// 2026-03-19 // by Elder0010
This machine belonged to the Raoul’s loot. It has a very important value for him, as it was his first own computer, where he developed his passion for computers and programming. The Amiga 1000 is the first model of the Amiga line, and it was released in 1985. It was a revolutionary computer for its time, with advanced graphics and sound capabilities, and it was widely used for gaming, multimedia, and creative applications.
The front expansion is the standard 256KB one, so the machine has 512KB of chip ram.
I found a DKB insider I expansion inside, providing: - 32 DIP sockets for 1 MB RAM - 256k×1 DIPs - autoconfigures under Kickstart 1.2, requires AddMem under Kickstart 1.1 - the starting memory address is DIP switch selectable to one of nine address blocks - the memory can be disabled / enabled by software - battery backed up clock (RTC)
Unfortunately the drive bay was hosting a standard 3.5” panasonic drive, which is not the original one. The original drive was a Mitsumi 3.5” drive, which is quite rare to find in good conditions. The panasonic drive was working, but it made the frontal panel impossible to close. Luckily, I found a good condition mitsumi drive in the spare parts box, and I replaced it. The original drive is now working, and the frontal panel can be closed again. The drive eject button was missing, so I 3D printed a new one and installed it.
The case was in good conditions, with some scratches and signs of wear, but nothing too bad. I cleaned it thoroughly with isopropyl alcohol and a soft cloth, and it came out looking quite good. The metal shielding was a bit rusty, so I repainted it with some metallic spray paint after applying a rust converter. The paint job is not perfect, but it looks much better than before.
I found a couple of usual RIFA filters in the power supply, so I replaced them immediately. Some of the caps on the secondary were bulging, so I replaced them as well. I turned on the power supply with a test load, and it was working fine, providing the correct voltages. The fan (which is a standar 230V one) was working but a bit noisy, I had to open it and clean it thoroughly to make it quieter. I also lubricated the bearings with some Singer sewing machine oil, which is a very good lubricant for this kind of thing.
Besides a deep cleaning, the motherboard was in good conditions and perfectly working.
Half of the keys were not working, and the other half were working
only sometimes. I had to disassemble the keyboard and clean it
thoroughly. All the problematic switches have been desoldered, cleaned
with isopropyl alcohol and reassembled. After this process, the keyboard
was working perfectly again. 
And here is where the fun begins! The DKB insider I expansion was not working at all, and it was causing the machine to crash at startup. There’s not much information about this expansion, but in short here’s what I discovered about it: - There’s a programmable PAL chip, handling the autoconfiguration and the memory mapping. I haven’t found a dump for it, so I had to blindly trust that it was working fine. - The RAM chips are controlled by the THCH4502B, a DRAM controller. - RTC (a Dallas DS1216E) is directly controlled by the CPU, but it’s not a standard one (like the usual OKI M6242B). More on this later. There’s a DIP switch to enable/disable the expansion, and to select the starting memory address.
There’s not much information about it online, but here are 2 working configurations I have found by trial and error:All the trapdoor expansions for the Amiga 500 are mapped to 0xC00000, so I guess that the autoconfiguring mode is the one that should be used for compatibility reasons.
The DKB Insider was installed and the Amiga booted fine, with memory recognized correctly. However Workbench reported no clock, and the DKB RTClock utility could not detect the RTC. The DS1216E appeared completely dead.
The DS1216E is not a simple RTC chip. It is a 28-pin DIP socket with an embedded microcontroller, quartz oscillator, and lithium battery inside the socket body itself. A small PCB sits inside the socket housing, and the ROM chip normally plugs into the top of the DS1216E, which in turn plugs into the board. The RTC function is completely transparent to normal operation - the ROM chip works normally at all times. It’s important to understand that there are 2 variants of the DS1216E - a ROM variant and a RAM variant. The ROM variant is used in the DKB Insider. The communication protocol for the ROM variant is quite unusual.
To communicate with the RTC, software must send a specific 64-bit unlock pattern to the chip using the address lines A0 and A2, after which the RTC registers become accessible for reading and writing. The chip outputs data on DQ0. This protocol is entirely different from the various RTCs usually found in the Amiga. The internal battery is a lithium cell permanently embedded in the PCB inside the socket body in epoxy glue. Dallas shipped these chips with the battery disconnected to preserve freshness, connected automatically on first power-up. After 40+ years, these batteries are universally dead.
The DS1216E contains two lithium batteries embedded in the internal PCB, which are now dead and causing the RTC to be non-functional. The first step was to remove these batteries. This involved carefully removing the epoxy glue on the back at first. The batteries were soldered onto the PCB, so I had to remove them and isolate their contacts. It was important to be careful during this process to avoid damaging the internal circuitry of the chip.
Rather than immediately assuming the chip was dead, I decided to test it systematically using an Arduino Leonardo as a bit-banging interface. This turned into a surprisingly deep dive into the DS1216E protocol. The correct wiring for the 28-pin DS1216E to an Arduino Leonardo is:
| Arduino | DS1216E Pin | Signal |
|---|---|---|
| D2 | P1 (A0) | RST |
| D3 | P27 (A2) | WE |
| D4 | P22 (DQ0) | OE |
| D5 | P20 (DQ0) | CE |
| D6 | P8 (DQ0) | A2 |
| D7 | P10 (DQ0) | A0 |
| D8 | P11 (DQ0) | DQ0 |
| 5V | P14 (VCC) | GND |
| GND | P28 (GND) | VCC |
The communication protocol for the ROM variant works as follows:
The recognition pattern is: {C5 3A A3 5C C5 3A A3 5C} sent (LSB first). After extensive testing I confirmed that the chip logic was at least working - the pattern unlock was working and the chip was responding. However no data was being retained between operations, confirming the battery was dead.
The DS1216E socket body contains a small PCB with the internal controller chip. This controller has two pins that require battery positive voltage - pin 4 and pin 14 of the internal controller IC. Both must be connected to the external battery positive for the chip to function correctly. Connecting only one of them is not sufficient - this was a critical discovery that had been causing intermittent and inconsistent behavior throughout testing.
The fix:
After this mod the chip began retaining time correctly between power cycles.
The DKB Insider does not use the standard Amiga battclock.resource - it comes with its own utility called RTClock on the Insider KwikStart disk. Workbench will not automatically detect the clock without some manual setup. The solution for Workbench 1.3 on the Amiga 1000:
C:RTClock date
The RTClock command reads the time from the DS1216E and sets the Amiga system clock. The date command is necessary on the Amiga 1000 specifically to propagate the time correctly to the OS. After this the Amiga boots with the correct time automatically every time.
A chip that appeared completely dead turned out to have perfectly intact logic - it just needed a new battery, correctly connected to both required supply pins. The DS1216E is a remarkably resilient piece of engineering. With a CR2032 it should run reliably for another 10 years. Total parts needed: one CR2032 coin cell holder and one fresh CR2032 battery. Total cost: under €2.
