Blitter socket install problems

[JoeFloyd]

Atari 1040 ST C103225 Rev 1 board

Machine was working fine with no issues. Here is what I was doing to break something.

Installed PLCC68 FUNKY SOCKET ADAPTER with Blitter chip per Exxos installation instructions.

Removed jumper solder blobs next to Blitter chip.

Prior to soldering the Blitter socket in place I installed the Blitter chip in the socket and booted the machine. At this point in time the machine booted. TOS 1.02 ('87) appears have been installed (the existing TOS is 2 chip 1Mbit 28 pin ROMs). There was no option to enable or disable the Blitter in GEM desktop. I recalled reading that early TOS did not have support for the Blitter and assumed this is the reason that GEM desktop did not show the Blitter option.

I then solder the socket in place, but did not test the machine after this step.

Given the old TOS I decided to update TOS to 1.04. I tried to use 27C1001 1Mbit EPROM (32pin) to update the machine to 1.04. I burned the EPROMS (I had just used these same chips in an Atari STE to upgrade to 2.06, and the STE supports these chips explicitly with jumper settings) and following the Exxos guide for wiring a 32 pin EPROM for a 28 pin ROM socket. I then installed the 2 EPROMS and tried to boot the machine.

At this point is where I first notice an issue. Upon power up there is no noticeable activity on the screen. The monitor I'm using detects a signal and turns on, but nothing is displayed. No color. No bombs. Nada. Power light on the keyboard is on just to confirm that power is being applied to the board.

I removed the 27C1001 EPROMs and placed the original TOS ROMs back on the board and same behavior observed. So, at this point the only modifications to the board was the installation of the Blitter socket and chip, and the jumper associated with the Blitter. I removed the Blitter chip and put the solder blobs back on the Blitter jumper pads. Blitter socket remains in place. Same behavior. No activity on the screen.

At this point I began to wonder if the hack job to for the 27C1001 didn't damage something related to address or data bus.

I should have stopped at this point, but I was on a mission.

The 28 pin sockets on the mother board are too close to allow the 32 pin 27C1001 EPROMs to be installed with some physical modification to the EPROM. In order to try out the 27C1001s I had to stack a couple 28 pin sockets together to offset the EPROMs vertically to allow the EPROMs to be installed. This solution would never have allowed the power supply to be installed, so I burned a set of 6 TOS EPROMs 27C256 (I have used these in a different ST with older main board that can only accept a 6 chip TOS) and installed these EPROMS in the locations described by the Exxos TOS upgrade guide. I adjusted solder blobs to move the blob from 1Mbit to 256Kbit.

I tried booting the machine and same behavior. No activity on screen.

Last thing was to desolder the 68000 CPU and install a socket. This is for a future upgrade, but while this adds to the uncertainty as to what may be wrong with the machine. The chip desolder and socket install went smoothly. No lifted pads or damaged traces. The 68000 has clean pins.

Final machine configuration:
Blitter socket installed
Blitter chip installed in socket
Blitter jumpers set for active Blitter
6 Chip TOS 1.04 ROMS installed
TOS ROM jumpers set for 256Kbit
CPU Socket installed
Original CPU in socket

This afternoon I have poked around with an OSCOPE on the 68000 address and data lines, eprom address and data. What I have observed is that upon power up the 68000 A1-AX, etc are showing activity. I have not tested all address lines. All 6 of the ROM chips I see activity on the chip enable, address and data lines. The 68000 seems to have activity on the data lines. Again I have not checked every line. The activity lasts for about 2 seconds and then stops. Nothing further is observed on these lines after this initial burst of activity.

A primary question is that I only see activity on the 68000 address and data lines upon power-up. Reset button does seem to toggle the 68000 reset line, but I don't see address or data activity after reset. Is this normal?

Given when the fault occurred (right after soldering the Blitter socket) I''m guessing the I have shorted something under the socket. If the address or data bus has lines shorted does the observed behavior seem reasonable?

My next attempt will be to remove the Blitter socket.

Any other suggestions?

Joe

[exxos]

Given when the fault occurred (right after soldering the Blitter socket) I''m guessing the I have shorted something under the socket. If the address or data bus has lines shorted does the observed behavior seem reasonable?

If anything is shorted out and of course the machine won't boot up.. I would suggest you go around the blitter socket pin by pin the measure for short-circuits all the way around..

Of course as I've been saying for a long time, and others are now realising, you only have to slightly manhandle these machines and they just die out of the blue. There may not be anything actually wrong with the blitter install at all, a slight twist on the board in the wrong place and some solder joints break somewhere unrelated.

My next attempt will be to remove the Blitter socket.

Yikes! That will not be a easy job...I would simply just check for shorts first.

[JoeFloyd]

I buzzed out the pins for the Blitter socket testing adjacent pins for short and all pins for ground. Attached is a JPEG with two adjacent pins which appear to be shorted. I only found two ground pins on the socket. I don't have a schematic for the board revision.

Inked20200517_000328_LI.jpg (331.96 KiB) Viewed 2928 times

Inked20200517_003505_LI.jpg (363.38 KiB) Viewed 2928 times

My motherboard is the same as is seen in this post https://www.exxosforum.co.uk/forum/viewt ... =56&t=2393

The blitter pinout from https://www.exxosforum.co.uk/forum/viewt ... f=31&t=149 shows two grounds. so that seems to match what I see.

So, given what I'm seeing it does look like two pins are shorted. I did place a small piece of cardboard under the socket per the installation instructions. My best guess is that heat from soldering the may have melted some solder on the adapter pin to funky socket adapter causing a bridge. I'm going to apply the "proper amount of flux" and reheat the pins to see if I can clear the short...

Joe

[JoeFloyd]

After using the "proper amount of flux" and reheating the Blitter pins the short was removed. Installing the original mask ROM TOS version and after resetting the 1Mbit jumpers, the machine booted. This was without the Blitter chip installed in the socket. I installed the Blitter chip and tried to boot. This did not work and the screen showed random data.

I then removed the Blitter chip and everything returned to working order.

I then turned to the 6 chip TOS upgrade to 1.04. The issue here is that one of the EPROM pins was bent. After installing the 6 27C256 EPROMs and setting the 256Kbit jumpers the machine booted into TOS 1.04

Here is where things get a bit confusing. I can't completely recall the state of the Blitter jumpers during the following, but inserting the Blitter chip after having TOS 1.04 working resulted in a failure to boot. I must have had the Blitter jumpers set, since after removing the jumpers I now have a working Blitter.

20200517_020650.jpg

Thanks go to Exxos for providing the kit, chip, forum, and trouble shooting advice.

Joe

[exxos]

a relitivly simple fix then

Those solderpads, its always worth checking they are not still soldered with a meter. I've been caught out a few times where they look isolated, even under a magnifier and they are not !

[JoeFloyd]

Brief update. During testing of the Blitter install, I had the power supply balanced on the side of the case. Everything was working and then I bumped the computer and the power supply fell onto the main board. The machine rebooted and started displaying either 4 bombs (~10% of the time) or 20 bombs (~90%) of the time. As far as I can tell one of the power supply support legs made contact with the ST cartridge slot.

The cartridge slot is on the same bus as the processor and it appears that the processor data or address lines have been damaged.

As luck would have it, as part of this project I had installed a socket for the 68000 processor and after installing a replacement 68000 processor the machine is alive and well again.

This is a good example of trying to take a short cut (power supply not installed where it is safe) which then blows up in your face.