[sdiy] NVRAM for TR606?

[brianw]

Kenny,

That makes sense about the Pixie emulator. Most modern MPU chips have everything that would have been on a separate chip, like SRAM, in the TB-303 era. Plus, they have Flash memory that doesn't need a battery, but does need different programming.

I got the impression that Antti has a stock Roland TR-606, without anything like the Pixie to emulate the old features.

I went looking for a modern NVRAM replacement for a Buchla Thunder memory card after it's battery died, and found that they're available in a model that doesn't even need an external battery. However, these chips still have battery chemistry inside, because they're not Flash or FRAM. Fortunately, I fixed the battery, because having a battery inside the chip seemed like it would be even more difficult to replace than the original design.


Antti,

I like the suggestion to use a lithium battery and diode. I've never seen a CR2032 leak - they just lose charge after a decade or more. The only challenge might be to isolate the SRAM so that the battery doesn't also feed anything else. Thus you'd want a second diode from the original power source, so that the SRAM will take power from anywhere it's available. You'll see this circuit in many vintage synths.

On the TR-606, looks like IC 8 is the CMOS RAM; a µpD444C part. Pin 18 looks like power, but that same trace feeds power to every semiconductor on the board. You'd need to either cut the trace or use a socket and daughter board to separate IC 8 power from the rest of the chips, then use a couple of diodes so that original power or battery power can feed the RAM. You'd just need the data sheet for the µpD444C to look up the minimum voltage that it needs to preserve memory. Sometimes that voltage can be lower than normal operating voltage. Of course, any such modifications to vintage gear could either reduce (or increase) its resale value.

One diode protects the battery from being "charged" by the original power supply. These batteries are not rechargeable, and they don't like being fed voltage or current. The other diode stops the battery from being drained by keeping the CPU and other logic chips on all the time.

Brian Willoughby


On Apr 1, 2023, at 5:59 AM, kenny at beatkamp.com wrote:
> Hi Brian,
> 
> In this specific instance, the Pixie implementation for TB-303, TR-606 (and I believe
> but cannot confirm TR-808) from Paul at RE-303 negates the need for external RAM.
> This is internalized on the CPU itself. The Pixie is running the original Roland
> code with a tiny bit of MIDI stuff hung on to it. ie) D650C emulator
> 
> Cheers,
> ./k
> 
> On 2023-04-01 01:26, brianw wrote:
>> Do you have any data sheets or part numbers for NVRAM?
>> Last time I checked, they either have a battery built-in, or they require different code to work with SRAM interface timing. In other words, I don't think you can retrofit. If someone has a part number that would work, I'd like to know how it works.
>> The other solutions people are talking about use a new CPU design where the code can handle modern non-volatile memory.
>> Brian
>> On Mar 31, 2023, at 11:36 AM, antti.s.pitkamaki--- via Synth-diy <synth-diy at synth-diy.org> wrote:
>>> I hate leaky batteries. Does anybody know if there is an NVRAM solution for the Roland TR606? I don’t want to keep batteries in it, but it looses memory contents without them…
>>> At least for pinball machines some have replaced the original volatile memory chips with NVRAM equivalents, but googling about a solution for the TR606 yields nothing.
>>> Cheers,
>>> Antti