Any Issues with 70ns fpm memory?

[toddbailey]

Originally Posted by mech

i have been mixing fast page mode and static column in many 3000's for years with dmac2/ramsey4 combo, generally with the ramsey 4 the fpm must be in the first bank.
If you are buying ram from UTSOURCE expect much problems, i have sampled their zips and 1 of ever 8 or so are bad. I have a simcheckII ram tester with zip addon. Dont put it past them being remarked slower rams etc.they are a simple chip broker and certify and test nothing. I tested a batch of ram for one guy that were marked with 1mx4 part numbers and ended up being remarked 256x4's.

i have also noticed one strange behaviour in a 3000t i had. When running a A3640 in the 3000t if you get memory crashes and odd scsi behaviour,the WD proto -04 or -04 (and earlier)can cause this. Upgrading it to a wd-08 or amd33c93a usually fixes it.

It would make more sense to install some compatible ram instead of desoldering a ramsey from a working 4000. If you need the ram tested let me know, i can do that for you for the cost of shipping.
i am in the usa also.

Mech

Thanks, can I drop you a few bucks for the effort?

for the record, I am running OS vers 3.9 on 3.1 roms, A3000 rev 7.5
super buster -11, fat gary - 2, ramsey -4


The vendor I purchased from is named Quest Components out of So Ca.
ever hear of them ? me either, but price was right except until it came time to use the components. This little project is taking far too long, it should be a simple plug and play. With so many variables how do you really determine what's at fault. It it the memory chips, the jumper setting, the version of the ramsey chip, the revision of the board ?

I recently replaced the the wd33c93a-pl rev 00-04 (proto) with a amd part and haven't had (too many) new scsi issues since. Problem is the part is rated for 16, running the system @ 25 mhz may cause the part to error or crash. I lost a partition on a iomega 1gb jaz disk. But I was running ffs so got caught up in that mess, error validating etc. What happened is I tried to abort a lha extract/install and had to reboot to kill the process, so it might be my fault. later I discovered pfs all in one. haven't really tested it much so too early to tell if problem is fixed

I need a rework unit any way to do an upgrade to the a3640. When I turn off cache, the memory behaves, so far but I can't boot from floppy even with running the memory at 5/8 speed I see 8000 - 0004 errors. because I can't turn off cache. This by itself is really dumb cause you loose a lot of cpu performance when you do so.


I do like the idea of testing the memory, maybe I can find a local company that can validate what I have. Since I only have a few days left to return or keep the parts I need to act pretty soon.

[SpeedGeek]

Originally Posted by mech

i have been mixing fast page mode and static column in many 3000's for years with dmac2/ramsey4 combo, generally with the ramsey 4 the fpm must be in the first bank.
If you are buying ram from UTSOURCE expect much problems, i have sampled their zips and 1 of ever 8 or so are bad. I have a simcheckII ram tester with zip addon. Dont put it past them being remarked slower rams etc.they are a simple chip broker and certify and test nothing. I tested a batch of ram for one guy that were marked with 1mx4 part numbers and ended up being remarked 256x4's.

i have also noticed one strange behaviour in a 3000t i had. When running a A3640 in the 3000t if you get memory crashes and odd scsi behaviour,the WD proto -04 or -04 (and earlier)can cause this. Upgrading it to a wd-08 or amd33c93a usually fixes it.

It would make more sense to install some compatible ram instead of desoldering a ramsey from a working 4000. If you need the ram tested let me know, i can do that for you for the cost of shipping.
i am in the usa also.

Mech

I suspect that you have not done much Ramsey 04 FPM testing on 16MHz 030 A3000s (and you probably have not considered that Ramsey 04 might actually perform differently than with the 25MHz A3000s).

The WD -04 SCSI chip can not cause any memory errors. It could cause SCSI transfer errors under certain (but rare very conditions). But since this problem affects floppy disk usage also this rules out any SCSI transfer errors.

Originally Posted by toddbailey

Finally a comment that explains why things don't work the way they should,

the ramsey chip I have is
390541-04
51-ramsey


My a4000 has a -7 version, too bad it's soldered in place
If I can find one, I'll give it a try. Finding parts is a bit problematic.

If not, then I'm stuck with scm memory then ?

If I exchanged the a4000 & a3000 ramsey chips, would the a4000 have memory issues then?
It has 2/16 mb simms but don't know what mode they operate in


I tried amigakit, ebay and versalia, no listings for the ramsey 7, any suggestions where I might find one?
I also found an interesting site amiga auctions but doesn't appear to be fully working,


The a3640 while marked as refv 3.1 has a fair number of 3.0 including u209
I found a retailer on ebay who has upgrades to rev 3.2, but also has a kits that remove unneeded wait states. Only problem with this purchase, I'll also need to buy a smd rework station and learn how to solder using a scope and hot air device. <sigh> it's only money right ?

I have over 100 floppies, some are propriety and non standard. manually updating those that can is not a solution, especially when they boot fine when running the old scm memory type.

The non-standard floppies should not be a problem since they use custom boot code which seldom messes with the cache/burst settings. But if you don't want to edit your standard floppies s-s. You could use boot menu to boot with no s-s, run RamseyConfig NOBURST from the shell and then reboot from your floppies.

It never hurts to run a complete memory test program to make sure all of your memory is good (but I doubt bad memory is the cause of your problems).

Your options seem quite clear to me:

1) Always make sure Burst is disabled before the cache is enabled
2) Overclock your 16MHz mobo to 25MHz so you can use the 25MHz jumper settings (don't worry the SCSI chip uses a separate 14MHz clock)
3) Upgrade to Ramsey to -07 ($F)
4) Upgrade your A3640 to U209 -03 and eliminate any Burst related problems
5) Get rid of your page mode RAM and find some more expensive static column RAM

P.S. You can't expect simple "plug and play" with a A3000. This is essentially a prototype machine with especially few 16MHz models ever made.

[toddbailey]

Originally Posted by SpeedGeek

I suspect that you have not done much Ramsey 04 FPM testing on 16MHz 030 A3000s (and you probably have not considered that Ramsey 04 might actually perform differently than with the 25MHz A3000s).

The WD -04 SCSI chip can not cause any memory errors. It could cause SCSI transfer errors under certain (but rare very conditions). But since this problem affects floppy disk usage also this rules out any SCSI transfer errors.



The non-standard floppies should not be a problem since they use custom boot code which seldom messes with the cache/burst settings. But if you don't want to edit your standard floppies s-s. You could use boot menu to boot with no s-s, run RamseyConfig NOBURST from the shell and then reboot from your floppies.

It never hurts to run a complete memory test program to make sure all of your memory is good (but I doubt bad memory is the cause of your problems).


Oh and yes instead of the rom ships plugged into the system board, they are installed in a separate card that plugs into the system board. Not sure what they were thinking here, replacing 4 sockets for 4 more ? but hey if it works...

Your options seem quite clear to me:

1) Always make sure Burst is disabled before the cache is enabled
2) Overclock your 16MHz mobo to 25MHz so you can use the 25MHz jumper settings (don't worry the SCSI chip uses a separate 14MHz clock)
3) Upgrade to Ramsey to -07 ($F)
4) Upgrade your A3640 to U209 -03 and eliminate any Burst related problems
5) Get rid of your page mode RAM and find some more expensive static column RAM

P.S. You can't expect simple "plug and play" with a A3000. This is essentially a prototype machine with especially few 16MHz models ever made.

The scsi chip was replaced with a 16 mhz AMD part because the version I originally had installed was documented as having lock-up issues with more that one device connected, which is was I was experiencing. I wasn't aware that the scsi chip also affects floppy disks, I was getting format failure at track 0 from time to time, disk to disk copy was very problematic- I figured either the drives are out of alignment or the disks were bad. I currently have 2 iomega jaz drives, a cd-rw + dvd rom drive and a 2.1 hdd on the same wire

proto type, more bad news, I would have thought that with a rev 7.5 is would be "well refined and stable" at least as far as a CBM machine goes. I don't have any problems replacing the 32m with a 50m clock, but with all the other issues I thought I'd hold off a bit, considering the cost of chip and socket probably worth the purchase regardless

[SpeedGeek]

Originally Posted by toddbailey

The scsi chip was replaced with a 16 mhz AMD part because the version I originally had installed was documented as having lock-up issues with more that one device connected, which is was I was experiencing. I wasn't aware that the scsi chip also affects floppy disks, I was getting format failure at track 0 from time to time, disk to disk copy was very problematic- I figured either the drives are out of alignment or the disks were bad. I currently have 2 iomega jaz drives, a cd-rw + dvd rom drive and a 2.1 hdd on the same wire

proto type, more bad news, I would have thought that with a rev 7.5 is would be "well refined and stable" at least as far as a CBM machine goes. I don't have any problems replacing the 32m with a 50m clock, but with all the other issues I thought I'd hold off a bit, considering the cost of chip and socket probably worth the purchase regardless

The SCSI chip should not affect floppy disks. I just explained that since you were having the page mode RAM/Burst problem with floppies also, it ruled out any SCSI transfer errors.

[toddbailey]

Contrary to the post there aren't any issues, there are issues with fpm memory with my particular machine. Since SCM works, I'll add 24 chips of the same type and hopefully mfg. assuming the seller agrees to an exchange. The next enhancement is to see if the cpu can handle 25 mhz, if not we'll try 20 mhz. I got sockets and 50 & 40 mhz crystal Osc. in the mail Next mystery is to find out why the retina video card reports no memory
A lot of mods and improvements made over the last few months, maybe some day I'll actually be able to use the machine for it's desired use.


thanks to all who offered suggestions and work arounds.

[alexh]

Originally Posted by toddbailey

The next enhancement is to see if the cpu can handle 25 mhz, if not we'll try 20 mhz. I got sockets and 50 & 40 mhz crystal Osc. in the mail.

Are you sure you can try 20MHz? Speedgeek will know for sure

[mech]

Originally Posted by toddbailey

The scsi chip was replaced with a 16 mhz AMD part because the version I originally had installed was documented as having lock-up issues with more that one device connected, which is was I was experiencing. I wasn't aware that the scsi chip also affects floppy disks, I was getting format failure at track 0 from time to time, disk to disk copy was very problematic- I figured either the drives are out of alignment or the disks were bad. I currently have 2 iomega jaz drives, a cd-rw + dvd rom drive and a 2.1 hdd on the same wire

proto type, more bad news, I would have thought that with a rev 7.5 is would be "well refined and stable" at least as far as a CBM machine goes. I don't have any problems replacing the 32m with a 50m clock, but with all the other issues I thought I'd hold off a bit, considering the cost of chip and socket probably worth the purchase regardless

I ran a rev. 7 A3000 board and it was perfectly reliable for many many years. I think what Speedgeek meant was the A3000 was the first of the full 32 bit amigas and being first it was a learning experience for them to get all the bugs worked out. It was the first machine with zorro3. the best board imho are the 9/02 and 9/03 ,but i have not had any trouble with 7.x up(with buster 9 or 11 and the updated scsi chip.

You will need to run the machine at 25mhz also,not 20. i have never ran into a 16mhz a3000 that would not clock 25. But your experience could vary.

[toddbailey]

Originally Posted by mech

I ran a rev. 7 A3000 board and it was perfectly reliable for many many years. I think what Speedgeek meant was the A3000 was the first of the full 32 bit amigas and being first it was a learning experience for them to get all the bugs worked out. It was the first machine with zorro3. the best board imho are the 9/02 and 9/03 ,but i have not had any trouble with 7.x up(with buster 9 or 11 and the updated scsi chip.

You will need to run the machine at 25mhz also,not 20. i have never ran into a 16mhz a3000 that would not clock 25. But your experience could vary.

What wait, you can run 25 but not 20 ? I'm guessing it's not so much the cpu rating but timing issues with other chips?. Considering it's a $2 part, if figured it's worth a gamble. I'm using a socket for the mod so a simple replacement after I'm done.
25 from 16 will be a nice speedy improvement, if it works that is.

I have a stick-on cpu heatsink that I can use if needed.

So if I run the system at 25,
will my fpm memory issues go away,
my Michron Tech 8m memory card work and
will my retina video card also work?

They worked when I last had the A4000 wired up.

Actually, the fpm memory is no longer an issue, my supplier has provided an rma so I'll be switching to the same that came with the a3000 a few years back.

I don't have any problems replacing the 16 mhz cpu with one rated at a higher speed,
they are like $15 bucks on ebay. I also have a rework station coming in another week, so replacement isn't going to be a big problem,

One thing I'd like to find out is what is the differences between my rev 7.5 and the 9 series, finding any custom chips might be the deal breaker however. - ramsey 4 and dmac espc.

[NMI]

If I remember correctly Dave Haynie explained somewhere that when the A3000 was designed there was only supposed to be a 25MHz model. But very late in the development process management decides there would be a 16MHz version also, and the development team had to hurriedly get a 16MHz working.
And there is no doubt, the A3000 runs better at 25MHz. But that is of cause no guarantee that it will solve all your issues. After all, the hardware is over 20 years old.

But I suggest you try to run your current 16MHz cpu at 25MHz, it is very likely that it will work without any issues.
When Motorola started making the FE (plastic) version of the 030 they had already been making 030's for some time and the absolute majority of the chips worked reliably at 25 or even 33MHz. So it was rather Motorola marking the chips with 16MHz because that was what the customer had ordered than the chips failing the 25MHz test.

[toddbailey]

"..But that is of cause no guarantee that it will solve all your issues. "

whoa wait what, you mean replacing the 32 mhz clock with a 50 mhz one, won't make my teeth whiter, cause my hair to grow back, loose 25 pounds and make me more attractive to the opposite sex...

Damm!!! why am I doing this again?
.
.
Oh I remember now so I have a chance in hell that I can run fpm zip memory and use the latest ver 11 superbuster chip with my add on memory card (optional) and more importantly use the retina video card, and a little faster boot time never hurt.

after a long 8 day journey from Mn, Nd, Wa. and Or. and finally back to Wa, my clock and sockets arrived from digikey. so much for 2 day express delivery from USPS.

ok but wait, hours later, a3000 socket and 50 mhz clock installed and after a jumper change systems boots.
but wait there's more, swapped scm with fpm memory, still boots ok, then the ultimate test, allow cache and burst to start at power on
More success
and if you call now we have this special offer
the system in it's current state boots unaltered floppies,

we are most thanks again to all who shared their deep understanding of the a3000

the 68030 does run too hot for my liking, so I dug through the spare parts bins and found a low profile heat sink with unused thermal tape intact
a bit over size an inch and 3/4 covering the cpu that is apx 1 in, no fan on this one and not needed, the heat sink get warm maybe 90 F but much better that 130 with out

I have to wonder if I'll have the same luck with the a3640 rev3.1,... tomorrow is another day and another project

thanks again