FS: 2 x GVP-M TekMagic 2060 boards

[trilobyte]

-D- wrote:
On the software end, I suggest using Thor's mmulib package. The processor library works fantastic with the TekMagic, and then you have a nice set of tools that all work together... you can do away with the supplied software entirely. (For example, I use mufastrom instead of ksremap, and mumaprom for testing kickstart images before burning.)

I've been using the Phase5 software with great results, because I was scared of some possible conflicts with MuLib and Picasso96.  One thing I guess that needs to be looked out for is that some 68060 libraries will enable the FPU/MMU on 68EC060 chips, even if they are not supposed to be enabled (hence the EC designation).  Both of my TekMagic boards have XC68EC060RC60 chips on them, so I'm not sure how reliable the FPU/MMU could be on either.

On the hardware side of things I've had a lot of problems.

First of all, lots of random lockups and freezes not attributable to software. My "BigFoot" PSU was full of cheap quality (and 10 years old) JPCon capacitors, so those were all replaced.

Wish I could do that with my A2000 PSUs too... but I'm not an electronics guy

This helped significantly, but I was still encountering periodic freezes while the WB blanker was running. Same symptoms I encounter on accelerated A600/A1200's when the caps have failed. (Funny side note -- "blanker" seems to be a good hardware test, as my TekMagic would easily fly through hours of CPU intensive benchmarks (like AmigaMark), yet lock after 15 minutes of blanker, LOL.) I replaced all motherboard caps, afterward, no more freezing.

Third hardware related problem. After 30 minutes or so, some WHDLoad games exit back to WB, returning similar errors. Pinball Dreams in particular - just let a table run, and eventually it quits to WB, more frequently as the machine warms, until it exits almost immediately. This was directly related to the +5v rail and the lower FPGA on the TekMagic.

I'm amazed at the lengths you've gone to.  Far beyond anything I've done.  Thanks for sharing.  Sounds like you've definitely narrowed down the problem.  

On the plus side, GVP has offered to replace both FPGAs, though I'll need to send in the card since they require programing on the board. Waiting to hear details, but I'm happy it's able to be fixed.

I'm surprised GVP is still servicing these boards.  That's good news.

Do you have Burst Mode enabled?

I do, and I don't recall a difference with it off, but I haven't tried disabling it recently. I'll give it a shot and let you know if it makes any difference.

Since according to that webpage, the FPGA is what controls burst mode, and you have narrowed down your problems to that FPGA, I wonder still what will happen if you disable burst mode (JR1) altogether.

I found the following, while I was testing out these boards again after starting this thread (though I had sort of promised myself I was done with them
  :roll: )... neither board would allow the system to even POST with JR1 enabled and certain RAM installed.  (One of those boards would boot with JR1 enabled, and certain unmatched SIMMs installed (a pair of them), but it was such an anomaly I don't even consider it a success.)  Matched pairs, marked 60ns, some not marked 60ns... as soon as I took JR1 off, disabling burst mode, both boards would again boot.  

One of the boards (the one I used for many years) could not render SpaceFighters from LW3D 3.5 into a HAM6 anim without hard-locking the machine part way through.  The other board, with infinitely less hours on it, can render the same anim fine, with the same RAM and similar jumper settings (burst mode off, of course).  I say *similar* jumper settings because it wasn't until I had both boards next to each other again that I realized my more-used board has had the "ROM type" jumper set wrong for ... who knows how long.  As the machine seemed to lock when doing *disk access* while saving the ANIM frames out, I wonder if this ROM jumper could have had something to do with that board's problems, though I do doubt it.

Whether I'm going to do more testing to determine if this jumper was causing the locks during LW3D rendering disk saves, I don't know.  I'd rather just sell/trade the card.

Again, please do let me know if you find anything from changing the burst mode jumper...

thanks,
t

[Damion]

I think I've about got it resolved.

The card needs a healthy +5V supply, about +5.2V measured at a molex (translates to around +5.05V at at the accelerator and various ICs around the motherboard). Much lower and it gets flaky (starting here with the lower FPGA). Disabling burst mode *might* help the card to function more reliably with lower voltage. With the Tek disabled, or possibly even different accelerators (tried a Vector '030 here), the computer is stable with less voltage (even slightly below +5V at a molex).

I also suggest finding some 60ns FPM Micron - this works flawlessly here. EDO doesn't work at all (including the same Micron EDO which works fine in other Amiga accelerators), and some FPM RAM seems to work better than others. :/

I'm going to go ahead and cool both FPGA's anyway, and add a Nexus WaveAir to the PCI slot nearest the accelerator.

My guess is that with a good RAM and PSU combination, your cards should work correctly, burst mode included. After that, the motherboard caps might also have to be done if you're still getting random freezes or gurus.

[trilobyte]

Well, you were definitely on to something.  In fact, I thought you had it nailed, but I'm still frustrated.

I tested the power supply I'd been using in the test machine, and it indeed did test at ~4.8V.  I tested another PSU I had, and it tested at around ~5.2V.  Both are low on the +12V lines -- both read at a steady 11.7V.

So I swapped out power supplies, and still was not able to enable JR1 Burst Mode with my two 16MB single-sided SIMMs (which work great otherwise!)

So I thought I would try a couple of other SIMMs, and I did get the machine to boot with 2 or 4 single-sided 4MB SIMMs installed and JR1 burst-mode enabled.  Hooray!  But then I saw that I was getting the dreaded trashed pixels on my Picasso screen*.  Also the machine was not stable.  It crashed for no reason.  Decided that I would rather have 32MB of tested-good RAM, rather than 8MB or 16MB of burst-mode, unproven RAM.

For what it's worth, this TekMagic board will boot with practically any SIMMs installed, so long as I have JR1 turned off.  And it is still very fast.  But as soon as I enable JR1, I get blinky green power light as soon as I turn the machine on.  (That is, except with these exact 4MB SIMMs).

Otherwise this machine is rock solid, and I've seen the results of having this board in my other A2000s as well.  Same stuff.  

So now I wonder why switching out the power supply brought up trashed pixels on P96 screens, when I never saw this with the other power supply in the system

See why I just wanted to get rid of these boards?!  
 :pissed:  :-(   :-P

-t

* usually found from using the wrong (15/16, PC/non-PC) video mode in a Zorro II system ...
 

[Damion]

So now I wonder why switching out the power supply brought up trashed pixels on P96 screens, when I never saw this with the other power supply in the system

The card is simply damn fussy, worse than any other amiga hardware I've owned. Also, the computers are getting older, and things like caps need to be changed (particularly in expanded systems). It might be worth scouting ebay for some good SIMMs... I can give you the part # of the ones I have if you like, they pop up frequently, sometimes NOS.

[terminator4]

@trilobyte
> Both of my TekMagic boards have XC68EC060RC60 chips on them, so I'm not sure how reliable the FPU/MMU could be on either.

Actually, both of these chips you mentioned have no FPU/MMU.
I have some 68060 (XC68060RC50 with FPU/MMU) cpu's NOS (new old stock NOT recycled crap) if you/anyone else is interested in backup 68060 - drop me a pm.

[trilobyte]

terminator4 wrote:
@trilobyte
> Both of my TekMagic boards have XC68EC060RC60 chips on them, so I'm not sure how reliable the FPU/MMU could be on either.

Actually, both of these chips you mentioned have no FPU/MMU.
I have some 68060 (XC68060RC50 with FPU/MMU) cpu's NOS (new old stock NOT recycled crap) if you/anyone else is interested in backup 68060 - drop me a pm.

But for whatever reason, they both show up with FPU/MMU in ShowConfig, and FPU/MMU software all works without problem.

I realize this could be due to 68060.library forcing FPU/MMU to be enabled.  But I have never noticed a problem... I use P96 in MMU mode, CyberPatcher/OxyPatcher, Fusion/ShapeShifter, and Kickshifters, as well as LightWave/ImageFX/TVPaint/Imagine all in FPU mode.

-t

[Damion]

Just wanted to add some more info:

I never fully resolved my issues, although much improved, the computer would still freeze on rare occasion.

Also, I couldn't use my PIV - as soon as I attempted holding LMB and drawing a selection window, and the window touched any other image on the screen (icons, window border, backdrop, etc) the computer would lock solid. I tried every imaginable thing to resolve this problem... swapped out all the chips, tried all manner of software configuration, and couldn't fix it. No problem whatsoever with a CV3D.

Then, I buy AmigaPete's A2000. Everything now works perfect, using the _exact_ same hardware and software combination (PSU included) as before, with the only variable being the motherboard. No more lockups with the PIV, and the machine is finally stable.

The "problematic" motherboard was a Rev 6.3, while the working one is a 6.2. I have no idea if the Rev 6.3 board was damaged at some point in its life, or if (as I've read) the TekMagic is simply picky what motherboard it likes.

[terminator4]

@Damion

Here's my story (June 2009 now), which I'm sure you're familiar with (going back and forth on gmail).  This has been going on for at least half a year (?) with some breaks:
- using 2 tekmagic boards
- one works 99% of the time, the other crashes randomly.  Using the same hard disk on both.  So the only thing swapped is the tekmagic board.
- the one that crashes crashes on the same motherboard (A2000 motherboard is ok).  So while the tekmagic can be picky about a A2000 motherboard, it should like this one, since it works with the 1st tekmagic 060.
Items testted:
- tested the CPUs (68060) on Quickpak and on my other Tekmagic 060.  both work fine.  So CPU is fine.
- the ROMs are also fine.
- the oscillator clock is also fine.
All 3 work fine on the "ok" board with the tested A2000 motherboard.
The main question, what else should I check on this tekmagic board?  I'm testing it on a system that likes the 1st tekmagic060 and on the upgraded rom 3.9.  I may try swapping the RAM but rather not mess with it.

It's very interesting about, however don't think that applies to my board:
"Then, I buy AmigaPete's A2000. Everything now works perfect, using the _exact_ same hardware and software combination (PSU included) as before, with the only variable being the motherboard. No more lockups with the PIV, and the machine is finally stable."

Joy, its going to be fun to watch Amiga users buy some of those Tekmagic 2060 boards off softhut.  I doubt they are being redesigned by GVP.  If they are new, they still may be fussy about proper A2000 motherboards.

[Damion]

@terminator4

Interesting that one card works, while the other is flaky, in the same motherboard. Is the RAM the same between the two cards?

One thing: the majority of my issues (like the WHDLoad faults) were solved when I cooled the lower FPGA on the TekMagic, just a little fan blowing on it was enough. Increasing the +5v reduced the need for cooling the FPGA. Might be something to try on your flaky board.

If I get some time over the weekend, I'm going to drop the +5v down a bit and see if it starts misbehaving with WHDLoad again, or if the motherboard was the sole issue. The thing is fortunately running fantastic now... it's a real nice card when it's working properly!!

[gertsy]

@terminator4
Hi, What are the revisions for your old A2000 Motherboard vs the new one from AmigaPete?
I guess trying the new MB with the old pwr supply isolated your old motherboard as the issue. Be interesting to know the Revs ?