GBA1000 68060 Turbocard build project

hese

Well-known member
AmiBayer
Joined
Aug 24, 2008
Posts
1,685
Country
Finland
Region
Savo
Welcome to the GBA1000 68060 Turbocard build thread. This time I'll be building a GBA1000 68060 Turbocard (GBA1K-TK-02) for GBA1000 rev1-rev4 motherboards.

Without further ado, let's get started. Here is the empty turbocard PCB that needs a few components.


The assembly is best to start by soldering the small pitched (0.5mm) TQFP chips first.


Some are in the belief that these chips are very difficult to solder. Actually with the right tools and technique, soldering the TQFPs is really trivial. All you need is a soldering iron with bevel or chisel tip, some solder and flux.


The soldering technique I'm using is called sweeping. First the chip needs to be anchored to the PCB by soldering a few pins in place. I anchored the chip from two sides. After anchoring a side, it's best to check that the pins are still perfectly aligned and if not, realign the chip before progressing.


When the chip is in place, the chip is soldered one side at the time by sweeping. To do this, add some flux to the pins of an unsoldered side and add some solder to the bevel tip. Then place the tip on the corner of the chip and sweep the tip across the pins to the another corner. Each pin of the side should be nicely tinned with a single sweep.


After a few sweeps the chip was soldered in place.


Another suitable technique is to solder the pins by point to point. First flux the pins, then add some solder to the tip and just touch the pins by sliding the tip on the PCB towards the pins. With this technique you are soldering a few pins at the time, depending on how wide the tip is.

Here's a good video showing these techniques in action: https://www.youtube.com/watch?v=5uiroWBkdFY
Sweeping technique is shown at 1:41 and one version of point to point at 1:20.

While I was in momentum I put the rest of the TQFPs and also the memory chips in place.


The rest of the chips were easier to solder. I continued the assembly by populating the capacitors and resistor networks of the right-hand side.


The assembly is halfway finished, the more populated left-hand side is still to be done. And the bottom side with the small 0603 package capacitors (1.6mm x 0.8mm).


Update 27.12.2014:

I began filling the left-hand side by first soldering the smaller capacitors and resistors around the big coils (L1, L2). Then I added the bigger 100uF tantalum capacitors and the rest of the smaller components (diodes, ICs, oscillator). Finally I put down the big coils.


The legs of the JTAG connector filled the pads pretty much completely, so I used a bit different approach to solder the connector. I put some solder paste on the pads, placed the connector in place and heated the legs with a soldering iron.


Only a few more components were left to be put in place, such as the 68060 CPU socket and the 68030 connector. I had a proper socket for the 68060 CPU, the 68030 connector was crafted from round pin strips.


The assembly is now finished. Next it's time to program the Xilinx chips and test the board.

Update 28.12.2014:

I chose to program the Xilinx chips onboard using JTAG. JTAG provides a handy way to reprogram the chips if needed. No need to desolder the chips, reprogram them and solder back.

Here is the JTAG pinout and the JTAG programmer connected to the turbocard.


The chips are programmed using the Xilinx WebPACK's iMPACT software. The WebPACK is freely downloadable from the Xilinx website.

Plug the JTAG programmer to the computer and start iMPACT. Let the software create a new project.


Let iMPACT automatically detect the Xilinx chips of the turbocard by choosing "Automatically connect to a cable and identify Boundary-Scan chain".


After the scan iMPACT should have found three programmable chips (XC9572XL and two XC95144XLs). Assign the jedec files (*.jed) for each chip as shown below. The jedec files can be found from the a1k_tk.zip archive found from Georg Braun's website.


The default settings how to program the chips are fine.


I programmed the chips by first selecting all chips with Select all button. Then started the programming by clicking Program. After a few moments iMPACT reported that the programming had succeeded.


The 68060 turbocard is now ready to be plugged to the GBA1000 motherboard. Almost... First the Boot ROM of the motherboard must be programmed with a boot code for the 68060 turbocard. The turbocard does not start without the boot code. The 68060 boot code can be found from the GBA1K_TK060.zip archive from Georg Braun's website.

The programming is done using the onboard 68030 processor of the motherboard. First I removed the S5 jumper from the motherboard so the Boot ROM could be programmed. Then I started the computer and erased the Boot ROM with erase_flash command and programmed a new boot code with execute gb_a1k_060_ram.


After that I turned off the computer and put the S5 jumper back to enable the Boot ROM. The 68030 processor was replaced with the turbocard. Also the 50MHz oscillator was removed from the motherboard, as the turbocard is now handling the CPU clock signal.


With excitement the computer was turned on. Did the turbocard work? Yes, I was greeted with the 68060 boot intro.


Here are a couple of benchmarks...


I used a MC68060RC50 rev6 71E41J processor with the turbocard.

The jedec files found from the archive are set to run the 68060 with 100MHz clock speed. If you have a 68060 CPU that can't handle 100MHz, you have to modify the ABEL program codes and compile new jedec files from the modified ABEL files. The ABELs are also found from the same archive. I have tested the turbocard also with 50MHz clock speed.
 
Last edited:
Seriously clean workmanship there and well explained.

Bryce.
 
Thanks for the feedback, I have updated the first post with more pictures and progress.

@Rakeim132
We will know that soon. :)
 
I got the components from a group buy organized on the A1k forum. Big thanks to halbvier for organizing it.
I've been working on a shopping cart on Mouser for components for the 68060 and Picasso cards.
 
I have updated the first post. The turbocard is now up and running.
 
Perfect job hese! (y)
How many GBA1000 do you own? You're going to receive the three ones you had ordered here to build now!

I have begun to source the components for the three expansions for the GBA1000.
Did you create a project onto Mouser? And are you going to share it for all members from Amibay who have bought the PCBs with us?

- - - Updated - - -

Sometimes ago, I have tried to download Impact software from Xilink... but I have never found the link.
I found link only for Vivado Design tool and the ISE Design Tools. Where to find Impact? Is it into one of these two archives?
 
Last edited:
I have found iMPACT... it's a part of ISE Design. Not really clear.
 
N° NAME VALUE EXT TYPE
1 L8 50R 0.2A L-1206
2 L1-L2 33uH 33UH L-33UH
3 L3-L5-L6 48R 6A L-1206
4 IC3-IC4-IC15-IC16 74ALVT16543 ABT16543 TSSOP56
61 C55-115 10nF C-0603-K@3 0603@3
1 C26 10nF C-0805-K 805
27 C19-C22-25-C27-41-C46-C47-C50-52-C117-C122 100nF C-0805-K 805
1 C116 22nF C-0805-K 805
3 C18-C20-C121 100nF C-1206-K@2 805
4 C2-C12-C13-21 100nF C-1206-K@5 1206
2 C5-C9 15nF C-1206-K@5 1206
2 C4-C8 220pF C-1206-K@5 1206
2 C1-C3 22nF C-1206-K@5 1206
4 C14-17 10uF C-1812-K@4 1812
9 C42-45-C48-C49-C53-C54-C118 4,7uF C-3528-T 3528
4 C6-C7-C10-C11 100uF C-7243-T 7243
2 C119-C120 47uF C-7243-T 7243
2 D1-D2 2A-50V D-2A-50V D-2A-50V
1 IC9 ICS502 ICS502 SO-8
4 IC11-14 K4S56163 K4S56163@2 TSOP2-54@2
1 L4 n.b. L-1206 1206
1 L7 n.b. L-1206 1206
2 IC5-6 L5972D L5972D SO-8
1 IC1 68030-SOCKEL MC68030RC@2 PGA-128M
1 IC2 MC68060RC MC68060RC MPGA-223
1 X1 MOL-4-GW MOL-4-GW@2 MOL-4-GW
1 IC18 20MHz QUARZ MM20SS
1 IC19 4MHz QUARZ-P MM20SS
1 R4 1k82 R-MELF@2 204
1 R2 2k21 R-MELF@2 204
1 R8 4k02 R-MELF@2 204
2 R1-R7 4k7 R-MELF@2 204
2 R3-R9 6k8 R-MELF@2 204
1 R45 10k R-MELF@3 805
2 R42-R49 1k R-MELF@3 805
5 R18-R23-24-R28-R47 22R R-MELF@3 805
1 R44 2k2 R-MELF@3 805
5 R17-R27-R29-R39-40 33R R-MELF@3 805
5 R19-20-R41-R43-R46 4k7 R-MELF@3 805
1 R25 6k8 R-MELF@3 805
8 R5-6-R10-15 33R R-NETZ@2 R-NETZ@2
1 R16 4k7 R-NETZ@2 R-NETZ@2
7 R21-22-R35-38-R48 1k R-NETZ@3 R-NETZ@2
6 R26-R30-34 33R R-NETZ@3 R-NETZ@2
X2 JTAG SERVO-8 LSI-SMD
X4 SL2-1 SL2-1 SL2-1
S2 SWITCH SWITCH@2 SWITCH1
2 IC8-IC17 XC95144 XC95144 TQFP-100
1 IC7 XC9572 XC9572 TQFP-100
 
I have found iMPACT... it's a part of ISE Design. Not really clear.

I've used this one in the past. Although it's old (Like me) It has worked great for me, and it's free too. Though my programmer is a parallel port one. Like this one..




http://www.xilinx.com/webpack/classics/wpclassic/ and pick ISE WebPACK 4.2WP3.0 (It's not part of ISE Design) but run it in XP compatibility mode, or better still a old laptop with XP installed......
 

Attachments

  • parallel port.JPG
    parallel port.JPG
    14 KB · Views: 1
Last edited:
Thanks a lot britlord, I have downloaded the Linux version. If problem, I'll download this version to use with XP onto an old laptop!
 
Thanks a lot britlord, I have downloaded the Linux version. If problem, I'll download this version to use with XP onto an old laptop!

Glad to help you... errr When do think that you'll get our boards..
 
Last edited:
@Tuxbar81
I have been working on the component lists on Mouser but they aren't ready yet.
I'll share them when they are finished.

Or if someone has already done the lists, feel free to share them.
 
Hi. Will a 68060 running at 60MHz work with this board..............
 
@britlord

It's possible. The speed is determined by the speed of the oscillator (IC18) and the multiplier setting of the PLL.
The PLL multiplier is set in the ABEL code, possible combinations are 2x, 2.5x, 3x, 3.33x, 4x and 5x.
The multiplier in the compiled jedecs should be 2.5x, so putting a 24MHz oscillator gives 60MHz speed.
 
Last edited:
I've been told that changing the quartz is not a guaranty of success. Not all speed will work correctly. It can work or not because of "timing problem". It's seems to be not so simple... But the best to do is to try :)
 
Last edited:
@britlord

It's possible. The speed is determined by the speed of the oscillator (IC18) and the multiplier setting of the PLL.
The PLL multiplier is set in the ABEL code, possible combinations are 2x, 2.5x, 3x, 3.33x, 4x and 5x.
The multiplier in the compiled jedecs should be 2.5x, so putting a 24MHz oscillator gives 60MHz speed.

OK. Though does the jed and other codes have to be changed... You see I can't find a 100MHz one on eBay... But I can get 60MHz one....

Oh which ABEL code has to be changed...
 
Last edited:
Back
Top Bottom