Tom's Line OCTPUS tuning

A forum devoted to mod, tips and suggestions for upgrading and rehousing your VERY CHEAP commercial stompbox to near boutique excellence.
Post Reply
User avatar
Posts: 23
Joined: 01 Jul 2013, 20:30
my favorite amplifier: the BME V-12 I built
Completed builds: no stomp boxes to date, just guitars/ Tube Amps
Location: Yuma, AZ
Been thanked: 3 times

Post by ReverendBow »

So I have been listening to Rival Sons way too much lately, and wanted to try out a Polyphonic Octave... Just because.

I scoped Ebay and picked up a Tom's Line Engineering OCTPUS... cheap... It came in the original box with original packing, but not manual..

I will say it is an interesting critter... BUT it isn't exactly "in Tune" with the note being played...

This is probably why Scott Holiday of Rival Sons paid the cash for a real POG... :roll:

Has anyone ever played with one of these, or one of the copies out there?

I am wondering if it is tweakable to play in tune

User avatar
Posts: 6
Joined: 20 Feb 2020, 19:45
Been thanked: 4 times

Post by spacecarrot »


You're probably out of luck there... In order to make any modifications to the firmware you'd have to dump it, and it is likely to be on read-protected flash storage inside of the signal processing chip. In that case you'd have to rewrite it from scratch, which is a good exercise but not really time efficient.

If you open it up and check out what hardware is being used, you might be lucky and find an external storage chip that holds the program. If you can dump that it's just a matter of working with the binary - not easy either, but with software like radare2 or the NSA's ghidra you might be able to figure out where signal processing happens. It's not a "first day of programming" thing though.

Whether or not the hardware is capable of being in tune is another question: Most polyphonic pitch shifters analyse the signal in bins: This means there's always some inaccuracy, for example an implementation might check how much signal there is in the range between 438Hz to 442Hz, and then interpret all of it as 440Hz. The number of bins is hard limited by processing speed, sample rate and available memory, so if there aren't any free resources you might be unable to increase total accuracy. However, we're only guessing that it uses an algorithm like this, it might be something else entirely too.

If it is, how the bins are interpreted and centered is another thing; if it is a sensible implementation, they might use 12-tone equal temperament standard tuning as a reference, but this is pure speculation. In this case, you might squeeze out some extra performance by fine-tuning the processor clock to shift the bin centers to match 440Hz tuning. Without modifying the binary, that would mean the processor needs to already run on an external clock - if you find a crystal in there, you could figure out its center frequency and replace it with an active temperature stable oscillator, fine tune it a bit and hope that the bins are properly spaced. You might need to transplant it to a bigger case to fit the extra hardware, This of course doesn't work for bends, but at least for the standard western tuning notes. From a programming perspective however it's unlikely that they went with proper 2**(1/12n) spacing for a low cost implementation, so you might be out of luck here and all efforts are in vain without hacking the binary.

Long story short: We personally wouldn't go for it, too much time effort and low expectations of improvement. But if you wanna take it apart and take detailed pictures of the internals and write down the markings of all the important-looking chips we could tell you more. If a ROM dump is possible we'd be interested in having a peek at the binaries, but we currently wouldn't have the time to fix things even if the hardware permits it.

All the best,

Post Reply