Simple DIY dsPIC Project

[JovianPyx]

The last sample has clipping. The clipping came from the output amp having too much gain. This sample is the same piece, but played on the synth with lower gain and the clipping is gone.

---

My_Brain_Is_An_Electron.mp3
Description:	My Brain Is An Electron	Download
Filename:	My_Brain_Is_An_Electron.mp3
Filesize:	1.49 MB
Downloaded:	159 Time(s)

---

[JovianPyx]

Here's a sample of what it might sound like to have 2 Harpies. I generated a recording of this sequence and then again at an octave higher. I offset the higher one in time from the lower one by some milliseconds and made this mix in Audacity.

---

Sometimes_I_Repeat_Myself.mp3
Description:	Sometimes I Repeat Myself	Download
Filename:	Sometimes_I_Repeat_Myself.mp3
Filesize:	2.36 MB
Downloaded:	143 Time(s)

---

[Cynosure]

It sounds awesome. Amazing job!

[JovianPyx]

I have two more samples. They are both from a MIDI file I adapted for the Harpie.

One has the reflection filter Fc low, the other has the setting higher.

The piece is "Entry of the Gladiators" by Julius Fucick. Many of you will recognize it. I was playing different MIDI files through the Harpie and twizzling the timbre knobs and I like the way it plays this. If you listen, I hope you like these.

---

Fucick_Entry_of_the_Gladiators_lower_filter.mp3
Description:	Lower reflection filter setting Entry of the Gladiators by Julius Fucick	Download
Filename:	Fucick_Entry_of_the_Gladiators_lower_filter.mp3
Filesize:	2.74 MB
Downloaded:	51 Time(s)

---

---

Fucick_Entry_of_the_Gladiators_higher_filter.mp3
Description:	Higher reflection filter setting Entry of the Gladiators by Julius Fucick	Download
Filename:	Fucick_Entry_of_the_Gladiators_higher_filter.mp3
Filesize:	2.76 MB
Downloaded:	57 Time(s)

---

[elmegil]

Got any Scott Joplin files?

That sounds awesome. I know, I need to find a new adjective.

[JovianPyx]

I would like to thank Tom Wiltshire (who may be a member here) and Blue Hell for their help in getting things started up with dsPIC development.

[elmegil]

And thank you for documenting your steps, so there's a chance some of us might follow (eventually...backlog.... )

[Blue Hell]

Glad I could help Scott, you're doing great stuff with it!

[JovianPyx]

I've built another DIY stripboard dsPIC with a Microchip 23K256 SPI SRAM. This is a 32 kilobyte SRAM organized in bytes. I connected the SRAM to the dsPIC with 2.5 inch wires and used no termination. The SPI SRAM operates nicely at an SPI data rate of 10 megabits per second (max for the dsPIC).

One thing I discovered is that the SRAM bar-CS pin cannot be tied low to force the IC to always be selected. The documentation could be a bit more clear about the actual function of bar-CS. They call it "chip select" which indeed it is, but it also serves to terminate the previous command. If bar-CS is simply help low, the SRAM does not function correctly. Each full transaction to or from the SRAM requires:

1) Bring bar-CS low
2) Send command
3) If write, send data. If read, accept the read data
4) Bring bar-CS high

In the documentation it is mentioned that bringing bar-CS high terminates the previously executed command. It does not clearly state that each SRAM transaction must be bracketed by a bar-CS low and bar-CS high. When bar-CS is low, it selects the SRAM and causes it to enter command mode. When bar-CS is raised, the SRAM terminates whatever it was doing (which is fine if it is timed so that the previous transaction is fully completed).

EDIT ADD:
In order for read from SRAM operations to be successful, it is necessary to make sure that the data being read from the SRAM SDO pin remains synchronized at all times. The dsPIC will shift data on SDI in whether it is a read operation or not. It is up to the driver to figure out how to do this. In my simple test program, I just wrote code to read the buffer register for each byte written to SDO. Most of these are dummy reads, but they keep the register synchronized so that when the real read data is there, it can be read out of the register. This is a bit confusing because one might think that if they are in the process of sending command data to the SRAM, then there's no valid data to read - and that is true, but if the dummy reads are not done, the SPI overrun indicator bit is set and further attempts to read the SDO line will fail. I am now looking into using DMA so that perhaps I can avoid using up CPU cycles to manage these reads. If the reads can done by the DMA controller, then this is work that gets done "for free" so to speak. (I do hope this last paragraph makes sense, it's late and I'm very tired, but wanted to add this information).

[JovianPyx]

Here is a sample of an experiment I did in C that I wish to apply to the Harpie synth. It is a bowing algorithm. In the sample, what you hear is an arpeggio that is repeated several times. Each time it repeats, the reflection filter for the karplus-strong system is "opened" allowing more high frequency energy to be retained. The model I developed allows control over the bow speed, bow pressure and KS reflection filter cutoff. In the Harpie hardware, the filter would be controlled by a pot. Bow pressure would be controlled by channel pressure and bow speed by modultion wheel.

---

karplus-strong_bowing_algorithm.mp3
Description:	bowing algorithm test from C program	Download
Filename:	karplus-strong_bowing_algorithm.mp3
Filesize:	638.85 KB
Downloaded:	38 Time(s)

---