a utility that would let you set the maximum volume (i.e., velocity) and reprocess the song with that value in mind?

 

 

[ivop]

Hmm, I cannot seem to find a program that normalizes velocity values like you would if you normalize an audio file with, e.g., Audacity. That is, search the highest value, adjust that to 127 and adjust all other values by the same relative amount. Some MIDI/DAW packages can adjust velocities, but they take normalize to mean "make them all the same", or you can add a fixed value to all velocities. Even http://www.midish.org can't do it.

[Kyle22]

Yeah there is quite a dynamic range to contend with, and I have found a couple of MIDI files that do tend to push the envelope such as a few Roxette songs. But if you adjust for them, then everything else is way too low in volume. Having a variable resistor would allow for that, but even then those songs would end up having passages that are very, very low in volume. I wonder if there is a utility that would let you set the maximum volume (i.e., velocity) and reprocess the song with that value in mind?

 

happy-thanksgiving.png

 

 

Post the .MID files so we can try them.

[+mytek]

 

Post the .MID files so we can try them.

 

 

Here you go: Roxette Midi Song Files.zip

 

This Zip contains 2 song files, the one titled LOVEIS.MID is unprocessed, and the other titled LOVEIS2.MID has been reprocessed with the program linked to below. I used both compression and normalization when processing the file, with all the settings at their defaults. The unprocessed version would begin to distort during the high passages, but after processing it sounded much better. Have a go at using different settings to see what happens.

 

FREE Windows program (HERE) that lets you do compression and normalizing of MIDI files. Plays well in WinXP which should keep you happy Kyle .

 

BTW, using 22K resistors instead of 33K for the audio mix feeding the 1088XEL makes a substantial difference in volume level, so it's probably worth doing. Otherwise you'll find yourself having to change volume settings when going between MIDI and POKEY playback.

[+Stephen]

BTW, using 22K resistors instead of 33K for the audio mix feeding the 1088XEL makes a substantial difference in volume level, so it's probably worth doing. Otherwise you'll find yourself having to change volume settings when going between MIDI and POKEY playback.

Sounds like a dual pot 20k-50k would be a great option for adjustable control?

[Kyle22]

Both of those files sound good in WMP11. Nice song.

[+mytek]

Sounds like a dual pot 20k-50k would be a great option for adjustable control?

Yes you could probably go with a 20K pot and a fixed 15-20K resistor as Ivop suggested. However for the majority of song files I've downloaded, and I have quite few now, the fixed 22K resistors work good. And for the few songs that are a bit too low or high in volume, the Normalizer program will make those work much better.

 

 

Both of those files sound good in WMP11. Nice song.

Glad to hear that. Yeah I like quite a few of Roxette's tunes, and most of them have been sequenced for midi and available for download.

[ivop]

Good to hear you found a solution. I was pondering on writing a small midifilter utility so i could do:

for i in *.mid; do
    midifilter --normalize-velocity 100 < "$i" > "normalized$i"
done

but I see your new found tool also supports batch processing. Only downside (for me) is that it's a windows tool.

[+mytek]

I see your new found tool also supports batch processing. Only downside (for me) is that it's a windows tool.

It seems to work well under WINE in linux, so I would assume the same would hold true for a MAC as well. Not sure how to get the built-in midi player to work, but I can always preview the results under some other media player instead.

[Mr Robot]

Ok where was that mcu software? Off to the 1088xld thread I go!

[Mr Robot]

Looks like I'll have to wait for a pic2stik version. No biggie, I'm still working on the case and controllers and multijoy and wifi R: device :-)

[+mytek]

Looks like I'll have to wait for a pic2stik version. No biggie, I'm still working on the case and controllers and multijoy and wifi R: device :-)

I thought I uploaded the firmware in both the hex and JOY2PIC versions someplace??? No matter next time I get on my desktop I'll either link to it or re-upload it here. Really need to create an entry for it at my AtariBits website .

[+DrVenkman]

I'm pretty sure you did somewhere in the 1088XEL Builders Thread at least a time or two along the way.

[Mr Robot]

I found some source code and a midi file in the xld thread but didnt think to look in the builders thread. Ill check it tomorrow

[Mr Robot]

Found it. It's in this very thread, one page back.

 

http://atariage.com/forums/topic/261147-1088xel-alternative-mother-board-project/page-53?do=findComment&comment=4134039