Enhanced Memory ME-1 for Kawai K5000
Kawai K5000W
The early project phase
23.8.2006
I received an email from Jan, asking if I could manufacture a memory expansion for a KAWAI K5000. This is the beginning of my project.
There are just pictures of the ME-1 at this point. However, there is no chance to get an original ME-1 board.
ME-1 component side
ME-1 battery side
The pictures show the following major components:
- TC554161 by Toshiba. A 256k x 16 Bit static RAM. Exotic part made in Japan.
- TC1027 by Microchip. Voltage reference plus quad ultra low power comparators.
- SMD connector
- Battery holder with CR2032 Lithium battery
3.12.2006
There is one problem concerning this component. It is a SOT23-5 package with Marking Code H4K. I found no manufacturer or datasheet for this component. There is no entry for H4K on any of the common cross references.
4K SOT23-5
Any hint is welcome to my below mentioned email address. Other pictures where this component may have an alternate marking are also welcome.
The function of this component could be reverse engineered using an original ME-1 board.
An original ME-1 board is also needed for the final circuit analysis. Unfortunately, I have no original ME-1 board.
For this reason, the project is put on hold now for indefinite time, until somebody can provide me an original ME-1 board.
22.12.2006
Thanks to Jan we are able to grab a K5000 along with a ME-1 memory board from the "worldwide online market place". This will be the starting point of the reverse engineering.
5.1.2007
The mentioned IC seems not to be a TC1027 by Microchip. It makes no sense to connect a TC1027 this way. The logo is quite similar to Microchip's logo, but not exact the same.
It is a SO-16 widebody IC with 5,5 mm body width. The numbers 643 and 1027 are printed on it. I did not find the logo in the internet IC logo catalogues. So I am still searching for manufacturer and type of this IC.
19.1.2007
Found out that "1027" may be a MM1027 by Mitsumi. A company normally known from disk and CD drives. It is a NVRAM controller with system reset, backup battery switchover, chip enable output and a two level low bat warning. A circuit for converting SRAM into nonvolatile RAM.
Having the datasheet now I can go on analyzing the circuit. The MM1027 is obsolete since years, so I will have to find a replacement.
24.1.2007
The circuit is completely analyzed, revealing also the part of the H4K. It's a single OR gate, gating the chip enable line from K5000 on power down, preventing the RAM from data loss.
There are different manufacturers for NVRAM controllers, as Dallas/Maxim, Texas, ST, and Fujitsu. Some have a CE-gate built in, some have a low bat monitor. All of them are expensive and hard to buy if you want only a few.
I checked also a new part named MR2A16A by Freescale Semicond. I is a nonvolatile MRAM memory, but only suitable for 3,3V systems.
26.1.2007
I decided to use a Maxim MAX691A. It has battery switchover, CE gating and a powerfail comparator. And I found a supplier where I can order from 1 pcs. up for a relatively low price.
A low bat monitor draws nearly no current; a powerfail comparator however needs a resistor network, dividing the battery voltage down to 1,25V. A resistor network will always draw a small current from the battery.
Using a 1µA current for low bat detection, and 0,5µA standby current for RAM and MAX691A, I calculated a battery lifetime of 17,5 years...
Time to think of using a smaller battery.
