sammich playing wrong notes **FIXED**

[prein öno]

Successfully finished building a sammichSID last week. Everything went fine.

It wasnt even that difficult to build. :cool: You did really great work with the kit an the superb build guide! :thumbsup:

The only issue is, that i cannot trigger the right notes via midi.

Between C3 and C4 all is played well except G3 (plays C3) and A3 (plays D3).

Below C3 only B2 is correct. Going deeper results in large steps heading C-2.

Above C3 there are also incorrect notes in each octave.

Sysex communication has no problems. I could upload the preset sets and i can use the MidiBox V2 Editor.

(Can watch the parameter changing on the display when editing on the screen)

So i assume sysex and the midi communication in common is working.

I tried a clean patch, with one OSC, no detune, no finetune and no transposition.

For every factory preset the problem is the same depending on the transpose setting.

BL and LD mode makes no difference. I am using Rc37.

I crosschecked with another HW-Synth. Seq, cable and midi-out is okay.

Did i overlooked some "special" scale option or what can go wrong in this context?

[Wilba]

That's a very strange bug!

It is possible that one or more of the data lines to the SID are not working properly... so that most of the time things are working but you are noticing incorrect frequencies being written to the SID's data registers.

You could try the interconnection test:

http://www.ucapps.de/mios/mbsid_interconnection_test_v2.zip

to make sure the tracks between the 74HC595 and the SID are working correctly.

See the README.txt file inside the zip for instructions.

Check also there are no shorts between the pins on the SID sockets, and the 74HC595 below the SIDs.

Let's say there was a short, then in the case that both pins are outputting the same value (0V or 5V) then you wouldn't notice it, but when they are outputting different values, then they'd both be 0V, I'm guessing. Or could be something else depending on stuff I don't know about electronics (which is a lot). Either way, it might make some bit patterns still work fine, like most note frequencies in a given note range, and then be totally wrong all the time for some other note range.

You might want to test this theory by setting the pitch bend range to something large, and then hold a note down and do some pitch bending.... if you hear jumps in the pitch instead of a smooth glide, then the problem is definitely between the PIC/74HC595 and the SID and (most likely) nothing to do with the firmware or MIDI events being interpreted wrong, etc.

[prein öno]

You might want to test this theory by setting the pitch bend range to something large, and then hold a note down and do some pitch bending.... if you hear jumps in the pitch instead of a smooth glide, then the problem is definitely between the PIC/74HC595 and the SID and (most likely) nothing to do with the firmware or MIDI events being interpreted wrong, etc.

Thats exactly what happens! I got jumps on the above mentioned notes (A2, G3, A3)

But would this not also affect other registers?

Are the same Values for pitch used like here: http://archive.6502.org/datasheets/mos_6582_sid.pdf (Appendix A)?

I found out there is a correlation between the failing notes and the 11th databit!!!

Thats the reason the Octave from

midi b1 - table b2 - 123 hz - 0x0817

to

mid a#3 - table a#3 = 233 hz - 0x0F46

are not working.ALL lower notes work.

In the upper octaves all notes with 3rd digit > 7 produce wrong notes

[prein öno]

Sid's Pin 18 / D3 is shorted to ground :shocked:

[prein öno]

Sid's Pin 18 / D3 is was shorted to ground :queen:

[Amanita]

Sid's Pin 18 / D3 is was shorted to ground :queen:

Accidental circuit bending! :thumbsup:

[Wilba]

Awesome! I'll reply to the questions for completeness...

But would this not also affect other registers?

Yes, but it might not be affecting the ones that count, or are audible. i.e. filter cutoff frequency might also be wrong, but you wouldn't notice unless you were doing slow filter sweeps.

Note also that the frequency is sent in two bytes, so both bytes of the register would be wrong, but you wouldn't have heard much difference in the low bits, and probably half the time, it wasn't wrong (i.e. it was a zero bit anyway).

Are the same Values for pitch used like here: http://archive.6502.org/datasheets/mos_6582_sid.pdf (Appendix A)?

Yes, but I think TK recalculated this table using the formula in the datasheet, as there are some rounding errors in the table, so what is in the firmware is not identical to the table.

I found out there is a correlation between the failing notes and the 11th databit!!!

hehehe I was going to check if there was, but it was late and finding a short would have proved the theory right a lot quicker. :wink:

Can I ask if you used the interconnection test to discover the short to ground, or just tested each pin with a multimeter to find out what was wrong?

[nILS]

Yes, but I think TK recalculated this table using the formula in the datasheet, as there are some rounding errors in the table, so what is in the firmware is not identical to the table.

<klugscheisser>Yes, but no? ;) Wilba's right as usual, they're not identical. The table in the datasheet is off.</klugscheisser>

[Wilba]

Paraphrase fail. It is actually (yes or (yes and no))

[prein öno]

Yes, but it might not be affecting the ones that count, or are audible. i.e. filter cutoff frequency might also be wrong, but you wouldn't notice unless you were doing slow filter sweeps.

Now i notice! Some presets sound different and even the startup sound was not audible before -

But when you don't know how they should sound than you think: okay - not so nice preset :wink:

The tuning error for the 3rd bit in the low byte was apparently to small to notice.

Can I ask if you used the interconnection test to discover the short to ground, or just tested each pin with a multimeter to find out what was wrong?

No - was to lazy :) Just had a look on the SID spec to narrow the search area, played some notes and than it was obvious. (especially when having a look on the completely failing octave)

At least the highbyte values are the same as in the table.