G-Force 030 Combo hangs

[Pat the Cat]

One general problem for all Amigas is dodgy system clock signals that are distorted beyond the point which accelerator cards will accept. This varies by card, and can "drift" over time.

There are tolerances involved, and sometimes it is very much a case of "tuning your Amiga" to work with accelerators. The older through hole Amigas (A500, A2000, A3000, A1000) are more prone to this, and the A2000 is the biggest area board, so maybe hardest to get just right.

The A600, A1200, and A4000 are SMD, usually hold their frequencies better, but still might need some adjustment to get a very faster accelerator working. Sometimes the accelerator has off timings too. Gayle and Budgie issues means sometimes you HAVE to change a resistor or similar just to get the thing to work with an accelerator.

There wasn't much scope to test this issue at the time - not many people had storascopes or other diagnostic gear.

[Pat the Cat]

RP2 is a resistor pack, and shows damage. It has been getting hot all of it's life, probably. These are typically used to pull a signal up to (+5V) or down to (GND) if no other part of the system is directly driving the signal line at the time. On the inside they are very simple, just a bank of resistors lined up to one common rail. Saves a lot of connections than using individual resistors.

Typically, some RP designs have lower or higher resistance values than are necessary for continuous operation. Also, if they are sinking more power (Volts X Amps = Watts) then they were designed for, they will get very very hot and eventually some of the individual resistors in them start burning out. Resistors have a wattage value. If they sink more current and draw more power (watts) than they are rated for, they will die, sooner rather than later.

The snag with resistor packs is, every resistor in one has a similar resistance. Sometimes you want a bit higher, to reduce the amount of power being drawn. Swapping out just one resistor isn't possible without hacking the motherboard. If you do replace RP2, try it 2 ways - one with the same resistance values, one with about double that or less. The double one will run cooler, but might not be able to "assert" the line correctly.

You might find some of the logic chips around that area, which connect the fast RAM sockets to the rest of the system, have gone bad. You can get 74 series chips in both TTL and CMOS. Most equipment was fitted with TTL. CMOS take way less power and does the same job, but requires careful handling as it is static sensitive (this is an issue when noobs start poking around with hardware).

What revision of A2000 are you using it on? Problem might not be the card, it might be a mod done to bring early rev 4 boards up to rev 6 standard. Or there might be issues with both Amiga motherboard and card daughterboard.

The root cause I think is, accelerators need slow downs when talking to or reading from the Amiga custom chips, or chip RAM. That part of the bus only goes at 7MHz (AGA goes 14MHz), so accelerators have to slow down access to those areas, or the information gets garbled on transfer. For some reason, that used to work, but does not anymore.

The problem is vastly reduced on the A3000 and A4000, which were ALWAYS designed for faster CPUS to access the Amiga parts of a system without a hitch. OK, sometimes you have to plug in newer Busters, but the design side was there from the start. Those were designed asynchronous from the ground up. A2000s just were not, it was something they were adapted to by modification.

[Pat the Cat]

All that stuff in the yellow box is socketed FOR A REASON. It was designed to be replaced in time.

It can be difficult to source replacements in that package, at least you have all the chip numbers written on the top of them.

4.3? Yeah, time to hack it up to 6. Which was what happened at the time, some of the mods may have gone bad, and the card is actually OK.

[Pat the Cat]

You overclock any system, you add coolers, and you kiss your warranty goodbye. It goes with the territory. Having said that, ceramic 68K processor variants are generally tolerant, it's just the rest of the design that can have problems.

Overclocks are hugely individual to the specific board, and somebody else with the same board might not get the same results.

GALS (Gate array logic chips) are not common now, but they can be got, and they don't need programming. They are specific chips for specific jobs. So you CAN replace them easily, the tricky part is finding the things working and for sale (or for swap). They are socketed for ease of replacement, but the only part "exclusive" to GVP is the EPROM. Everything else is stock. PALs are different, you have to burn the gates in, so they can be made "proprietary".

Read off some chip numbers, try Google or datasheets. That should tell you if it was standard component or not. True custom gate array chips were very rare - even CBM did not make their own, they didn't have facilities to make them. Chips like Gayle, for example. Too many connections for their chip making equipment.

[Pat the Cat]

Sounds like a reasonable explanation. I can't find any info in the documentation about disabling the FPU entirely after removing it.  Any secret jumper setting I missed or didn't understand?

Usually handled by the autoconfig ROM - FPUs were an option on every system. If there was one, it got picked up automatically. No need for a jumper generally.