How to upgrade GVP 030 combo card

[ModMumpitz]

Hi,

Does anybody knows how to upgrade this card to more then 16MB of Ram?

Thanks

[hyami]

Is it actually possible to upgrade this card beyond 16MB? I hope my card can be repaired, but such an upgrade would be a great bonus.

[SpeedGeek]

Been there... Done that!

Well not exactly, there are some small differences between Combo 030 and G-Force 030. Anyway, here are the details:

Hello guys,

Here is my G-Force 030 with super hacks:

Hack one: 36MB Fast RAM (Theoretical max. is 52MB!)
Thanks to tbtorro for the GVP 16MB PCBs!

Hack two: 54.2 MHz overclocking
The faster accelerator card is the better accelerator card!

Yes, it's all possible!*

*Note: There is a difference between what's possible and what will become available. Such matters are completely out of my hands.

[hyami]

I've read the Amibay thread. Has any progress been made regarding the approval? Would be cool to be able to upgrade these cards.

[SpeedGeek]

AFAIK thebajaguy has been authorized to make a G-Force 030 memory upgrade kit available but the details of this are still WIP. You should wait for his announcement on the details.

[alexh]

Another great project.

I see that tbtorro has stopped selling 16MB GVP SIMMs:

https://www.amibay.com/threads/16mb-...dules.2440847/

Is there another source?

[SpeedGeek]

While progress has been delayed on the memory upgrade kit (again) I decided to post some tips on the 40 to 50 MHz upgrade topic.

1) There is a good chance you might be able upgrade from 40 to 50 MHz by replacing the oscillator @ U1 and (later) upgrading to a 50 MHz capable* 68030.

2) First check the PAL located at U36. Look for a checksum on this PAL of 90C6 or 90C4. Either should work fine @ 50 MHz.

3) Change the following jumpers to 50 MHz settings:
CN8 1-2
J4 Closed
J5 Closed
J6 Closed

4) You can test this configuration with the original 40 MHz 030 (for a short time) but the RP package variant was never intended for 50 MHz operation and will probably over-heat after a few minutes.

*If you are satisfied with the test results in step 4, then you will need a 50 MHz capable 68030, the RC package variant in particular. This package variant usually over-clocks very well so it need not be rated for 50 MHz.

Optional clock signal improvement hack:
Change SMD resistor pairs R3/R4 and R38/R39 to 1K Ohm.

DISCLAIMER:
Use at your own risk, no warranty expressed or implied, etc.

[hyami]

why is there delay on the memory upgrade kit? Is it a technical issue or an agreement that needs to be made?

[SpeedGeek]

Originally Posted by hyami

why is there delay on the memory upgrade kit? Is it a technical issue or an agreement that needs to be made?

It was a technical issue. The problem is too many logic revisions for G-Force 030 cards and some undocumented bugs.

There will probably have to be two variants of the memory upgrade kit and limiting each kit to just a certain revision of cards. Hopefully, thebajaguy will have better results with testing the updated kit variant.

[dalek]

Maybe they can be persuaded to just release the code and the community can sort it out.

It's not like they are going to make any significant money off this endeavor - we already have open-source alternatives (TF530, TF534, Matzes' 030TK) and closed-source alternatives (TF536, ACA500, HC533) that most people would rather pour their money in to.

At least releasing the code would give some old cards new life and save them from waste and a legacy that is slowly being forgotten.

[A4000Bear]

Originally Posted by SpeedGeek

While progress has been delayed on the memory upgrade kit (again) I decided to post some tips on the 40 to 50 MHz upgrade topic.

1) There is a good chance you might be able upgrade from 40 to 50 MHz by replacing the oscillator @ U1 and (later) upgrading to a 50 MHz capable* 68030.

2) First check the PAL located at U36. Look for a checksum on this PAL of 90C6 or 90C4. Either should work fine @ 50 MHz.

3) Change the following jumpers to 50 MHz settings:
CN8 1-2
J4 Closed
J5 Closed
J6 Closed

4) You can test this configuration with the original 40 MHz 030 (for a short time) but the RP package variant was never intended for 50 MHz operation and will probably over-heat after a few minutes.

*If you are satisfied with the test results in step 4, then you will need a 50 MHz capable 68030, the RC package variant in particular. This package variant usually over-clocks very well so it need not be rated for 50 MHz.

Optional clock signal improvement hack:
Change SMD resistor pairs R3/R4 and R38/R39 to 1K Ohm.

DISCLAIMER:
Use at your own risk, no warranty expressed or implied, etc.

I have a 40MHz card, with a 50MHz 030 fitted, so I decided to try it at 50 MHz. Unfortunately it was a miserable failure.

At 50 MHz, on power up, the power light is dim then goes bright as normal, but the machine remains dead.
At 44.9MHz, while there are signs of life, the machine will not boot successfully.
At 42MHz, the machine boots and runs OK, but will fail to boot again on reset or power cycle after a few minutes, unless I remove my Picasso II card.
At 40MHz, normal reliable operation.

None of the chips or the card itself ever gets more than mildly warm, even though the problem at 42MHz appears to be heat related.

My U36 has a checksum of 90C4. Jumpers are exactly as advised above. Performing the clock signal resistor hack made no difference at all. My board is a Rev 3.


Any suggestions?

[giantclam]

Have you checked the voltage on power rails?

[A4000Bear]

Originally Posted by giantclam

Have you checked the voltage on power rails?

Yes, I have 5.05V on the 5V rail, and 12.1V on the 12V rail.

[SpeedGeek]

Originally Posted by A4000Bear

I have a 40MHz card, with a 50MHz 030 fitted, so I decided to try it at 50 MHz. Unfortunately it was a miserable failure.

At 50 MHz, on power up, the power light is dim then goes bright as normal, but the machine remains dead.
At 44.9MHz, while there are signs of life, the machine will not boot successfully.
At 42MHz, the machine boots and runs OK, but will fail to boot again on reset or power cycle after a few minutes, unless I remove my Picasso II card.
At 40MHz, normal reliable operation.

None of the chips or the card itself ever gets more than mildly warm, even though the problem at 42MHz appears to be heat related.

My U36 has a checksum of 90C4. Jumpers are exactly as advised above. Performing the clock signal resistor hack made no difference at all. My board is a Rev 3.


Any suggestions?

You can't use the G-Force 030 card with the 8MB Auto-Config Fast RAM option and a Picasso II. The Picasso II needs 2MB of the Auto-Config Fast RAM space.

First try removing the Fast RAM. If it solves the problem then you may need to find components with better tolerances. According to thebajaguy some GVP cards have 2K vs. 4K refresh DRAM timing problems. This is one of the "Undocumented" bugs which may or may not cause any problems.

Next, try removing the FPU. I've had mostly good luck with overclocking the 68882 but there have been a few cases where it fails (e.g. "Your mileage will vary"). Note: The FPU must run synchronously with the CPU clock at 40-50 MHz clock speeds.

[A4000Bear]

Originally Posted by SpeedGeek

You can't use the G-Force 030 card with the 8MB Auto-Config Fast RAM option and a Picasso II. The Picasso II needs 2MB of the Auto-Config Fast RAM space.

First try removing the Fast RAM. If it solves the problem then you may need to find components with better tolerances. According to thebajaguy some GVP cards have 2K vs. 4K refresh DRAM timing problems. This is one of the "Undocumented" bugs which may or may not cause any problems.

Next, try removing the FPU. I've had mostly good luck with overclocking the 68882 but there have been a few cases where it fails (e.g. "Your mileage will vary"). Note: The FPU must run synchronously with the CPU clock at 40-50 MHz clock speeds.

I always have the RAM set up as 'Extended' (I also have a bridgeboard as well as the Picasso II). I removed all RAM from the G-Force (except for its 4 MB built in RAM). Also removed the FPU. After doing this it actually booted at 50MHz, however it didn't recognise the 4MB built-in RAM on the G-Force, and as a result, it runs out of memory before Workbench fully loads as there is now only 2MB of Chip RAM only.

After the machine has been on for a minute or two, it will fail to reboot, I get a blinking power light and no other signs of life. Nothing on the G-force gets hot.

I should mention I had been running the G-force at 40MHz for many years, but with the FPU clocked separately at 54MHz and it has been totally reliable.

[SpeedGeek]

Originally Posted by A4000Bear

I always have the RAM set up as 'Extended' (I also have a bridgeboard as well as the Picasso II). I removed all RAM from the G-Force (except for its 4 MB built in RAM). Also removed the FPU. After doing this it actually booted at 50MHz, however it didn't recognise the 4MB built-in RAM on the G-Force, and as a result, it runs out of memory before Workbench fully loads as there is now only 2MB of Chip RAM only.

After the machine has been on for a minute or two, it will fail to reboot, I get a blinking power light and no other signs of life. Nothing on the G-force gets hot.

I should mention I had been running the G-force at 40MHz for many years, but with the FPU clocked separately at 54MHz and it has been totally reliable.

Well, 40 MHz is probably the upper limit for asynchronous FPU clocking but I still don't recommend it. The DSACK1 skew from U7 is already over the 3-5ns spec. Also, the FPU_CS is marginal and at the limit. Running totally reliably with occasional FPU use is not the same as working the FPU hard under the full set of thermal and voltage loss conditions.

You did not mention if it was the Fast RAM or FPU removed which enabled the boot @ 50 MHz? But there is an opportunity to try some of the lesser overclocking frequencies you previously mentioned.

Other than reviewing the PAL checksum and speed ratings, I have no more suggestions at this time.

[A4000Bear]

Originally Posted by SpeedGeek

Well, 40 MHz is probably the upper limit for asynchronous FPU clocking but I still don't recommend it. The DSACK1 skew from U7 is already over the 3-5ns spec. Also, the FPU_CS is marginal and at the limit. Running totally reliably with occasional FPU use is not the same as working the FPU hard under the full set of thermal and voltage loss conditions.

You did not mention if it was the Fast RAM or FPU removed which enabled the boot @ 50 MHz? But there is an opportunity to try some of the lesser overclocking frequencies you previously mentioned.

Other than reviewing the PAL checksum and speed ratings, I have no more suggestions at this time.

Presence or absence of the FPU did not seem to affect my results, it is a 50MHz one anyway. While I was not aware of the problems clocking it asynchronously, I had been clocking it synchronously in my earlier tests for convenience.

Done some more testing now. All tests are with the FPU and expansion memory removed from the G-Force. All Zorro cards except PicassoII removed. All jumpers set for 50MHz. Power is good, 5.05V on the drive connectors, 4.9V on the pins of the clock oscillators in the G-Force.

50MHz - Will boot when cold, but built in memory on G-Force not recognised. PicassoII not recognised.
48MHz - No boot at all, even when cold. Power LED changes brightness. Hard drive light may flash, followed by the bypass relay on PicassoII sometimes giving a brief buzz or power LED may blink.
44.9MHz - Boots from cold OK, machine runs OK, but will fail on reboot or power cycle (behaves just like at 48MHz). Heat related? Nothing on G-Force card gets hot though.
42MHz - Seems to be perfectly reliable.

Here are the details on the PAL chips on my board:
U34 - 74F4, 5nS
U35 - 5F54, 10ns (this chip made by MMI, all others by AMD)
U36 - 90C4, 7nS.
U37 - 4D7B, 5nS.
U38 - 38CC, 7nS.
U39 - F029, 7nS.
U40 - B520, 7nS.
U53 - A2B1, 7nS.


U7 - 4214, 7nS.
U6 - D794, 20nS. (I included this as none of the pictures of this board on the net seem to have this checksum).....I'm starting to think this may be the problem.

EDIT: Since then, I have been testing with the FPU installed, and clocked synchronously with the CPU. Its presence has not altered the results I was previously getting without the FPU.

Edit 2: Seems I have two problems simultaneously (for some reason that ALWAYS happens to me!). With the hard drive removed and J14 closed, I now get fully reliable operation at 44.9MHz, even with all Zorro cards, RAM and FPU fitted. Still fails to boot at 48 or 50 MHz though.

Now investigating my SCSI setup (I have a Guru ROM v 6.14, a SCSI2SD v6 and a CDROM). Correctly terminated at both ends. Just have to find out why it falls over above 42MHz.

[SpeedGeek]

Originally Posted by A4000Bear

Presence or absence of the FPU did not seem to affect my results, it is a 50MHz one anyway. While I was not aware of the problems clocking it asynchronously, I had been clocking it synchronously in my earlier tests for convenience.

Done some more testing now. All tests are with the FPU and expansion memory removed from the G-Force. All Zorro cards except PicassoII removed. All jumpers set for 50MHz. Power is good, 5.05V on the drive connectors, 4.9V on the pins of the clock oscillators in the G-Force.

50MHz - Will boot when cold, but built in memory on G-Force not recognised. PicassoII not recognised.
48MHz - No boot at all, even when cold. Power LED changes brightness. Hard drive light may flash, followed by the bypass relay on PicassoII sometimes giving a brief buzz or power LED may blink.
44.9MHz - Boots from cold OK, machine runs OK, but will fail on reboot or power cycle (behaves just like at 48MHz). Heat related? Nothing on G-Force card gets hot though.
42MHz - Seems to be perfectly reliable.

Here are the details on the PAL chips on my board:
U34 - 74F4, 5nS
U35 - 5F54, 10ns (this chip made by MMI, all others by AMD)
U36 - 90C4, 7nS.
U37 - 4D7B, 5nS.
U38 - 38CC, 7nS.
U39 - F029, 7nS.
U40 - B520, 7nS.
U53 - A2B1, 7nS.


U7 - 4214, 7nS.
U6 - D794, 20nS. (I included this as none of the pictures of this board on the net seem to have this checksum).....I'm starting to think this may be the problem.

EDIT: Since then, I have been testing with the FPU installed, and clocked synchronously with the CPU. Its presence has not altered the results I was previously getting without the FPU.

Edit 2: Seems I have two problems simultaneously (for some reason that ALWAYS happens to me!). With the hard drive removed and J14 closed, I now get fully reliable operation at 44.9MHz, even with all Zorro cards, RAM and FPU fitted. Still fails to boot at 48 or 50 MHz though.

Now investigating my SCSI setup (I have a Guru ROM v 6.14, a SCSI2SD v6 and a CDROM). Correctly terminated at both ends. Just have to find out why it falls over above 42MHz.

The PAL checksums and speed ratings appear to be consistent with the other 40-50 MHz variants of the G-Force 030 cards. U6 is not likely the cause of the problem (even if the checksum is uncommon).

CN8 is a 3 pin jumper without any pin numbering - is it possible this was set incorrectly?

NOTE: The U53 checksum confirms it has the MAPROM bug (it only works with the first bank of Auto-Config Fast RAM). Don't use the GvpCpuCtrl FASTROM option. Use CPU FASTROM if you want to remap the Kickstart ROM.

[Thomas Richter]

The 40Mhz version and the 50Mhz version are different, so do not expect that this will work. As far as I recall, the 50Mhz version had one additional wait state in the RAM, so it wasn't even always faster than the 40MHz edition anyhow.

Also, without the RAM on board, the card will be seriously limited. Go with the 40Mhz, and be happy that you may have a card that is even faster than the 50MHz edition GVP offered. It only worked by hand-selecting components and due to slower RAM speed.

Not worth the trouble...

[SpeedGeek]

Originally Posted by Thomas Richter

The 40Mhz version and the 50Mhz version are different, so do not expect that this will work. As far as I recall, the 50Mhz version had one additional wait state in the RAM, so it wasn't even always faster than the 40MHz edition anyhow.

Also, without the RAM on board, the card will be seriously limited. Go with the 40Mhz, and be happy that you may have a card that is even faster than the 50MHz edition GVP offered. It only worked by hand-selecting components and due to slower RAM speed.

Not worth the trouble...

<sigh> The extra wait state is determined by jumper CN8. The refresh time interval is determined by jumpers J4-J6. If you check the well known Amiga Hardware websites you will see both the 40 and 50 MHz variants have essentially the same PAL set. Except for a few prototype cards, all GVP accelerator cards were factory produced. The hand-picked components were the CPU, FPU and oscillator.