Corcomp 9900 disk controller problems

[+Ksarul]

I believe the CTIUG meeting this month is today. . .

[justacruzr2]

I believe the CTIUG meeting this month is today. . .

 

AAAHHHHHH! Too late for today. But then I guess it's the first Saturday of the month. Anyone know where they are meeting these days?

[justacruzr2]

I will take a couple of pictures of my Micro Expansion boards this weekend(have the box apart till I get it functional again) and I may already have posted a picture of my 9900 PEB card somewhere here on the forum, will search and post here. Found this at mainbyte, same as mine.

 

If this picture of the early CC9900 you posted is yours I can tell you right now that one of the problems is that WD2793A on it. Made the same mistake myself. Didn't realize that the "A" suffix does mean something with this chip. Specifically, this is a variation of the WD2793 for Apple computers and there is something very different about it. Forgot where on the web I found that info but get rid of it and get a WD2793. The WD2793A I mistakenly put on my board blew several ICs and messed up my floppy drives so that they don't read/write at all (but they do spin) and blew one of my floppies. A costly mistake. After I get my cards working right I'll have to figure out what went bad on the floppy drives and fix that too. Fortunately I have another floppy drive that wasn't connected at the time and I'm using that for testing purposes. Also noticed the 2 tantalum caps on the voltage regulator and wondering why they are there? Also, unless you have some unusual floppy drives that require faster seek times, the switches on the 8 position DIP are set incorrectly. They should be :

 

1 - ON

2 - OFF

3 - ON

4 - OFF

5 - ON

6 - OFF

7 - ON

8 - OFF

 

This is the factory setting for nearly all drives (as per the CorComp manual that comes with the card).

 

Edited June 5, 2016 by justacruzr2

[Omega-TI]

Sometimes capacitors are used to prevent spikes on startup as well as cushion for quick draw-downs that can occur when an item starts active use. Not seeing the circuit, it could also be a filter for AC interference which could have an adverse affect on the cards operation.

 

I'm sure someone will come up the right answer, I'm just guessing here.

[+Ksarul]

Oooooohhhh, In Search of Space. . .I have that on vinyl and on CD. . .along with a whole lot of additional Hawlwind/Hawklords/Pinkwind/Nik Turner/Alan Davey/Robert Calvert material to keep me happy. . .good info on the 2793A too. I'll have to go digging there. . .

[justacruzr2]

Sometimes capacitors are used to prevent spikes on startup as well as cushion for quick draw-downs that can occur when an item starts active use. Not seeing the circuit, it could also be a filter for AC interference which could have an adverse affect on the cards operation.

 

I'm sure someone will come up the right answer, I'm just guessing here.

 

I think that's a pretty good guess and makes sense from what I understand about caps. Thanks for the enlightenment. I may consider making that mod once the cards are working right again.

[justacruzr2]

Oooooohhhh, In Search of Space. . .I have that on vinyl and on CD. . .along with a whole lot of additional Hawlwind/Hawklords/Pinkwind/Nik Turner/Alan Davey/Robert Calvert material to keep me happy. . .good info on the 2793A too. I'll have to go digging there. . .

 

Ahhh. A fellow Hawkwinder. That's one of the best album covers I've seen and I frequently use it as my avatar. I don't think I have as much as you though. Only have the first 6 Hawkwind albums and RC's Captain Lockheed and the Starfighters.

 

As for the 2793A....just don't want to see more people make this same mistake. Usually the suffix part on a chip doesn't mean that much but in this case it does.

[+Ksarul]

What is really odd is that none of the datasheets mention this particular craziness. . .so it would only be found as a result of plugging in the wrong one. Note that there are also B, C, PL02, and PL02-4 varieties of the 2793.

 

And on Hawkwind, I've been listening to them since the mid-seventies, so they have their hooks in pretty deep by now.

[justacruzr2]

What is really odd is that none of the datasheets mention this particular craziness. . .so it would only be found as a result of plugging in the wrong one. Note that there are also B, C, PL02, and PL02-4 varieties of the 2793.

 

And on Hawkwind, I've been listening to them since the mid-seventies, so they have their hooks in pretty deep by now.

 

I just did a Google search too but couldn't find exactly what I was looking for. Part of what I remember reading was that the "buffer tank" (whatever that is) operates in a different manner than the standard WD2793. Maybe the voltage is different or is applied to the wrong pin of a chip on the receiving end. Wish I could find that again or remember more. It was about 2 - 3 years ago that I read that.

 

That's about the same time I came across them too.

Edited June 5, 2016 by justacruzr2

[RickyDean]

 

If this picture of the early CC9900 you posted is yours I can tell you right now that one of the problems is that WD2793A on it. Made the same mistake myself. Didn't realize that the "A" suffix does mean something with this chip. Specifically, this is a variation of the WD2793 for Apple computers and there is something very different about it. Forgot where on the web I found that info but get rid of it and get a WD2793. The WD2793A I mistakenly put on my board blew several ICs and messed up my floppy drives so that they don't read/write at all (but they do spin) and blew one of my floppies. A costly mistake. After I get my cards working right I'll have to figure out what went bad on the floppy drives and fix that too. Fortunately I have another floppy drive that wasn't connected at the time and I'm using that for testing purposes. Also noticed the 2 tantalum caps on the voltage regulator and wondering why they are there? Also, unless you have some unusual floppy drives that require faster seek times, the switches on the 8 position DIP are set incorrectly. They should be :

 

1 - ON

2 - OFF

3 - ON

4 - OFF

5 - ON

6 - OFF

7 - ON

8 - OFF

 

This is the factory setting for nearly all drives (as per the CorComp manual that comes with the card).

 

No this picture isn't mine but one I pulled online, but mine is similar so I will have to look at it tonight after work, will let you all know tomorrow.

[justacruzr2]

Googled for W R Moseid but found nothing of use. However, I was looking at the CorComp manual for the CC9900 card and saw that Craig Miller was involved with this project. It may be that he was privvy to some of the details of the PALs. It might be worth contacting him to see what he knows about them...if anything. Other than that, we just have to hope that we can come across a good set that can be copied. It's frustrating that this is holding up putting these cards back to use.

[+Ksarul]

I just had a thought on this: since the PALs are socketed, what happens to the board when there are NO PALs inserted? I ask this because the problem only appears after a minute or two--and that makes me think the PALs may actually be working with the problem somewhere else on the board. If the board still heats up with no PALs, you have a problem somewhere else. . .most likely a hidden short at one of the solder points. That bit is actually helpful, as you can start tracking it by following every single 5V power trace and making doubly sure that no solder blobbed over to a ground trace anywhere. Pay close attention to the pins in your sockets, as it is pretty easy to get too much solder there.

 

BTW, I know the board won't actually DO anything with no PALs--this is purely a power overload test. If it still overheats, try it again with NO chips installed at all. If it still overheats that way, you probably have a shorted resistor, capacitor, or diode somewhere on the board. . .

[justacruzr2]

I just had a thought on this: since the PALs are socketed, what happens to the board when there are NO PALs inserted? I ask this because the problem only appears after a minute or two--and that makes me think the PALs may actually be working with the problem somewhere else on the board. If the board still heats up with no PALs, you have a problem somewhere else. . .most likely a hidden short at one of the solder points. That bit is actually helpful, as you can start tracking it by following every single 5V power trace and making doubly sure that no solder blobbed over to a ground trace anywhere. Pay close attention to the pins in your sockets, as it is pretty easy to get too much solder there.

 

BTW, I know the board won't actually DO anything with no PALs--this is purely a power overload test. If it still overheats, try it again with NO chips installed at all. If it still overheats that way, you probably have a shorted resistor, capacitor, or diode somewhere on the board. . .

 

Already did that and already checked over all the solder joints with a 30x magnifier and already checked all the circuit board traces for breaks. The only thing that overheats is the voltage regulator. No other chip on these boards gets hot. All the IC's have been replaced along with every other electronic component. Essentially these are now new boards except for the PALs. Don't know how much you know about the PALs but I located the datasheet for them and learned quite a bit. These chips are all fusible links. They are not programmed in the conventional sense (ie. assembly or "C" code which is then burned onto them like a ROM or PROM). They are programmed by burning out or leaving intact the fusible link. Each pin has one and 2 are reserved for the programmer to protect the chip from being duplicated (this can be overcome though and not everyone does that anyway). So by programming the fusible links you permanently set the pins either to an active high ("1" - intact) or active low ("0" - burned out) state. Since these cards took a static surge hit it's very likely that whatever fusible links weren't originally burned out are burned out now. It takes only 400ma to program (burn out) a fusible link according to the datasheet. A static surge would have delivered much more than that. CorComps cards (all of them) were notorious for this since they did not provide any buffering on the cards to stop such an event dead in its tracks. One thing I plan to do when I get the extra time is to make a schematic of where every trace from the PAL pins goes to. If one of the 5 people I've already contacted can't come up with good sets of the PALs this may provide some insight into how it was originally programmed and maybe it can get done that way. Oh...BTW...with no chips installed the voltage regulator does not overheat.

Edited June 25, 2016 by justacruzr2

[+Ksarul]

I understand that--which was why I asked that the card be checked exactly the way I asked. It doesn't heat up with no chips installed--that is good. Does it heat up with all chips except for the PALs installed? The problem here (and I am one of the folks Arcadeshopper connected to offline on this) is that we don't have good dumps of the PALs. CorComp, Rave99, and SNUG all blew the lock fuses to make reading their PALs a whole lot of hard work. I have the capability to read a PAL--but I don't have experience reading PALs with blown lock fuses (and I haven't tried to research that problem either, as I haven't had need to do it yet). At some point, I may even try to go through all of the cards I have with PALs on them for which we don't have JEDEC files and try to read them for servicing purposes, but that is a long project for another day. . .and unfortunately, unless you get lucky with a set of CorComp PALs and find a set that isn't locked, you will be stuck with the analyze the circuit and build a new set of equations from scratch method. I wish I had a better answer here--but I don't.

 

The main reason I ask the questions they way I have is that you have stated that ONLY the voltage regulator gets hot. If the problem was a short through one of the PALs, that chip would get VERY hot too, as it would be the path with excess current draw. So far, I see no evidence of that, which is why I fall back to the methodical, check everything slowly and carefully method of troubleshooting. Something is missing. It may BE the PAL--but we haven't truly eliminated everything else yet based on the symptoms you've described, especially since you say the card works for a few minutes (which wouldn't happen if the PALs were completely blown, so we know that they are at least partially working).

 

Troubleshooting this here in a forum thread is good too--as it will definitely help the next person with a similar problem.

[+Schmitzi]

hmm, maybe he can mount/fix one of this "active ceramic piezo coolers" (?) on one chip after the other on the card,

maybe starting with one PAL first, or the VRM, to see on which active cooled chip maybe the card is working some

measurable minutes longer, before it crashes, or so ? Just to identify where it comes from.

(I do not know what I am talking about here !!)

[justacruzr2]

@ Ksarul

 

OK. Wasn't sure what you already knew. It helps if we're both on the same page. When I de-populated the board and started re-populating it one at a time the voltage regulator did start getting hot. I did not have to get every chip back in place before this happened. I replaced starting from the left side of the card to the right and from the top to the bottom on both. I wasn't really paying attention to how many chips I had replaced when the volt reg started getting hot again but I believe it was about 5 or 6 and I believe the PALs had already been put back. I'm going to do this again and pay close attention. Does it matter the order in which I do this? Should I put the PALs back last or first? Also have a new set of the MG ROM chips which I could temporarily swap to confirm that the original CC ROMs aren't the problem. I didn't do this up to this point since the cards do work for a short time and I can load the Disk Manager and catalog a disk or load a file into the Gramulator with no problems. That's about the time the cards shut down. Both behave exactly the same way. One thing that crossed my mind was the PEB power supply. Was it possible that the voltage supply to the card slots also absorbed some of the static charge and fouled something up there too. Unsure if the PEB supplies more than one voltage line to the slots and if there are a couple of different voltages (ie. 12V, 5V) supplied to the slots for cards to use. Because none of the other cards in my box have volt regs that are overheating I discounted this but maybe I'm wrong. Maybe the CC card uses more than one voltage?

One other thing I mentioned before in a previous post was that the early model FDC has some modifications on the solder side of the board that look like CC mods but I have no way to confirm this. But since both cards are acting the same way (and the Rev A card has no mods) I'm thinking they probably aren't the problem.

I agree with everything you have said. There is a piece of this puzzle missing yet and in my enthusiasm to get these working again I'm rushing thru things. So as I said above I will do it again and pay attention to how many chips are back on the board (and which ones) when the volt reg starts getting hot again. Also agree that this is a good place to sort this out. I'm aware that there are also a few other people out there with CC9900 cards that aren't working who could benefit from this discussion. I also have a Rave99 512K card and yea I saw that it had a PAL too. But it does function correctly. But I think Rave provided buffering on this card.

One last thing. One of the chips on both boards is the SN7438. At the time, DigiKey didn't have them and a Google search revealed that the SN74S38 can be substituted for it. I had wondered whether or not this might have something to do with it so I downloaded the TI datasheet for the SN7438 family of chips which includes the SN74S38 and SN74LS38. According to the datasheet there is a difference in the low level output current. The SN7438 puts out 48ma whereas the 74S38 puts out 60ma. My local electronics guy said this shouldn't make a difference but I still wonder. However those chips were among the last ones I put back and the volt regs were already overheating.

Edited June 25, 2016 by justacruzr2

[+Ksarul]

OK, that gets us somewhere--the first things to put back onto the card are the various 74LS chips (and I'd suggest replacing the 74S38 with a 74LS38). I may also have some 7438s around here somewhere, as I have an assortment of 74XX series chips lurking about. For the initial test, you don't want the DSR chips, the PALs, or the TMS9901 installed.

 

Let me know if it heats up when tested to that level. If it does, you have just eliminated all of the components that are NOT installed as potential problems. One other question here: have you done all of this testing with NO drive cables connected? It is asked to again eliminate a possible cause of the short. Also, does your PEB have the tall dividers in the back or the half-height dividers? Either way, it might be a good idea to put a strip of black electrical tape along that part of the card (front and back) to prevent the card from shorting out there. Lastly, is all testing being done from the same slot? Try a different PEB slot and see if the results are the same. . .use a known good slot (like one where one of your working cards is currently). The disk controller card is only using the 8V line in the PEB, with the regulator pushing out a nice stable 5V to the rest of the chips on the card. The RS-232 card is one of the few that also uses the 15V lines (bringing them down to regulated 12V signals).

 

For work like this, having a PEB extender card is also a very good thing, as it lets you check signals and voltages without diving into the center of the box. I do still have a couple of those from the board run I did a couple of years ago. I think I still have a few of the connectors too.

 

One other thing that just came to mind: these are both cards that come with a metal CorComp shell, correct? When testing them without the shell, you have to screw a heat sink onto the hole with the Voltage regulator in it or the voltage regulator will go crazy hot. . .when it is in the shell, the shell acts as a heat sink, so it works fine, as long as the metal sleeves for the screws (note that some are metal and some are nylon) are used on the screw for the voltage regulator. Note that sometimes all of the sleeves are metal, but that some have nylon shims to keep them from contacting the board. These too are important, and belong on all screws except the one for the voltage regulator. Make sure the heat sink doesn't short anything to ground by accident (and use heat sink compound if you have any). Note that there is an area under the regulator that expects that kind of contact--that's the only place it is OK. . .

Edited June 25, 2016 by Ksarul

[+Ksarul]

We can deal with the early model modification later--the key is to get stable 5V on the boards first.

[marc.hull]

Have you tried putting a heatsink on the 5 and 12 regulators ? My CC card has a factory metal clam shell that acts as a heat sink. May clear things up (if you Haven't tried before. If you have then sorry.)

[+Ksarul]

It only uses the 5V regulator. Marc--but you are spot-on reference the heat sink. That was why I said he needs to use one when testing the card without using the clamshells. . .

[justacruzr2]

Yes. I asked him about that and he confirmed your identity. Thanks for taking the time to help.

Anyway.........

At first I did not. In a previous post I mentioned that I had mistakenly replaced the WD2793 with a WD2793A that, when I first turned the system back on to test the card, it blew several support chips and both my floppy drives and the floppy I was using in the drive (hope a disk utility program will get that back). That was the early model card. At the time I did not yet have the Rev A card done. So my next tests were without the drives connected. My thinking at the time was that maybe signals from the floppy drives boards might be causing the problem.

OK. Just checked my PEB and it has the 1/2 height dividers. But I'm going to check for interference with the card anyway. It has the spring loaded tangs that do touch the card but from what I can see these cards have the fat ground trace running along the outer edge of the card and I figured that was just an extra safety measure by giving them a secondary ground source to the PEB.

I have been testing it in slot 7 since I'm putting it in and out so frequently it made it easier to attach the ribbon cable and to also put my finger down the side to feel the volt reg. I actually was going to ask you if using slot 7 instead of 8 could be the problem. Thought maybe there was a reason slot 8 is supposed to be used for the FDC other than because of the ribbon cable. Like maybe there were slightly different lines that run to that connector than slots 2 thru 7. But your suggestion answers that.

As for the volt reg output, I did a simple test where I temporarily soldered 2 wires (30ga strapping wires), one to pos one to neg to see what the voltage was during actual use. I put my volt/ohm meter on them and got a reading of 5 volts. It was steady..

My Rev A card came with the metal clamshell but the early model that I bought from a guy at the Yahoo TI group site came bare. Getting a clamshell for it was one of the things I wanted to do for it after I got it working again. And my Rev A card/clamshell uses the metal screws with the nylon washers. The nylon washers are only used on the 3 screws that are not the volt reg. Actually the washers are used to keep the card even in the clamshell. Only the volt reg screw shoulder is taller. The other 3 are shorter and need the washers to keep the card from bending at the corners when put all back together.

Hope you're not going to kill me for this one but so far I've been doing this without the heatsinks attached since the screw that goes thru them attaches to the clamshell and has no nut. Didn't think that would make much difference over the short haul but now with what you and marc.hull have said I'm going to go out to the garage and find a small screw with a nut and put them back on for these next tests. I do have heat sink compound.

So my next tests will be with the heatsink attached. First I will just see if THAT is the problem (am I bad?). If the problem still persists I will then do as you say above starting with the 74 series chips and no DSRs, PALs or 9901 installed. I will be doing this right after I get offline so I'll let you know later what happened.

Edited June 26, 2016 by justacruzr2

[marc.hull]

When PALs go bad they usually go bad catrostrophically (system freeze up or DC never works.) I don't think the PAL is your issue. I also suspect the drop out is just overheating from the VR being isolated.

 

I would go back to square 1 and verify the cable and your termination.

 

You have 2 cards with the same behavior correct ?

[justacruzr2]

@ Ksarul

 

It seems you were right. When I put the heatsink back on the early model card it works without shutting down and locking up. I had it going for about 10 minutes to be sure. But it's not reading the disk. When I press 1 for Disk Manager it goes out drive by drive looking for it but doesn't find it even though the disk is in the drive and I know the disk (and file) are good because I've loaded it before during these tests. However, I did that using the Rev A card not the early model card. The early model card uses 2 trim pots which I replaced but didn't adjust because I have no idea what voltage they are supposed to be delivering and no idea where to take that reading. There are pins in several locations on the board that look like they are for hooking up a meter but which one and what should I be getting for a reading is unknown. The datasheet on them states their voltage adjustability as being +- 0.01%. On a 5V input that would be 0.0005 volts. Didn't think that was significant enough to be a problem but maybe something on the board is that sensitive. This may have something to do with not being able to see the disk but I don't know. Or it could have something to do with the mods on the back side.

As for the Rev A card it didn't even come on at all...no LED...no CC menu screen. And why it did that I don't know since it was fine just a couple days ago when I tried it last. I'll have to investigate that. But for now it would seem that I will get the same result as the early model card when I use the heatsink for that one too.

Anyway, it looks like I'm making some progress and at least on the early card the PALs are not bad. BTW, do the volt regs have something like a bi-metal strip in them that breaks the connection if they get too hot?

 

Will keep you posted as things progress.

 

@ marc.hull

 

Yes and after reading the above we may have solved the overheating issue which prevented any extensive testing more than a minute. I'm not totally out of the woods yet but getting closer. Thanks for your help.

[+Ksarul]

If you have a spare 9901, you might want to try using that with the card that you have coming to life. . .the 9901 is usually the first suspect in any problem with a CorComp floppy controller.

 

It is good to see progress too!

 

Since you do know how to read out PALs, you might want to do a read of the set you have to see if you get lucky. The lock fuses will probably be blown, but if they aren't, that's one set of PALs preserved for posterity.

Edited June 26, 2016 by Ksarul

[justacruzr2]

If you have a spare 9901, you might want to try using that with the card that you have coming to life. . .the 9901 is usually the first suspect in any problem with a CorComp floppy controller.

 

It is good to see progress too!

 

Since you do know how to read out PALs, you might want to do a read of the set you have to see if you get lucky. The lock fuses will probably be blown, but if they aren't, that's one set of PALs preserved for posterity.

 

I do. I kept all the chips I replaced just in case they weren't bad after all. Even though they were replaced with new ones along with everything else I'm going to check it again anyway along with the DSRs and the WD1773. The WD1773 and 9901 came from China. Maybe they weren't so good after all although I bought 2 9901s from them for both cards. They were the only place that had them. Getting tough to find the specialized chips these days.

 

Believe I will need an eprommer to do that though and I don't have one. Maybe I can find someone who will lend it just for this purpose. Getting a read on the ones from the early card would be a nice thing to accomplish since the early card is more rare (I think) and the PALs are not interchangeable between the 2 versions of the cards. The chip used (PAL12L6) is the same but the PCB traces are different. Both PALs do the same thing on both cards but the traces come to different pins so I think just the order is different.