Corcomp 9900 disk controller problems

[+acadiel]

On 4/27/2023 at 6:44 PM, justacruzr2 said:

Here's a question I'm hoping someone can answer.  I am going to be testing those Fujitsu M2251A floppy drives this weekend on my PC.  The connector I need just arrived today that will convert the PC clone twisted floppy cable to the TI standard flat ribbon cable.  Should they fail, I might be able to get 2 TEAC FD55FV drives.  These are 720K drives with 96TPI.  If I understand floppy drives correctly, the capacity and TPI are governed by the controller card.  So if the CC9900 controller card can only do DSDD 360K then the TEAC drive should still work and would only format them for 360K.  I just wouldn't be using it to it's full capacity.  Also the TEAC drive writes its tracks at half the width of the Fujitsu's.  So, if what I read somewhere on the web is correct, that shouldn't be a problem either.  The heads on the TEAC should still be able to read the wider tracks of the floppies that were originally created with the Fujitsu drives.  It just doesn't work in reverse.  The Fujitsu would not be able to read the smaller width tracks.  Is this correct?  I just don't want to buy 2 drives that aren't going to work with the CC9900 Controller card.

See here for the Teac FD55 line to make the 80 track versions act as 360K:  https://www.cpcwiki.eu/forum/amstrad-cpc-hardware/modding-a-teac-fd-55-for-double-step-(80-to-40-track-conversion)/

 

There's a link to a German site in the post here that explains how to get them to do 40 track stepping properly, as a 360K drive would do it.

[justacruzr2]

On 4/27/2023 at 6:21 PM, Ksarul said:

Will the FD55F drives work? Yes, with a really big caveat: the drive controller does not know to double-step the heads, so it only sees the first 40 tracks on each side of the disk. Why is this datum important? the disk will work fine in any 80-track drive with your CorComp, but it will not even be readable in a 40-track drive, as that drive will only see 20 formatted tracks per side. It "may" be possible to set the drive to automatically double-step, but you would have to read the technical manual for the FD55F drive to see if that is a possibility. On a side note, when used in a Geneve9640, your CorComp controller will actually be able to format and use 80-track drives, so you could get the full 720K formatted capacity the drives are capable of.

 

It would also have issues reading the normal 40-track disks, for the same stepping issue identified above. The controller does not know it needs to double-step, as the firmware in the CorComp expects a 40-track drive.

That's good news.  And I was right that once I start copying the old disks from the Fujitsu to the Teac, the Fujitsu will no longer be able to read them.  I will still save those old disks from the Fujitsu assuming I am able to get them going again.  And would you please look at the following post.  You seem to have more knowledge about these things.  And Thanks.

[justacruzr2]

Using that cable to bridge from the PC's twisted floppy cable to the straight TI cable didn't work.  So I don't know whether the floppy drives are the problem or not.  I have tried both drives with both the first release CC9900 controller and the REV A controller.  The results were the same with both.  I did connect a different drive awhile back that I bought off eBay and it did work.  But now that drive doesn't work either.  Anyway, that's why I think the problem is with the drives.  But it did make me wonder if there's something wrong with the cards that's killing the drives although I don't know how that could happen.  Maybe it's just a coincidence.

 

Let's assume for this discussion that the drives are OK and it's the controllers.  When I turn on the PEB, the drive spins up and the LED comes on until I turn the console on.  That's normal.  When I turn the console on, the CC9900 screen is displayed, and the drives shut down after a  moment.  That's normal.  When I press 1 for Disk Manager, the drive comes on and attempts to load it but beeps after it checks drive 1 and 2 and can't find it.  That's normal.  When I press 2 for TI Basic it goes right into TI Basic.  That's normal.  When I press 3 for TI Extended Basic the drive comes on and attempts to find and run a program on disk named LOAD but can't find one so the drive shuts down and returns to the TI Extended Basic ">" prompt.  That's normal.  When I press space bar for color bar I get the color bar.  And if I press the space bar again I get the regular TI menu with "Press 1 for TI Basic" and "Press 2 for TI Extended Basic".  That's normal.  This is just how I remember it when everything worked.

 

So with all that appearing normal, could there still be something wrong with those cards?  Both of these cards were rebuilt.  I replaced everything including the IC's with the exception of the PALs since they're not available.  Would the cards still appear to work even if the PALs were bad?  I know the CC9900 screen is running out of the ROMs.  It was my assumption that if the cards work normally, then the PALs are probably OK.  But maybe not?  And it appears that the PALs are different (the equations in them, not the ICs' themselves) between the first release card and the REV A card.  I tried swapping them once and the cards didn't even come on.

 

And one more thing.  I also checked both ends of the boot between the console and the PEB.  They just contained a couple of 74LS244's and 245's and some resistors.  Nothing looked unusual.  I was thinking that maybe there was something wrong with communication between the console and the PEB.  Not knowing what that spike traveled thru and what acted like a fuse makes me suspect everything.

[RickyDean]

Well good news tonight, I've scoured these threads on the repair issues of the Corcomp, and tonight after viewing @Helocast's Rev A PCB for the Corcomp FDC and @Ksarul's schematic and some photos I'd posted here of the second Corcomp I'd bought, took some photos of and added them here, found that on this particlar the 2nd, FDC, I'd forgotten a couple of bodge wires that Corcomp had added due to missing traces and I had a bad connection on the U12 Pal socket, pulled bottom trace. And a little solder reflow on a few pins on the Controller chip itself. I now have a working Corcomp DSDD controller. Just read some files on my HxD and loaded a file in XB. This project has been on the back burner due to work, but I've been off a month due to a rotator cup issue on my shoulder, so I'm trying to play catch up.

@justacruzr2 did you ever get yours fixed? Did you pull old sockets and replace with new? Take photos of them front and back and replace any missing bodge wires?

Now I've got to see about burning another U12 PAL/GAL for the other, and see what happens, may take a bit. Have a good night, I go to sleep happy tonight!!

[RickyDean]

Well last night and today I discovered that there may be at least a couple of floppy controller chip possibilities for the Corcomp FDC, that use the WD2793 chip. I have tried a Wd2797 chip and was able catalog a disk with it, and the same for a TMS2793NL chip also. That is as far as I've tested so far, but I thought it was cool that I could do that.

[+Ksarul]

3 hours ago, RickyDean said:

Well last night and today I discovered that there may be at least a couple of floppy controller chip possibilities for the Corcomp FDC, that use the WD2793 chip. I have tried a Wd2797 chip and was able catalog a disk with it, and the same for a TMS2793NL chip also. That is as far as I've tested so far, but I thought it was cool that I could do that.

That is very good to know. . .

[RickyDean]

This will be a bit long, but here goes:

 

I have two PBox Corcomp FDC controllers, we'll call them FDC1 my original, I've had since the 90's and FDC2, my newest one, for brevity. I tore down FDC1 several years ago to rebuild it with new parts including tooled sockets. It had an issue seeing a drive, kept getting error 16, after rebuilding, but while I was testing it one day I blew the U2 Pal and seems that I damaged U1 inside. So it was dead. I bought and used FDC2 for some time then decided to take it apart and rebuild it to, taking pains to be sure I got it soldered good, and it didn't work afterward either. Put it aside while I've been working a lot of hours the last year.

 

Fast forward to a couple of weeks ago, I've been off for 6+ weeks with a work related shoulder issue. I'm keeping myself busy working on these other "projects" to include these controllers. I find 2 weeks ago that I had left a couple of factory bodge wires off of FDC2, after looking at photos I took and placed here on Atariage, before initial teardown. It fired up, but I did find one trace I damaged, so I ran a bodge wire and it works again with the original Pal chips in U1 and U2, though it acts a bit flakly, some times having to try to load a file two or three times before it reads it in. Maybe a timing issue with a 9901? 

 

So last week I ordered Gal chips and using @Fritz442's jed files, I tried programming and replacing  the Gal onto U2 of the good FDC2, to see if it worked, but it would not display a screen when powered up. I got with Fritz442  and sent him saved files from my programmed Gal, so he could verify that it was good. He said it was, but we could not get a set of Gal's, programmed with his files, to work on FDC2. I went ahead and tried the Gal's programmed with his files on FDC1, and replaced both U1 and U2 on it, and lo and behold, I got a screen with the Miller Graphics logo. Still have the original problem seeing the drive.

 

So tonight I checked out every chip, replaced those that I could not test from my working FDC2, except for the eproms and the Gal's, still not seeing the drive, but that's not the weird part. I then decided to change the Eproms on the two units, Now both sets are the same, bought new from @atrax27407, the eproms that skip the Blue screen. When I put eproms from FDC2 onto the FDC1 and powered it up, it had no screen showing, with the Gal's in place, but when I placed the Pal's from FDC2 onto it, the screen came up again. I then put the eprom's, originally on FDC1, onto FDC2 with the Pal's in place and it would not show a display again. I put Fritz442's Gal's into the sockets, now remember they would not work before, but with this set of eproms it did display the Miller Graphics logo when powered. But it is not reading a disk now either.

 

So I'm going to place the original parts back into FDC2 to get use out of it. But I've said all this to ask this, what could cause this weird behavior, could my eproms on FDC1, have been damaged from the Pal blowout? I think I should read both sets and compare them for differences? At the same time FDC2 can't use the Gal jeds, due to some issue, either, with the eproms that had been on it, though they have worked for Fritz442 on his equipment. Any ideas, and/or suggestions?

 

Sorry for the long post.

 

[+Atari2600PAL]

The MG roms don’t work with early corcomp controllers, could it be that?

 

(sorry if I’m totally misunderstanding your post above)

[RickyDean]

6 hours ago, Atari2600PAL said:

The MG roms don’t work with early corcomp controllers, could it be that?

 

(sorry if I’m totally misunderstanding your post above)

No, that's not it here. That is probably true on Wd2793 boards versus WD1773 boards. But these two are the pretty much identical, with the WD2793 FDC chip, just different PCB runs of the same. That's why FDC1 had one bodge wire on the back side of the 74LS74, next to the LED light and FDC2 had 2 wires in different locations. I assume to deal with trace issues by the PCB manufacturer.

 

I'm leaning to one of two things, either a mis-burned eprom on FDC2 or the one on FDC1 one was damaged during the initial testing of the FDC1 that had the U2 Pal blown out. Reading both sets on my Minipro, and comparing them to each other and a known good set of eprom files, should narrow that down. I'll be doing that tonight after I get done with my doctors appointment. Somewhere I also have the original eproms for both units, I had just replaced them because I was refreshing the boards with new stuff. But the fact does remain that the eproms on FDC2 work with the Pal chips, but will not work with the Gal chips

[+Ksarul]

One other thought here--you may have a trace somewhere that is making a bad connection (might even be connected, but not well enough to carry sufficient current) or you might have a partial or complete short between pins somewhere on the board that is pulling a signal lower than it should be or combining two signals that generally need to be separate.

 

The intermittent nature of your observed problems tends to point me this way. . .

[RickyDean]

18 minutes ago, Ksarul said:

One other thought here--you may have a trace somewhere that is making a bad connection (might even be connected, but not well enough to carry sufficient current) or you might have a partial or complete short between pins somewhere on the board that is pulling a signal lower than it should be or combining two signals that generally need to be separate.

 

The intermittent nature of your observed problems tends to point me this way. . .

Yes sir that is probably the issue.

 

I just did a comparison between the two eprom sets and they are identical, unknown why one set works with the Gals and the other set doesn't though, that doesn't make sense. Maybe a signal issue difference between them?

 

 

[+Ksarul]

Minor voltage output level differences between the two chip types. One (the one with slightly lower outputs) may be being drawn below the level of efficacy by a partial short, while the one with slightly higher outputs may ave just enough residual power to mostly work.

[humeur]

2 hours ago, Ksarul said:

One other thought here--you may have a trace somewhere that is making a bad connection (might even be connected, but not well enough to carry sufficient current) or you might have a partial or complete short between pins somewhere on the board that is pulling a signal lower than it should be or combining two signals that generally need to be separate.

 

The intermittent nature of your observed problems tends to point me this way. . .

Bad welding

[+OLD CS1]

28 minutes ago, humeur said:

Bad welding

A bad signal trace...?

 

 

[RickyDean]

1 hour ago, humeur said:

Bad welding

I don't disagree, but it is not shouting at me as to where. All the "weld" points look good on the backside and I can't find bridged points on top. gonna have to trace down with a multimeter from point to point. just redownloaded the schematics for the Wd2793 last night. 

Edited October 26, 2023 by RickyDean
spelling

[humeur]

15 hours ago, RickyDean said:

I don't disagree, but it is not shouting at me as to where. All the "weld" points look good on the backside and I can't find bridged points on top. gonna have to trace down with a multimeter from point to point. just redownloaded the schematics for the Wd2793 last night. 

Look with a magnifying glass, and if I understood correctly from time to time it works by tapping on the circuit with the handle of a small screwdriver, if there is a bad solder the vibrations can help to locate these solders , but the problem can also be the oxidation of the supports or the legs of the integrated circuits.

Despite this it will not be possible for me to tell you where the fault comes from, if it is a solder it is a bit of luck to find it, ditto it could be a pad of the printed circuit which is cut right at the edge.

 

Just patience and time

 

[Fritz442]

Just to put the jedec issue to rest, can anyone else here please burn this CC U2 with a 2793 controller to a GAL16V8 to see if anyone else has issues with it.

I have burned it numerous times with no issues on either of my two cards.

Corcomp_DSDD_FDC_U2_w-2793_(DCCU2)_GAL.JED

[+Ksarul]

I'll look to see if any of my 16V8s are handy (I think I may have a 2793 card--I'll have to verify that first) and try to burn one.

[RickyDean]

@Ksarul, running a multimeter between pin and points, I am asking a question here, It seems like in this schematic pin 28 of the 2793 is connected to the pin 31on the 9901. When I check the circuit and look at the under and around both sockets I don't see any points of connection. The pins appear to be just dead ends, is this wrong on the schematic? Thanks

 

[RickyDean]

On the continuity testing, I've finished point to point on all the chips per the schematic, no problems there, except maybe the answer to the post above. Before I go checking for shorted pins again, I have a question on two areas on the back of the FDC. One is the backside of the 74LS74. The original board before I depopulated, to socket it, had pins 2 and 5 tied together with a wire to the +5 line between pin one and the LED light. Also pin one of the WD2793, the enable precomp line was tied to ground as this photo shows.

I've recreated these when I replaced the sockets. I've since pulled the one off the 74ls74, but based on responses here, may reinstall it, if necessary. I had asked the question before here: https://forums.atariage.com/topic/246297-corcomp-9900-disk-controller-problems/?do=findComment&comment=3854864. Should these be tied like they are in the photo, I don't see this in the schematics and my other card doesn't have these two mods, nor do I see them on other photos of similar Corcomps?

 

Edited October 28, 2023 by RickyDean
changed wording.