A3000 Intermittent Heat Related Problems?

[Brosol]

I personally run 2-32GB kingston pro cards off my UWscsi on the cyberstorm ppc with Acard adapters to cf adapters wired for proper DMA. its been working 100% for 2 years+ now.

I only run SFS 1.279 on my stuff. read the full archive docs included 1.277.. it suggests max transfers for scsi and ide,and if the full dma scsi values dont work,drop down as suggested.

I'm not having much luck playing with SFS 1.279.  I set things up via WinUAE and test them on the A3000. I tried suggested max transfers and mask values but still getting boot problems:

Mask:0xFFFFFFFF
Max Transfer: 0x7FFFFFFF,  0x1FFFE, 0xFFFE

What SFS values should I try for A3000 SCSI, SCSI-IDE bridge & CF combo?

[Castellen]

Wow if not even you found a solution for the poor air circulation and heat build up....then there's no hope?

Actually I never really looked for a solution.  Suspect it would probably be better with decent air circulation, etc, but that involves cutting holes and mounting fans.  There's probably other solutions, but I lacked time to try things.  Plus I know the Cyberstorm mk2 is running a bit hotter than it should be.  As mentioned, it's only the backup computer, so doesn't get a lot of attention.

Seems you may be introducing a host of unknowns with many IDE adaptors and the likes.  If you're able to, it would be worth returning the A3000 to a fairly simple, basic, state.  Including a real SCSI disk with FFS.  It's slow, but it does work reliably enough to prove a point.

These kinds of problems can be a real pig to track down at times.  So my suggestion would be to start with as simple system configuration as possible then work your way up.

Another stupid idea; if you have a large chest freezer, run the A3000 in there to prove a point with the heating theory.  The inside case temperature will end up 30-40 degrees C less than what it currently is when operating at ambient room temperature.  Most disks and other consumer electronics are only rated down to 0 degC, so if you put it in the freezer and get it running within 20 minutes or so, the heat generated will keep things above the minimum operating temperature.

As an example, I'm currently soak testing a HSDSL modem outside on top of a mountain here in Antarctica.  Even though ambient is -20C, I can remotely see the modem's internal temperature is +24C.  It's been working happily all week so far.

[Brosol]

Hi,
   Assuming thats good, reseating chips-pushing on them is useless, pull them and reseat them(i doubt this is the trouble though) but we have 20 year old contacts here!Use a proper plcc puller, on those old sockets!. The pals and stuff on the side of the board you mention run quite hot normally,but usually run along happily that way.

I don't think heat is your problem but may aggravate the situation if you have a borederline component.  Bad ram may very well be the trouble,and hot ram chips can be a sign of bad ram.This could cause most of the problems you mention and have effects system wide.
Mike

You were right about the ram being the culprit but so was the temperature factor.

I stripped down the A3000 to most basic running setup and starting reseating all non-PLCC chips (I don't have a puller yet).  Nothing seem to work until after reseating the zip ram....the system was stable again I was able to boot, no more R/W errors or random crashes...etc.  My problem might be unique to the A3000 since it uses Zip RAM which are prone to work themselves out of their sockets due to cycles of cooling & heating.  I found this link by Robert Davis which explains the cause and the fix.

http://www.nyx.net/~rdavis/AmigaHints2.html#ZIPRAM

The Deneb is working again with unexpected benefit which may be of interest to A3000 Deneb users.  Even though I have Buster-09 and Deneb Flash ROM upgrade to support Buster-09 I was never able to use DMA mode without resulting in a ustable sytem.  I tried using Deneb in DMA mode after the Zip RAM reseating fix and now DMA works!  The system seems stable in Deneb DMA mode & I able to large filesize R/W without stability problems.  Perhaps other weird A3000 Deneb related problems might be fixed by reseating all socketed components, especially the Zip RAM.

Thank you everyone for helping!

[Brosol]

Try adjusting the flicker fixer potentiometer and see if it clears up,these variable resistors do degrade over time.

mike

That fixed it, Thanks  

[mechy]

I'm not having much luck playing with SFS 1.279.  I set things up via WinUAE and test them on the A3000. I tried suggested max transfers and mask values but still getting boot problems:

Mask:0xFFFFFFFF
Max Transfer: 0x7FFFFFFF,  0x1FFFE, 0xFFFE

What SFS values should I try for A3000 SCSI, SCSI-IDE bridge & CF combo?

It sounds like to me you have covered all the mask/max transfer values.is there any reason to suspect the new GAL's?

any news on this btw? find a fix yet?

[mechy]

You were right about the ram being the culprit but so was the temperature factor.

I stripped down the A3000 to most basic running setup and starting reseating all non-PLCC chips (I don't have a puller yet).  Nothing seem to work until after reseating the zip ram....the system was stable again I was able to boot, no more R/W errors or random crashes...etc.  My problem might be unique to the A3000 since it uses Zip RAM which are prone to work themselves out of their sockets due to cycles of cooling & heating.  I found this link by Robert Davis which explains the cause and the fix.

http://www.nyx.net/~rdavis/AmigaHints2.html#ZIPRAM

The Deneb is working again with unexpected benefit which may be of interest to A3000 Deneb users.  Even though I have Buster-09 and Deneb Flash ROM upgrade to support Buster-09 I was never able to use DMA mode without resulting in a ustable sytem.  I tried using Deneb in DMA mode after the Zip RAM reseating fix and now DMA works!  The system seems stable in Deneb DMA mode & I able to large filesize R/W without stability problems.  Perhaps other weird A3000 Deneb related problems might be fixed by reseating all socketed components, especially the Zip RAM.

Thank you everyone for helping!

Glad to see you have fixed it! Just for some 3000T users out there.. i have fixed 5- a3000T's over the years with crashing/scsi problems and intermittent boot crashes by replacing the soldered in chip ram zips.This is not always the case,but if you have eliminated everything else then it may be a good place to look.I don't know if the ram was flaky from the start,or soldering had a effect on it or commodore got a bad batch of chips. On all 5 machines the chip ram was the same brand(i cant recall the exact part # and cant find my notes,but i think they were toshiba brand 256x4 zips) . this can literally make you pull your hair out chasing the trouble.Also a side note, NONE of the ram testing programs picked up on the ram as bad! I finally figured it out by loading up chipram until it crashed and burned.
I have also discovered quite a few different revisions of ramsey and dmac chips that don't play together nice. so realize if your messing with this stuff, a dmac2 isn't always the same as another dmac2 or ramsey4 as a ramsey 4

Over the years i have bugged castellen on a few things,and anthony has been infinitely helpful.You are a really good amiga resource!

Mike

[save2600]

I tried using Deneb in DMA mode after the Zip RAM reseating fix and now DMA works!  

So the Zip's weren't actually bad themselves... just were no longer making as good of a connection? Hmm... next time I have my A3000 apart, I'll try that. Can only currently use PIO mode in my system and I've got a higher revision Buster. Thanksf or the tip and glad you got your system all squared away  

Kudos to Mike and Anthony too for their expert knowledge!

[Brosol]

So the Zip's weren't actually bad themselves... just were no longer making as good of a connection? Hmm... next time I have my A3000 apart, I'll try that.

I discovered while reinserting the 16mb of zips, it was too easy to accidently bend the legs.  I used tweezers to carefully straighten any bent legs.  Be patient during reseating & examine if all zips legs are properly inserted in their sockets.  Be static safe.

[Selles]

"hot" chips usually means over-voltage. The chips you mentioned do not even get warm in my Amiga 3000.
 
If the battery has leaked at all, then you will have problems with the Amber circuitry, even if you have removed the battery and cleaned the area. When the corrosion starts, it is like a cancer. It will continue to slowly eat away at the traces. This is why it is so important to find an Amiga 3000 motherboard with a battery that has NEVER leaked and remove that battery. Amiga 3000 motherboards like that are hard to find and command very high prices.
 
The Amiga 3000 computer was never really designed for expanding it. That is why it only has a 135 watt power supply. The most popular expansions people added to thier Amiga 2000 computer was Ram, SCSI, and processor boards. So, Commodore included all of that on the Amiga 3000's motherboard. They also included a FF/SD. Commodore did not think that people would need to install anything else, hence the 135 watt power supply. Amiga 3000s run nice and cool if you do not install any boards. The more boards you install, the hotter it runs. The Amiga 4000 is the same way, just not as bad as the 3000.
 
If you want to add a bunch of boards to an Amiga computer, get an Amiga 2000, 3000T or 4000T. These three Amiga models were actually designed to be expandable.

[save2600]

If you want to add a bunch of boards to an Amiga computer, get an Amiga 2000, 3000T or 4000T. These three Amiga models were actually designed to be expandable.

OR purchase a BigFoot replacement power supply or mod your own from an orphaned PC  

Strange post. If a computer was not designed to be expanded, they wouldn't have included a Zorro II or III compatible busses with fancy Busters, let alone an accelerator slot.

Real issue here as you pointed out - is the power supply. Cheapo, weak power supply was a cost cutting measure to be sure. Commodore wasn't trying to dissuade people from upgrading their systems, just short changed some of those that wanted to  upgrade at a later date on the initial purchase was all.

[mechy]

"hot" chips usually means over-voltage. The chips you mentioned do not even get warm in my Amiga 3000.
 
If the battery has leaked at all, then you will have problems with the Amber circuitry, even if you have removed the battery and cleaned the area. When the corrosion starts, it is like a cancer. It will continue to slowly eat away at the traces. This is why it is so important to find an Amiga 3000 motherboard with a battery that has NEVER leaked and remove that battery. Amiga 3000 motherboards like that are hard to find and command very high prices.
 
The Amiga 3000 computer was never really designed for expanding it. That is why it only has a 135 watt power supply. The most popular expansions people added to thier Amiga 2000 computer was Ram, SCSI, and processor boards. So, Commodore included all of that on the Amiga 3000's motherboard. They also included a FF/SD. Commodore did not think that people would need to install anything else, hence the 135 watt power supply. Amiga 3000s run nice and cool if you do not install any boards. The more boards you install, the hotter it runs. The Amiga 4000 is the same way, just not as bad as the 3000.
 
If you want to add a bunch of boards to an Amiga computer, get an Amiga 2000, 3000T or 4000T. These three Amiga models were actually designed to be expandable.

Thats ridiculous. I've owned/fixed about 13 A3000's in the past,and alot of the chips he mentions run quite warm. There are many differences in chip revisions/board revisions and PAL chips that can cause it to be hot or not. I have had boards where the PAL's would just about burn you,but run on happily,and boards where those PAL's were just warm/hot.There are a good few board revisions out there and one size does not fit all when it comes to problems on the A3000.As dave haynie says,many A3000's that were not intended to be sold and were intended for just demo units but were sold because the 3000 was hot(no pun intended) at the time.

The 135watt power supply i find to be quite robust for most things except maybe a toaster flyer.Unlike power supplies today its actually rated to run at 135watts.this is more than enough power to run the motherboard and some cards/harddrive.. The reason we see failures is internal heat inside the case(add a fan or 2 with extra cards) and the power supplies are 20+ years old.Half the PSU's deemed bad and trashed were probably only in need of capacitors.Just because a PSU shows good voltages does not mean its good, leaky capacitors can let dirty power to the 3000 and cause all kinds of fun.

If the A3000 was not intended to be expanded it would of had zorroII/III slots and a video slot.

About the only point i agree with you mostly on is that battery leakage.So many good boards ruined. I will say i have some boards that have had leakage from over 10+ years ago that i cleaned up and fixed  that still run fine today. Boards that get severe corrosion inside the layers are probably hopeless.Lets just hope amigans have the sense to save the chips off them before tossing the pcb.

[Zac67]

Real issue here as you pointed out - is the power supply. Cheapo, weak power supply was a cost cutting measure to be sure.

135W were pretty standard in the day (for a small box at least). Still got a PC-45/III PSU here - admittedly slightly older but it's just 110W.

I'd guess the mainboard with 16 MB RAM is still below 60W, so that'd leave plenty of margin for upgrades.

[...]
Just took a quick measure: A3000/30 + 1 GB HDD + 12 MB ZIP + Merlin Gfx 4 MB + A2065 + Maestro = 97W primary (108W peak, 92W with HDD powered down).

[save2600]

Yeah... I'm probably rolling my perceptions of past C= power supplies here, but damn were some of them junk. A1200's being weaker than the A500's (WTF?!) and both needed extra care paid to 'em as far as a user adding to their system. Then there was the 8-bit line whose non-user serviceable brick that most C64's and all C64c's were shipped with, was not designed to power much more than a cartridge or tape drive. Heck, you're not even supposed to plug one of their REU units into 'em without modding the REU or replacing your power supply. That's what I've read anyway. Have yet to even plug in my 1750 and am kind of scared to since I've read all the horror stories.

Think I'll post a new question instead of littering this thread  

[mechy]

Yeah... I'm probably rolling my perceptions of past C= power supplies here, but damn were some of them junk. A1200's being weaker than the A500's (WTF?!) and both needed extra care paid to 'em as far as a user adding to their system. Then there was the 8-bit line whose non-user serviceable brick that most C64's and all C64c's were shipped with, was not designed to power much more than a cartridge or tape drive. Heck, you're not even supposed to plug one of their REU units into 'em without modding the REU or replacing your power supply. That's what I've read anyway. Have yet to even plug in my 1750 and am kind of scared to since I've read all the horror stories.

Think I'll post a new question instead of littering this thread  

the Wedge amiga power supplies were not good. they were marginal at best,the 500 one the better of the bunch.Its not hard to overload a 500psu tho with some modest expansion.
The c64 power supplies were really bad(the potted bricks). Some early ones were repairable tho. most tried to kill the c64 when they died. The 1570 reu came with a c128 powersupply modded with a c64 plug. it was a power hog.
I don't recommend you try and use the reu with a stock c64 brick.either get a c128 psu and mod it with a c64 din plug or build a suitable supply.