View Single Post
Old 4th March 2013
Lives for gear
was silents's Avatar

Thread Starter
Roland MKS-70 Upgrade/Modification beta testers wanted

This great guy Fred has been working on a CPU upgrade for the MKS 70. I believe there will be extra software envelopes that will be several times faster than the stock ones. He is at the stage where he needs beta testers and has requested same over at the JX Analogs Yahoo group. I thought I'd pass along the info as I know all us Gearslutz are interested in this.

By the way, if you have an MKS-70 or another JX synth, the yahoo group is pretty active and helpful.

rolandjxanalogs : Roland JX Series Analog Synths

Here is a couple of the thread replies:

> Perhaps an overview would be helpful for someone considering upgrading
> their MKS 70? Will you be selling a replacement assigner board or just the
> CPU/ROM and the user will have to replace the CPU chip?

At this point, I have a prototype sound board code which adds 2 envelopes and 3 LFOs. Can assign each one independently to each DCO, VCF, VCA.

> Any other hardware modifications?

Not yet, but it should be possible to add real PWM to the sound boards. On the assigner board, it would be easy to add more sounds in ROM (4 banks total).

> Will you have a utility to convert existing tone sysex to the new CPU or
> will they be compatible?

Yes, the idea is that the new code will update the internal data structures to
the new ones.

> Any idea of pricing?

My intent is not to make money from this. I'd be happy if I get some contribution (to buy a JX10 and adapt the code to it, for example)


> Will the new envelopes be much higher resolution / snappoer? I see this as
> the greatest shortcoming of the MKS 70, though I don't no how much
> improvement you can get them with the existing processor. If they could be
> full 8 bit values (like the MKS-80 I think), that would be truly awesome.
> Think of the potential bass sounds....

With the 8031, this is not possible. The code is written without using any
timer, and relies on each operation to have a well-defined execution time. The 6 voices are updated one after each others, and some CV are updated twice (with a linear interpolation). The A/D has 12 bits resolution. A complete cycle (update 6 voices + some extra stuff) takes 10ms.

With the new CPU, which runs about 2-3x faster (we could run even faster by
changing the crystal too), I use a timer to achieve the same 10ms update cycle. So I could run ENV3 and ENV4 at, say, twice the speed of ENV1 and ENV2.

About using 8 bits : that would be difficult because of the way that data is
sent to the sound board. Without entering in the details, the 8th bit is used to mark commands. Note that the 7 data bits are converted into a log scale (using a table). The generated envelopes are all log based.