A2000 A2630 hardware on/off switch

[Spedez]

Any of you hackers have idea is it doable to have hardware on/off switch to enable/disable A2630 accelerator card? I know it operates modes (68000/68030) via software using r-mouse press on boot but that is pretty annoying to use. It won't even stay in 68000-mode after reboot. The board has jumpers mainly for setting onboard memory on/off but that's it which is not very useful.

[Matt_H]

I don't believe this is possible with the community's current knowledge and understanding of the 2630. Most accelerators with a disable function have it implemented in hardware, but the 2630 does it in software (well, firmware), so I think adding in a hardware switch is going to be complicated. The schematics are available if you're electrically inclined and want to investigate...

[Spedez]

I kind of guessed so but I still had to ask . Would be cool to have such a feature . Thanks for reply.

[SpeedGeek]

This isn't a very difficult hack or mod project. You need good soldering skills and the A2630 schematic. If you look at the schematic you will notice the PAL @ U305 has a signal called "68KMODE". This signal is asserted at a logical high state and negated at a logical low state. This signal is normally output from the Auto-Config register @ U303. This register normally comes out of RESET at a logical low state but can be changed to a logical high state by code in the A2630 boot ROMs.

So, if you disconnect the "68KMODE" signal from the Auto_Config register and connect it to a pull-up resistor and toggle switch to GND, you would then have a manual override circuit which bypasses the Auto-Config logic.   
   

[Spedez]

Wow, thanks. I might give it a go some day when I have time. I guess I can't break the board with such wirings, huh? They used to make everything bulletproof back then 

[Matt_H]

I stand corrected regarding this being possible. Still beyond my technical ability, though.

[blakespot]

Is there any chance that a hardware mod could be done to physically switch the A2620 (68020 accelerator preceding the A2630) off?

I am noticing more and more demos that suffer from the '020 being active and the slight lag it introduced to the system vs. a 68000 for tightly coded demos targeting the base hardware.


bp

[blakespot]

I want to try this with my A2620 - the schematics seems similar:

https://www.amigawiki.org/dnl/schematics/A2620.pdf


So, I would break the 68KMODE signal line coming out of 303 and connect it to a pull-up resistor to ground and also to 305 with a switch and that would do it? Can you give more info as to what type of resistor I would need here? Can you expand a bit on your suggestion looking at the schematics, as I am fine with soldering but not so much digital logic as far as my experience vs a class 30 years ago in my CS degree pursuit. Thanks!


bp

[SpeedGeek]

The A2620 is functionally the same for mode selection as the A2630. The only thing you missed was that the pull-up resistor is connected to VCC. If it were connected to GND then it would be a pull-down resistor.

Regarding the value, Commodore liked to use 1K ohm resisters but I prefer to use a higher value 2.2K - 4.7K ohm.   
   

[blakespot]

Thank you. Is there any chance you could sketch out this switch logic for me? Would a three-pole switch be used? One to ground, one to pullup resistor which goes to VCC and the 68KMODE line? I can't visualize it for certain - but solder well enough.

Or perhaps language that Chat-GPT could used to render such a circuit -- in my asks, I don't feel that I'm picturing this right based on the output.

Thanks either way. I'm trying to see if I can keep and use this accelerator (I need a switch, the softswitch boot select doesn't preserve beyond a reboot) or if I need to pull it and go pure 68000 for my particular needs. I need pure 68000 more often than accelerated. Thanks.


bp

[SpeedGeek]

What you need is already sketched out in the schematic. Look at the RAMSIZ, NORAM and B2000 jumpers. The only difference is that you would substitute an SPST toggle switch in place of the jumper.

[blakespot]

What you need is already sketched out in the schematic. Look at the RAMSIZ, NORAM and B2000 jumpers. The only difference is that you would substitute an SPST toggle switch in place of the jumper.

I see the piece you're talking about, but I guess it's just beyond me a bit -- I can't quite see what I need, from that. I can't clearly envision how the resistor and ground and +5 VCC would tie together with a switch. I will probably remove the A2620 and just run 68000 only. I don't do that much on Workbench, but my main focus is perfect execution of demos.

I may in the future if the slow Workbench is too painful look at a modern accelerator with an on/off switch as a feature.

Thanks.

[SpeedGeek]

There are three 4.7K pull-up resistors in RP201. Pin 1 is the common pin which is connected to VCC (even though it's not shown in the schematic). When the jumpers are open the three signals are pulled to a logical high voltage and when the jumpers are closed they are pulled to a logical low voltage.

If you can't clearly envision this simple circuit, then you are not (yet) at a technical knowledge level needed to perform this modification. I would suggest that you solicit help from someone with the technical skills to do the mod for you.      

[blakespot]

If you can't clearly envision this simple circuit, then you are not (yet) at a technical knowledge level needed to perform this modification. I would suggest that you solicit help from someone with the technical skills to do the mod for you.      

I could do the solder work, but I'm not 100% on the layout, and whether to cut a trace and solder there, or cut a pin and solder there, etc.

I keep looking for a switch-laden 3rd party A2000 accelerator coming thru eBay, but nothing. And I'm sure it would be quite pricey, looking at some of the things I do see coming through. (There's an 030 Derringer there now, but that seems not the ideal way to go on the A2000, direct to socket, etc.)

Are you aware of anyone in this forum community that would be willing and able to do this mod for me, for some level of compensation? Thanks.

I feel like I am kind of in a limbo with my A2000 having noted the issues that make me seek a switchable accelerator.

bp

[SpeedGeek]

I could do the solder work, but I'm not 100% on the layout, and whether to cut a trace and solder there, or cut a pin and solder there, etc.

I keep looking for a switch-laden 3rd party A2000 accelerator coming thru eBay, but nothing. And I'm sure it would be quite pricey, looking at some of the things I do see coming through. (There's an 030 Derringer there now, but that seems not the ideal way to go on the A2000, direct to socket, etc.)

Are you aware of anyone in this forum community that would be willing and able to do this mod for me, for some level of compensation? Thanks.

I feel like I am kind of in a limbo with my A2000 having noted the issues that make me seek a switchable accelerator.

bp

Finding a trace to cut probably won't be easy if the A2620 has a 4 layer PCB like the A2630. The A2630 has most traces on the inner 2 layers and VCC and GND on the outer 2 layers.

So that leaves 2 options, cut the output pin on the LS174 chip or disconnect the input pin to the PAL. Since the PAL is socket mounted it's probably easier to remove a pin from the socket.

The rest is just a matter of wiring and soldering the components which should be done with a low temperature and grounded soldering iron. Mech is a long time Amiga.org member who could do the work for you. You should send him a PM and ask him about it.