Amiga 3000D resurrection

[Worfje]

The design of the mouse/joystick switcher is clean and straightforward.

2x SN74ALS245AN - Octal Bus Transceiver
1x DM74LS74AN - Dual D-type Flip Flop
2x L101-2K2 - Commoned SIL Resistor Network 2200 Ohm
1x D-Sub 9 pins connector female
2x D-Sub 9 pins connector male



In the reverse engineered schematic, the clock inputs (pins 3 and 11) on the DM74LS74AN is left unconnected and this could result in a floating input. Looking at the function table and the schematic, preset and clear must both be high before any clock signal rise (low to high) would change the output Q and inverse Q. This is possible if both joystick fire and mouse left mouse button press is given. The data on pin data (pin 2 and 12) is connected to VCC (+5V) and the function table shows a high on Q and low on inverse Q.
In short, leaving the clock inputs open will never result in unwanted behaviour for selecting the mouse or joystick. Still, when connecting it to VCC, it could result in some lower power usage.

74LS can sink much more current (16 mA) than sourcing it (2 mA) and the design also exploits this.

I probably will add a resistors of 2k2 between VCC and clock, direction and data inputs.
Also, I am missing the 0.1 uF decoupling capacitors (ceramic) between ground and VCC.

[Worfje]

A schematic of the 4-player adapter was easy to find and I double checked it against the KMTech version on AmigaStore.eu.



No surprises there. It has a jumper J2 which has one pin connected to ground and the other pin to +5 V (don't put a jumper on this location, it is actually a header / connector!).
Jumper J1 is really a jumper, which will allow +5 V towards the connected joysticks. I do not know joysticks which require +5 V, but it doesn't harm to supply it. The change of shortcuts I guess is very low.

[Worfje]

Back to the power supply PCB.

As a basis, I wanted to use the Chieftec HPC-360-202.
When reading about the 'cross loading' requirements and finding some information on the quality of Chieftec PSUs, I eventually settled for a modern ATX 12V PSU of Seasonic. Overkill? Yes, for sure. This PSU has been very well received / tested and shows very good results on cross loading. I hope to receive acceptable overall efficiency of the PSU without using much +3.3V current and not needing to add a dummy load on the +3.3V.

Updated the 20 pin to 24 pin ATX connector (MiniFit Jr 5566-24 male, MOLEX 44206-0007) on my PSU adapter PCB.





It is not yet finished (earth plane missing at the top layer and no TICK generation yet).

Every supply (+12V, +5V, +VID, +5V_USER, -5V and -12V) gets a fuse. All voltages have been made available on a header, which in the future will be monitored by a Adafruit feather. I thought about monitoring the currents per supply with the INA260 chip, but it would require a significant amount of PCB area.

[Worfje]

TICK generation got me scratching my head.

The Amiga 3000 model allows you to choose between TICK coming from the PSU or VSYNC (from Fast Agnus).

I found this information from Dave Haynie.
Dave Haynie stated: 'Normally, you want the 50/60Hz line, as it's more accurate in most areas than the VSYNC signal.'. He was referring to an Amiga 2000, but I assume this also applies to an Amiga 3000.

The easy part was measuring the TICK and VSYNC signals by an oscilloscope (PicoScope 2204A):



red is VSYNC (duty cycle ratio is 8/92, high is 3.85 V for 19.19 ms and low is -0.6 V for 0.84 ms, 1 pulse takes 20.03 ms)
blue is TICK from PSU (duty cycle ratio is 50/50, high is 5.06 V for 9.95 ms and low is 0.27 V for 10.04 ms, 1 pulse takes 19.99 ms)

The obvious difference in signal characteristic can very well lead to problems in synchronisation of third party hardware.

[Jope]

In practice it's totally fine to use VSYNC tick. All A500/600/1200/A4000 users for example live without a mains derived TICK. :-)

[Worfje]

Short update:

I worked on a custom back plate of the Makerbeam XL case. Now, all standard A3000D connectors can stick through.



I took a step back with the PCB design for the PSU converter. Measuring all currents of all the different supply voltages made it too complicated for me. Still needs some reviewing before I can order it.

Ordered the SCSI chip AM33C93A-16PC (here). This chip fixes the bug on WD33C93A-PL 00-04 to 06 of freezing up the Amiga in case of attaching more than one SCSI peripheral (more info here).
This was required, since the WD33C93A-PL 00-08 is much less available these days and I decided to use an SCSI2SD solution plus a SCSI cdrom.

Ordered and installed KickROM 3.1.4 + AmigaOS3.1.4 from Hyperion (the extra reset due to kickrom remapping / changes due to AmigaOS 3.9 is gone, hurray).

Ordered longer SCSI cables which allow multiple SCSI devices to be attached.

Ordered and finally received my SCSI2D v6 2020c.
This device is awesome.
Put in a SD-card, connect through USB to PC/Mac/Linux, change some settings and choose disk sizes, connect to SCSI cable + floppy power connector, format the volume and there you go (in short).
After initial configuration it will also be possible to connect it again through USB to PC/Mac/Linux and using WinUAE you can boot into workbench from this. This allows for data backup and data transfer. Such a nice feature.

Currently installed AmigaOS 3.1.4 succesfully, installed lha (using this archive from aminet) and enabled the glowicons (which have been 'hidden' on the storage disk) with the help of [ Show youtube player ] movie on youtube.
At the end of the install wizard of AmigaOS 3.1.4, I was surprised to receive the information that the required cpu libraries are not installed. The FAQ of Hyperion is a bit minimalistic on information on this matter. Luckily I found this webpage who showed me how to install the missing parts.
I deviated a bit from his steps and did not do the expert install option, but the novice. It seems that all installed correctly. The message of missing cpu libraries has vanished.

Using WinUAE and the USB option of SCSI2SD, I made a hardfile of my workbench volume. It took some time, since the USB transfer speed is not that high.

Next step is to transfer data from my IDE harddisk to the SCSI2SD device and get them games working again!

[mousehouse]

Wow, nice stuff!

I only have the v5 SCSI2SDs and they work really well - just are not the fastest things around and I'm missing the USB function to hook it up to UAE. Ah well.

[Worfje]

I have finished the PCB for the PSU converter. It will allow connecting a (suitable) ATX power supply to an Amiga 3000:





In short, you can put in a 20 pins of 24 pins ATX connector into X2 and using a cable you can connect X4 with the power connector on the Amiga 3000.
Pins 1 and 2 of X1 must be connected to each other to turn on the ATX power supply.
Every outgoing voltage/supply has its own fuse.
The +5V of the original power supply is rated up to 17.25A, so this explains the much bigger fuse and track width.
Also, the board has a copper thickness of 2 oz (70 um) instead of 1 oz / ft2 (35 um).
This reduces the required track width for a certain current / heat dissipation.
+VID and also +5V_USER are made by providing +12V from the ATX PSU to two voltage regulators (Recom R-78B5.0-1.5).
Each outgoing voltage has its own led, so when a fuse would be blown you would know which one is dead.
I also added X3 which makes it possible to connect to the tick pin on the X4.

[jostberg]

Any chance to buy a pcb for the psu converter?

[Worfje]

Unfortunately, I ordered the PCBs with HASL containing lead, so due to regulations (RoHS Directive 2011/65/EU), I cannot sell any of the prints.
Also, I haven't tested any regulations that would be required for the declaration of conformity (CE).
I would be happy to supply you with the schematic / pcb layout and required parts, but using it would be on your own risk.

[obitus1990]

This is really nice. Do you have a parts/cuts list for it?

Originally Posted by Worfje

Finalized the placement and mounting of the main parts (main PCB, daughterboard, cdrom, disk drive, hard drive(s) and fans:

[Worfje]

Hi Kirk, I assume you are enquiring on a parts list of the Makerbeam XL parts?

For the main frame I used the:
- 15x15mm alu profiles of 500 mm (4x)
- 15x15mm alu profiles of 750 mm (4x) and cut them in half giving 375 mm (8x)
- 15x15mm aluminium profile corner cubes (12p)

[Hedeon]

@worfje

I have a spare A3000D and A3000T MB and am very interested in a case for either of them :-)

[Worfje]

@Hedeon: what is your proposal? :-)

[Hedeon]

@Worfje that I am willing to pay for a custom case for either or both an a3000d and a3000t board :-)

[Worfje]

@Hedeon: what is your budget? :-)

[Worfje]

Whoops, the modern ATX PSU doesn't supply -5V...
Also, I forgot to leave room for the clips of the male connectors on the ATX connector and Amiga power connector.

So, here is REV 1/03.
I added some pads that could allow measurement of the outgoing voltages (X8 and X9).
It might be nice to add a microcontroller that guards (and shows) the voltages and would shutdown the PSU (by disconnecting the shortcut on X1) as soon as it would go out of a certain range.





They have been ordered and with HASL leadfree and therefore allowed to be sold (with additional testing).

[A3000T]

If you replace the battery with a coin 2032 battery be sure to use a LiR2032 as it is a 3.6v
(like the original battery rechargeable battery) or you may have battery issues again. The regular 2032 is 3v non rechargeable cell.

[Jope]

2032 coin cell mods include a diode to prevent charging. The 0.6V difference does not matter for time keeping.