Ultimate 1 MB Issue with MyDos Ramdisk?

[+Larry]

I put my 800XL with U1MB back in service today and had to change the MyDos ramdisk settings. When I used the MyDos "O" configuration for the ramdisk, I immediately noticed a problem -- the ramdisk setup does not recognize the correct amount of ram in the U1MB with either MyDos 4.50 or 4.55 B4. If I set the Ultimate to 1088 extended ram, then MyDos sees 768K. If I set it to 320K Rambo, then MyDos sees only 192K. If I take a 320K XE with the "normal" hard-wired ram upgrade, it sees the correct amount -- 256K. Or in a Newell 256K 800XL, it sees correctly 192K.

 

In all other respects that I can see, everything is normal. If I use the amount of ram that MyDos reports, the system appears to work properly. If I tell MyDos what should be the correct amount according to the Ultimate menu, then the ramdisk shows garbage when I check the directory.

 

Then I ran XRAM on the system, and it reports same as MyDos -- 192K and 768K, so the issue seems to be in the U1MB. (pics attached)

 

Reflash? I did recently reflash the system a couple of times when I tried the new bios and ultimately reverted to the original, but the flashes using my programmer went normally. (?)

 

Any thoughts as to what might be the matter?

 

-Larry

 

 

 

 

 

 

[+Roydea6]

Larry,

My Ultimate shows 1024k and makes a ramdisk of 8170 sectors. And XRAM21 tests a full 1024 k. So you may have a problem with the Ultimate, but I use partition D9: as my ramdisk and leave the 'O' options setup as ramdisk NONE..

[Marius]

Odd. Forget the whole mydos story. X RAM already shows issues, so something is wrong.

 

Ask candle for support here.

[+Larry]

Thanks for the suggestions, guys. I've got a plan... First I'll re-flash a new rom with the stock (Lotharek) image and see if that cures the problem. Then I'll replace the MMU socket with a precision one (and I'll probably do that anyway on general principles). This is a V2-Candle Ultimate that was upgraded to the current XilinX program. But it did work properly for a long time, so I doubt that is the issue.

 

-Larry

[flashjazzcat]

ROM content won't make any difference here. A quarter of the RAM isn't banking in when selected via PORTB, so I guess some connectivity issue.

[+Larry]

Hi Jon-

 

I don't discount at all what you are saying, but I just switched roms with a couple of spares that I had previously burned successfully (?). The first was the original bios with my OS/rom slot changes. It shows 928K of the 1024 (58 banks). The other was your rom that I burned 0.24? (first version, IIRC), and it also shows 768K. Swapped the roms again with the same results. On my "good" rom, XRAM now shows 256K which I believe is the correct amount for the 320K setting. Swapped the MMU cable and it made no difference -- still 768 and 928K. Pins on all these roms are clean and straight. Socket contacts all look good. Put precision pins on the MMU cable connection (name?). --no difference. Strange!

 

I'm going to keep working on this, but for now I'll keep "Mr. 928" in there!

 

-Larry

[flashjazzcat]

Well the choice of flash chip can make a difference to the behaviour of the board, but I'm saying the content of said ROM (i.e. the BIOS) won't, since the extended RAM emulation is carried out by the FPGA, not the BIOS.

[+Larry]

I'm going to order some more. These are SST 39SF040. Have you had better luck with another? If so, I'll try some. I have seen with some of these that I'm using that I get several failed flashes (Mini-Pro burner) before getting a successful completion. Is that fairly common? I always clean the pins and make sure everything looks good before flashing, but seems not to matter.

 

-Larry

[flashjazzcat]

I had a few boards through my hands which had a special dislike for AMD chips. Symptoms included booting direct to the RAM Self-Test, which sort of speaks for itself.

 

I don't usually have problems with USB programmers (nor UFlash for that matter) if the PLCC pins are straight, but I've had a couple of Amic and AMD chips go south over the years. That said, I prefer to use the in-place flasher where at all possible, because every insertion/removal of the PLCC puts strain on the socket as well as the pins (even when using the correct extraction tool). One thing I found helped when my 1200XL started acting itself during the early PBI BIOS development days (before I wrote a convenient flashing tool) was making a little hook using a component leg or pin bent at right angles, and pulling the spring clips on the PLCC socket out a little so they made firmer contact with the chip legs.

Edited August 17, 2015 by flashjazzcat

[+Roydea6]

Larry,

Have you tried to test with XRAM.COM and a clean SDX boot from the same side cartridge?? If this show the same problem then it is probably an Ultimate problem and not a bad MyDos..

[+Larry]

Hi Roy-

 

Really quite sure it is not MyDos. I'll try with SDX, but XRAM021.COM shows exactly what MyDos shows in every case. In fact, I've thrown MyDos out of the process, except for booting the system and loading XRAM. XRAM is what shows the differing amounts of memory using the different flash roms.

 

BTW, look for a PM about your MyDos directory/file lister! (That is what started this whole process where one thing led to another, to another, etc.)

 

-Larry

[+Larry]

SOLVED! (I think!)

 

I decided to swap Antic. I have seen Antic affect these Ultimates before, but only affecting the display. When I looked at the Antic in there, it did not have the "right" part number. It was CO122960-01. Not the normal (IIRC) CO21697-xx. Note -- might be CO12296D-01?

 

As soon as I changed it, all my flash roms showed the correct amounts of ram.

 

Does anyone know anything about this different Antic? I thought there was one and only one XL/XE Antic. (?) It came out of its socket like it had been in there since day one, and I don't remember changing it, but I certainly can't rule that out.

 

I've had this XL a long time, and it was just fine when in "stock" form.

 

-Larry

[+bob1200xl]

The 12296 provides 7 bits of refresh. The 21697 outputs 8 bits of refresh. A 64K machine only needs 7 bits so the 12296 will run in any XL. When you go to 256K DRAM chips, you have to use 8 bits of refresh.

 

Your problem doesn't look like a refresh problem, anyway. Pull a couple of 12296 ANTICs out of working systems and try them in your 800XL.

 

Bob

[+Larry]

Solved! (Again...)

 

Well, with the "good" Antic installed, my 800XL U1MB started showing up with 768 and 192 KB again after leaving it powered on for several hours solid!

 

So I took my Lotharek U1MB out of a 130XE and put it into the 800XL (the "problem child" is an original V2 from Candle, updated with the newer firmware). Now every time I check, it shows the correct ram with either Antic and any flash rom.

 

Is it likely junk, or is it reasonable that if I re-flash it again, it might be OK? It seems like there is little to lose by trying. The update programming was without incident, so far as I could tell. Hate to toss it, but if that is what needs to happen, that's life on the upgrade circuit!

 

-Larry

[flashjazzcat]

Once again, I'd probably end up reflowing the FPGA before doing anything drastic. Reflashing won't affect CPLD behaviour. See if you can get someone to give the board a detailed "service".

[+Larry]

Sorry Jon, I don't quite follow.

 

Do you mean touching up the XILINX SMT solder connections? Looking at everything with a good magnifying glass, the XILINX connections all look good, but on the two ram? chips on the top side, there may be a minute bridge on two adjacent connections. Looks like I could get rid of that with a pin or needle.

 

-Larry

[flashjazzcat]

Yeah: a bridge on the RAM would do it too. You've inspected the Xilinx, which is exactly the kind of thing I'm suggesting.

[+Larry]

Jon, I think you nailed it!

 

I removed the "bridge" with a pin, but that made no change. But then I reflowed all the traces on the ram chips and now it reports the correct amount of ram. Sure glad I didn't have to touch that XILINX chip -- wow, those are tiny. I found that by resting my soldering hand on the work bench, it makes me steady enough to work on those ram traces. Now, just hope the fix holds!

 

-Larry

[flashjazzcat]

Glad it helped.