Thomas Henry Keyboard & sMs Integrated Keyboard Controller

[elektro80]

Just for the record, I want to make it perfectly clear that the Fatar group buy suggested by me on page one of this thread is not the same as the Randaleem Fatar group buy. This means that the Randaleem Fatar group buy is not an official electro-music.com community project. I don´t know who Randaleem is, I don´t know where he lives and I wouldn´t know anyone who knows him privately. Randaleem is representing a startup synth manufacturing business I know nothing about. If he wants to sell Fatar mechanisms really cheap, then that´s perfectly OK. This also means that all things Randaleem Fatar group buy should be associated with with his yahoo.com email addie and not the with any of the admins here and certainly not me. Even though I did make a Fatar group buy suggestion at the start of the thread, the current Randaleem Fatar group buy is not it. Will there be an electro-music.com Fatar initiative? I certainly cannot promise that and I wouldn´t completely write it off. Time will tell.

I am making this clear, not in order to make problems for Randal, but rather because I´m getting a fair number of PMs re this group buy from members who think this is my project off page 1 of this thread.

[State Machine]

Well said, and tactfully written. Thanks very much for that clarification. It clears things up for me Now, on to the original business of this thread, however, it was fun to see it get derailed and OT, but only for a little while.

[elektro80]

Bill, that´s right.. there will be no more diversions in here.

[State Machine]

[Dan Lavin]

Well folks, it's been a little while since the last update, so here goes:

The basic digital and analog portion of the design are complete. We did "clean up" the back end of the analog part as the result of me wiring a bad FET in. Don't worry, it is still very much in keeping with TH designs.

Bill is working on the midi portion which will allow note-on/note-off, midi channel selection and velocity control 0-127...although on a pot, not a true velocity sensitive kbd. Now remember, Bill is a very busy man between the UD-1 kits and the big black hole known as Klee, etc., so "patience" is the word now.

The pcb is about 80% done. Right now it is about 2.75" X 4.50" (70mm X 114mm), although the long dimension may have to get stretched a little bit....we'll see.

Now we've hit a snag because someone fried out the only working CPLD we had before samples could be recorded. Bill will be forwarding a new one shortly (probably along with his invoice!). Of all the parts to toast!

Anyway, as soon as the new CPLD arrives, I'll record some samples of the prototype controlling a SN-Voice so the group can get an idea of its operation.

[State Machine]

Just to add to Antman's update. I have finished blocking out the MIDI circuitry and it's associated clock generator today and starting the design entry process this weekend. As Antman said, it will output MIDI NOTE ON/OFF messages to control digital gear, the associated note number of the encoded key, and standard 127 level MIDI velocity data. The user will have the option of selecting MIDI channel numbers 1 to 16 upon which to transmit on.

The Velocity A/D chip has been selected and breadboarded with success. The chip used is a ADC0804 and used in a free running configuration and sampling a velocity potentiometer at 5.5 Ksps. More than enough.

Each time a key is pressed on this controller, the proper CV will be output by the analog section as well as a standard 3-Byte MIDI NOTE ON message to control digital gear. When a key is released, the MIDI interface shall output the standard 3-Byte MIDI NOTE OFF message with a corresponding velocity of "0".

So, the project is moving along !

and YES, Antman is doing a really fantastic job on a spiffy two sided printed circuit board !!! No worries about big IC's to solder either even though the CPLD is 84 pins. A through hole 84 pin socket, with 0.1" spaced pins, is being used and the CPLD chip will just snap into place.

Bill

[tomcat]

Some silizium left? Would be nice if the Midi IF gets inputs for two pots, so pitchbend and mod can be implemented. Not a necessarity, but ...

[State Machine]

[fonik]

i am still in the planning phase for my wildcat based performance synth. since it will be a performance synth it should incorporate a keyboard, i guess.
the fatar keyboard is diode matrix wired (two switches for each key):
diode matrix

i wonder if these diodes would do any harm used together with the TH keyboard controller circuit? if not, the fatar keyboard would be perfect.

[Dan Lavin]

Matthias,

The terminology is a little different. For TH's design, MK0, Mk1, MK2, etc. would be called row 1, row 2, row 3, etc. Then, T0, T1, T2 would be called Column1, column2, column3, etc. You would not use BR0, BR1, etc.

But to answer your question, the diodes are fine in that orientation. I had an old keyboard with the diodes reverse or the Fatar unit and those do not work, but you are fine with yours.

[fonik]

[Dan Lavin]

Matthias:

Sorry, you are right. I was thinking of the first velocity kbds which used a rest buss bar that the circuit would break then the keyswitch would make contact with an active buss bar, like a SPDT switch. This is how the old D510/E510 midi keyboard chips worked. Fatar now uses 2 parallel switches with one closing before the other, like a DPST switch.

[State Machine]

Glad there is some activity on this thread. I will be completing all the Klee parts orders by the weekend and I have got the "Integrated Keyboard Controller" board back up on the workbench It's been a while and sorry for all the delay but I was real busy over the past couple of months

The base digital design is completed and now the enhancements are being designed in.

So, expect more activity over the next month with this project

Bill

[goodrevdoc]

Excellent news! I had almost given up the ghost on looking for a suitable keyboard. Glad to hear this project is still being pursued. Now, if I could only actually play the keyboards worth a hoot...
-justin

[State Machine]

[fonik]

i just wanted to let you know that i ordered a fatar keyboard today. dieter doepfer sells them as spare parts so i had to pay something...

and then i am just curious about the controller...

[State Machine]

OK I spent time on Sunday designing in the MIDI engine in the keyboard controller and would have to say the design session was a success. One thing I feared after deciding to put a MIDI interface in this project was that I would run out of room in the CPLD before finishing completely. After about 3 hours of tweaking the device fitter properties I finally got the extra circuitry to fit with 99% macrocell utilization

The work is not done yet. I need to get the external 8-Bit A/D data into the device for the velocity portion of the message and the NOTE data, generated internally, into the message as well. I will need to pull off all the tricks in the book to get this to route to the right places in the chip.

The pictures shown here show the MIDI test setup consisting of a laptop computer with MIDI-OX software running in the role of MIDI data analyzer, a MIDI interface and the Keyboard Controller prototype.

Another photo shows a closeup of the controller with the MIDI connection to the interface.

The last photo shows a closeup of the data analyzer displaying the MIDI data coming from the controller as I simulate a pressing of a key with a jumper. The screen shows two message types. One is a NOTE ON event when the key is pressed and the other is a NOTE OFF event when the key is released. Each message consists of three bytes of data. The first byte of the message is the MIDI channel and message type (80h, 90h). Actually the "8" and "9" specify the message type and the "0" means MIDI channel 1. You want to consult the MIDI specification to refresh your memory of the MIDI word structure.

The second byte contains the NOTE number (08h) representing "G#" and the last byte contains the MIDI velocity, (55h). I hard coded a 32 bit message in the device just to prove out all the shifter logic and interface timing. I also wanted to prove out that the computer would recognize the data from the controller. It was a success as you can see.

Things have worked quite well and now to put the real data onto the MIDI message. That will be my next session.

My big fear is that I just wont be able to fit the rest into my device which means I will have to move it to the XC95144/256 instead of the 95108.

There about three other options of which I will discuss if I am not successful in fitting the remaining logic. I will see .....

Bill

---

controller_midi_test_1.JPG
Description:
Filesize:	87.78 KB
Viewed:	407 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

controller_midi_closeup.JPG
Description:
Filesize:	90.84 KB
Viewed:	372 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

controller_midi_analyzer.JPG
Description:
Filesize:	222.86 KB
Viewed:	390 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[State Machine]

I managed to fit everything into the XC95108 device without too much more trouble. I simplified the cloking logic that generates the keyscan and MIDI clocks. That left me enough room in the part to bring in the 8 bit A/D data and interrupt line for the velocity function, of which only 7 bits are actually used. The NOTE NUMBER data has also been successfully routed to the MIDI section via the chips I/O pins just in case someone wants to break into this data.

For now, the MIDI channel selection will be accomplished using a 4 position DIP switch with 4 bits representing the 16 MIDI channels for a single port.

After this goes through final checkout, the chips and parts will be shipped over to Antman for testing using his keyboard.

Bill

[Dan Lavin]

State Machine,

I'll be ready to roll! I'll try it on a number of sound modules from Emu, Yamaha, Roland, etc. to look for any anomalies. Oh yeah, and the GM Voice as well!

[State Machine]

[blue hell]

[State Machine]

[State Machine]

The "Integrated Keyboard Controller" is working perfectly. I was controlling my Nord G2 Modular demo software synthesizer sending note scales and various velocities into it with no problems !!!! It's rock solid The MIDI works spot on as they say and proves out the A/D converter interface and all the internal MIDI circuits ! I had also controlled my "Jack Dark" soft synthesizers with Abelton LIVE as "VSTi" host without any problems.

Last evening, I had to fit one last change in there because of a minor problem with note data capture in the MIDI section and I also stomped out a bug in the design that caused the "NOTE NUMBER" byte "2" MSB's (D6 and D7) to intermittently latch a logic "1"'s when they shouldn't have been! I cleaned that problem up with count truncation logic in the key scan state machine. Glad I caught that as this would have cause an intermittent "bad" note! I had to to modify the TRIGGER logic to produce a 2.5 millisecond pulse when a key is pressed just so I can gain some resources to fix these design issues.

I will send Dan "Antman" Lavins the new chips and he will do some testing for me. He will control some of his analog gear and his General MIDI sound modules that he and Thomas henry have been posting about. The results shall be reported.

After that, we will decide if and when a PC board will be done. One will most likely be developed. The design will also be released to the DIY forum on this thread in case anyone wants to build one. It can easily be breadboarded as you can see from my photo. I will supply the JEDEC configuration file for the CPLD device or I can supply a programmed CPLD chip for a small cost if you don't want to get involved with the configuration.

More to come !
Bill

[fonik]

that sounds amazing! i am really excited! thank you so much for your hard work.

[Scott Stites]