Enhanced Memory ME-1 for Kawai K5000
Kawai K5000W
The Testing
5.5.2007
There are several ways for the K5000 OS to find out if a ME-1 is installed.
One is using a free pin on the connector to send back a 5V signal. There are some pins on the ME-1 socket that might be used for that purpose. But none of them is connected on the ME-1 original.
The other and more popular way is accessing some RAM locations and testing if data keep stored. This seems to be the way the K5000 checks for an ME-1. Accessing RAM will only work with an unprotected CE line; this is the problem using a MAX691A with 200 ms delay.
The good news: Afte removing the short ciruit, I plugged the prototype into my K5000 and it worked. I tested the CE line to find out the time of the first access.
"Test bench": K5000 upon a IKEA chair. The scope probes are hanging down from the ME-1 prototype below
CH 1: Chip-Enable, CH 2: 5V supply. No access to CE for more than 200ms after power-up
More than 1000(!) ms after power-up: the first, only few µs wide CE burst
CE pulses after power up. Here the ME-1 is checked. Similar pulses can be seen while Bank E and F sounds are accessed.
A single CE pulse is about 200 ns wide.
The first ME-1 access is about 1 s after power up (K5000W with S/W4). The low battery feature is always operating. There will be testing neccessary on the K5000S and K5000R version.
6.5.2007
Got information that a Mitsumi PST575CMT reset controller is used in the K5000 series. The reset cycle time is set to 13 ms, as expected shorter than 200 ms. There are 4 or 5 OS versions for each variant. No guarantee for a 1000 ms startup delay on all combinations.
There are two ways to continue the project:
1. The existing design is tested on all combinations of hardware and OS version. On success I am through with this project.
2. I change to the Fujitsi MB3790 reset controller (<1 ms reset delay), adding a single gate for CE protection. The lead time for MB3790 is 17 weeks. There is no other NVRAM controller known to me with a comparable short reset cycle time.
The first way is the one I like much better, so I tested my K5000W with OS versions 3.0, 3.1, 4.00 und 4.03. The startup delay was always 1 s.
I am looking for K5000S and K5000R owners to test my prototype.
8.5.2007
Jan will do the testing with his K5000S, and Manfred with his K5000R.
18.5.2007
Today Jan and me tested the K5000S. The startup delay was 1,04 s for all 3.0, 4.00, 4.01, 4.03 and 4.04 OS versions.
After inserting an empty ME-1 board into a K5000 for the first time, the byte sequence "A5 5A" can be found at the very beginning of the memory area.
In binary format this is equal to "1010.0101 0101.1010", a typical bit pattern for testing memory presence and data line functionality.
000000 - A5 5A A5 5A A9 72 50 65 35 77 73 72 12 37 54 70 - ¥Z¥Z©rPe5wsr.7Tp |
19.5.2007
Today I received Manfreds test report. Everything works fine also on the K5000R. Here are some pictures:
ME-1 replica (prototype) mounted in a K5000R
K5000R with ME-1 replica. The text string "Enhanced Battery OK" appears at power-on, proving that the "R" recognized the ME-1 replica. OS version 4.04 is loaded.
K5000R accessing sound bank F .
