EQ parameter code logic - seeking experts

[arjepsen]

Hey guys.
I'm working a bit on deciphering the EQ handling at the moment, in relation to the project of trying to make a "new" software for the desk.
Most communication between the cpu unit and the two controllers in the desk are fairly simple ascii encoded commands, and most of it is relatively simple logic-wise to figure out.
However, the EQ handling is giving me some headaches:
when you change a parameter, a command is sent to the DSP board in the format "g*<HEX-CODE>h", where <HEX-CODE> is a 20-byte long number.
I have made simple python script that lets me choose a value for frequency and Q, and then iterates through every gain step, and maps out the code that is sent, but when I look over the codes and try to figure out the logic of it, my brain curls up in a corner and start making funny noises.....
Aaaanyway - if anyone feels like trying to go over some of the codes and see if they can find the logic/mathmatical patterns in it, it could be very helpful.
Regards
Anders

[Y-my-R]

I'm definitely NOT an expert in that kind of thing and just shooting the sh*t with this... but does the hex number start with F0H and end with F7H by any chance?

If so, then that's SysEx (a subset of MIDI commands, that can be custom-made by the vendor for specific devices), and the first few "words" after the F0H should be the vendor and model ID, followed by the data that makes the "adjustment" and then closed off with the F7H (kinda like </>).

But I'm pretty sure you already know what SysEx looks like... for the small chance that this is "news" I thought I'd comment. Sorry if I'm being captain TOO obvious, today

[arjepsen]

You're definately not being too obvious - I didn't know that about sysex commands.
Basicaly I remember something called sysex related to midi, but that's about my current knowledge extent in that regard

But alas, I dont think that's it.
For example, the code for the first band, all parameters at their lowest is:
0F CC 88 10 E0 7D 3A A0 0F B6 44 60 1F 82 C5 60 F0 7D 33 90

I do think I see *some* structure - I suspect that theyre five groups of four bytes.
It would be easy, if just part of it changed, when keeping frequency and Q static, and increasing gain, but it seems all five groups change when one parameter is changed.

Anyway, thanks for the input!

[Y-my-R]

Have you seen this thread?

https://www.d8bforum.com/phpBB3/viewtopic.php?nomobile=1&f=2&t=1484&start=30

Again, I have no clue about this stuff... but at the surface (I can't "grok" this any deeper), this looks related to what you're talking about (even with an EQ screenshot at the end).

I hope someone else may have more specific info...

Before I found that post, my guess would have been that it could still be SysEx (aka System Exclusive) since the D8B does A LOT of it's communication with that, anyway. So, I'd think it would save having to run everything through another translation or reference-table, if the entire communication would be handled in SysEx (but potentially while dropping the Vendor/Device IDs, since that's redundant if it's an internal data stream).
I then thought that part of the 20 bits, could as well be "addresses" like "EQ band 1" / "Q" / "+1 value" or similar.

But seeing that other post, I now also think it's likely NOT SysEx. Maybe Ralph/Munkustrap could help. Have you been in contact with him before? If not, I can PM you his e-mail or introduce you to him (if you prefer)...?

[arjepsen]

Thanks for the link - I'll have to glean over that when I get home from work.
I did previously have a written chat with Munkystrap - he helped me figure out how to double the speed on the cpu-brain communication. However, to my knowledge he did not work on the cpu-dsp communication.