2001-32n - Static character to random garbage

[dhoelzer]

New Symptom!

Turns out the charactor ROM was also bad. All fresh 2532 ROMs with good images are now installed. There's something flaky that's developed with the video that acts like a vertical drive problem, but I don't think it is. I think it's actually related to a new symptom I've found on the motherboard. This may also explain why it functioned temporarily and now 4 ROMs are bad.

I noticed a sizzling sound with some wisps of magic blue smoke coming from pins 24 and 21 on UD7. My first inclination (admittedly, I didn't actually hook up any test gear, but I didn't want to keep it powered on to try to scope it or anything!) was that C77 must be bad, likely closed rather than open, leading to lots of amps running across these two pins. I promptly replaced it with a new .1 capacitor but the sizzling remains...

Any ideas for a next step? I would typically turn it on and start testing voltages and signals but with the sizzling I'm really reluctant to do so.

Thanks!

 

[giobbi]

...are you sure your transformer is ok? You could unplug the power connector and check voltages.

I also read an article (http://www.8bit-homecomputermuseum.at/repair/bluepet/bluepet.html) about a wrong connection on the transformer somebody did; are you sure nobody in the past messed the transformer lines and/or its plug?

Indeed, 4 burnt ROM is quite strange! Did you check the voltage regulators with a multimeter to be sure they aren't short-circuited ? Should be easy to check and you don't need to power the board.... just a small check while you're waiting a guru's answer

More: are you *completely* sure there isn't any short-circuit under the sockets you mounted? In my (little) experience it could happen quite easily with 3032 boards, since the traces that pass between two pins under the socket aren't masked with green paint: if an excess of solder flushes through the hole, it can connect the socket pin and the adjacent trace together.
The test you should do is not only to check two adjacent pins (i.e. 1 and 2, 2 and 3, etc.) but also verify every pin and the adjacent trace. Often this trace lead to the opposite pin; so, assuming i.e. you're checking UD7, a good test would be to check for zero ohm between pin 1 and pin 24 and 23, pin 2 and pin 23 and 22, etc.
Since UD3-UD9 have almost all pins connected together (if I'm right only pin 20 isn't, but however see schematics), it's enough to check one socket. I solved at least one problem I had with this kind of test. I'm not sure a short-circuit can lead to magic smoke, but in every case you must to be sure your new sockets are ok.

If you used high-precision sockets, it would be easy to check: usually the plastic housing doesn't touch the board and permits a visual inspection using a strong light behind the board.

I use to change only one component at time and then I test it to see what changes. This way, in case of trouble I can pinpoint the culprit easily. Replacing all socket at time could take you more time to find a problem, admitting there's any.

 

[dhoelzer]

...are you sure your transformer is ok? You could unplug the power connector and check voltages.

Hmm.. I've always assumed it's correct since it seemed otherwise functional. I'll check it and the regulators.

I also read an article (http://www.8bit-homecomputermuseum.at/repair/bluepet/bluepet.html) about a wrong connection on the transformer somebody did; are you sure nobody in the past messed the transformer lines and/or its plug?

I think I'm safe on this account. I personally scraped the mummified mouse remains off of the transformer; they had been there many years.

Indeed, 4 burnt ROM is quite strange! Did you check the voltage regulators with a multimeter to be sure they aren't short-circuited ? Should be easy to check and you don't need to power the board.... just a small check while you're waiting a guru's answer

Well, I'm not thinking that they were "burnt." I'm thinking that they were 30 years old and went bad. Two of them were only partially bad, though it is interesting that the same range of addresses were bad (0 - 0x800).

More: are you *completely* sure there isn't any short-circuit under the sockets you mounted? In my (little) experience it could happen quite easily with 3032 boards, since the traces that pass between two pins under the socket aren't masked with green paint: if an excess of solder flushes through the hole, it can connect the socket pin and the adjacent trace together.

Yes. I always test as I solder.

I use to change only one component at time and then I test it to see what changes. This way, in case of trouble I can pinpoint the culprit easily. Replacing all socket at time could take you more time to find a problem, admitting there's any.

I usually operate in the same way but the condition of the system has not decayed from the time I replaced the sockets. Something went bad back right after I had it fully functioning and it remains in that same poor condition at this point.

Thanks for your suggestions!

 

[MikeS]

Well, sizzling and smoke coming from the Vcc pins of one of the ROMs would certainly suggest a short to ground, either in the EPROM itself or the socket and related traces.

Obvious, I know, but make sure the EPROM is not inserted backwards; we've all made that mistake more than once.

Measure resistance to ground (pin12), both with and without the EPROM inserted (with power off, of course).

I don't see how it could be a capacitor in this case, but where is C77 on the schematic?

 

[giobbi]

Obvious, I know, but make sure the EPROM is not inserted backwards; we've all made that mistake more than once.

I'm member of Guinness World of Record for that ;-)

In my (huge!) experience about ROM and EPROMs inserted backwards, the only thing I've seen is just a flash from the EPROM window (at least the 2716, I'm almost sure the 2732 with the adapter survived, at least for some seconds); I didn't fried ROMs (yet) for that reason; they seems to be harder to die than Bruce Willis, even with all my diligence putting them in the wrong way.

But... what's smoking, exactly? ROM or EPROM? Did you replace all the ROM with EPROM, or just some of them?

 

[dhoelzer]

Well, sizzling and smoke coming from the Vcc pins of one of the ROMs would certainly suggest a short to ground, either in the EPROM itself or the socket and related traces.

Obvious, I know, but make sure the EPROM is not inserted backwards; we've all made that mistake more than once.

Measure resistance to ground (pin12), both with and without the EPROM inserted (with power off, of course).

I don't see how it could be a capacitor in this case, but where is C77 on the schematic?

Wonderfully specific! Thanks Mike. I'll do what you recommend.

C77 on the schematic.. If you place the board horizontal with Tape 2 at the top, other connectors at the right and the expansion bus at the bottom then C77 is immediately to the left and slightly above the UD7 socket. I assumed (possibly incorrectly?) that either pin 21 or 24 is ground, but those two are tied together with a trace on the bottom of the board. The sizzling is coming from both of these pins on the top of the board beneath the socket.

Assuming they were connected to ground through C77 I thought C77 might be shorted (apparently not).

I'll likely have energy to do some testing tomorrow. To be clear (I'll post an answer to Giobbi about this too), the sizzling is coming from the pins themselves beneath the socket, not from the chip.

 

[dhoelzer]

But... what's smoking, exactly? ROM or EPROM? Did you replace all the ROM with EPROM, or just some of them?

All of the ROM have been replaced at this point. The original UD7 was completely gone (all FF), the character ROM was gone (same) and the others were as described (UD6 and 9 dead from 0x000 through 0x800, the rest intact).

The sizzling/smoking is coming from the pins on the top of the board beneath the socket, not from the eprom itself (thank goodness). The eprom continues to verify correctly and has not been corrupted.

 

[dhoelzer]

Hmm.. Hang on..

Hmm.. Hang on..

Ok. I finally took the time to look at the pinout on the chips (I should have done that a while back) and see that 21 & 24 are tied together to Vcc (as MikeS points out). I suppose these caps are for conditioning in that case... Verified that there are no bad joints and no incorrect continuity between pins between all of the sockets...

I have to research something else though.. Apparently one of the pins on UD6 is tied to the trace immediately to the left of it.. I have to verify that this should be the case.. It seems strange that it's the only one tied to this rail.

 

[dhoelzer]

Alrighty. I've researched the board a bit better and compared it to the logic diagrams. The trace next to UD6 is a Vcc rail. The trace to which UD7 is connected via the .1 cap is a Vss rail. No continuity between the two rails. No cold solder joints on any of the pins on any of the EPROM sockets. All of the Vcc and Vss pins have continuity as expected. All of the appropriate address and data lines match up on all sockets and there is no continuity between any of them.

I'm back to, "What do I try now?" I'm going to power it up in the morning and see what's coming out of the voltage regulators. Perhaps they're pumping out a bit more than expected.

Ok. I finally took the time to look at the pinout on the chips (I should have done that a while back) and see that 21 & 24 are tied together to Vcc (as MikeS points out). I suppose these caps are for conditioning in that case... Verified that there are no bad joints and no incorrect continuity between pins between all of the sockets...

I have to research something else though.. Apparently one of the pins on UD6 is tied to the trace immediately to the left of it.. I have to verify that this should be the case.. It seems strange that it's the only one tied to this rail.

 

[dhoelzer]

Woo! Last post tonight, I promise.

I feel confident it's a bad cap. I'm thinking it's one of the electrolytics but I don't know of any way to test them without removing them from the board. Any advice?

Here's what brings me to this conclusion:

Powered it up. The voltage to the logic chips is slightly below +5 (4.98) on all +5 rails. Looks good so far.

While powered up, CONTINUITY BETWEEN THE Vss and Vcc rails in the UD column!

Powered off I show a low (20 or so ohm) and very slowly rising resistance (acts like a capacitor, but very slow). Under suspicion are C14 and C16 since I can get them both to exhibit this behavior across their leads, but they may be more closely tied together causing the behavior of one to exhibit on the other.

Advice? Remove and replace them both? If I hook them up to my scope I would expect to see a spike that rapidly changes, correct? A slowly sloping line would be bad, if I remember right (but it was a very long time ago that I took that class in college. ).

Alrighty. I've researched the board a bit better and compared it to the logic diagrams. The trace next to UD6 is a Vcc rail. The trace to which UD7 is connected via the .1 cap is a Vss rail. No continuity between the two rails. No cold solder joints on any of the pins on any of the EPROM sockets. All of the Vcc and Vss pins have continuity as expected. All of the appropriate address and data lines match up on all sockets and there is no continuity between any of them.

I'm back to, "What do I try now?" I'm going to power it up in the morning and see what's coming out of the voltage regulators. Perhaps they're pumping out a bit more than expected.

 

[dhoelzer]

Alrighty. Another update.

I felt pretty sure that this looked like a cap issue. I've progressively removed every cap that bridges these two rails and I still have exactly the same behavior. The only thing that I can still see that could be connecting these two rails is the 75154 address selector at the bottom of the row.

Any suggestions on testing this in place? (I suspect you'll say, replace all of the caps and hook a scope up to it and see what the pins are doing.)

 

[dhoelzer]

Removal of a few other caps around the board is showing several that are way off. .1u coming in at .05 and lower, for example. I think I'm going to do a clean sweep of the caps and check again. I'd really rather not desolder the 75154 if I can avoid it. It appears to be hard to find in that particular package these days. Everyone's got the skinny package but no one seems to have the fat package.

Alrighty. Another update.

I felt pretty sure that this looked like a cap issue. I've progressively removed every cap that bridges these two rails and I still have exactly the same behavior. The only thing that I can still see that could be connecting these two rails is the 75154 address selector at the bottom of the row.

Any suggestions on testing this in place? (I suspect you'll say, replace all of the caps and hook a scope up to it and see what the pins are doing.)

 

[giobbi]

I'd really rather not desolder the 75154 if I can avoid it. It appears to be hard to find in that particular package these days. Everyone's got the skinny package but no one seems to have the fat package.

IMO eBay is the best place to find old ICs, etc. If you can't find on eBay, rarely you find them elsewhere.

http://www.ebay.it/itm/74154PC-4-16...432?pt=LH_DefaultDomain_0&hash=item3a6d1c54d0

Many years ago I decided to buy replacements for every IC of every C= I own (at least the hardest to find). You can still find SRAM,DRAM, etc. but not far in the future sources will dry and you will have to count on your stock only...

--Giovi

 

[MikeS]

Removal of a few other caps around the board is showing several that are way off. .1u coming in at .05 and lower, for example. I think I'm going to do a clean sweep of the caps and check again. I'd really rather not desolder the 75154 if I can avoid it. It appears to be hard to find in that particular package these days. Everyone's got the skinny package but no one seems to have the fat package.

I still don't see how a bad cap could possibly cause smoke and sizzling at the Vcc pins of an IC but since you're determined I thought I'd leave it with ya and stay out of it. You might want to check your part numbers though; I've never seen a 74154 in a "skinny" package, but you've now twice referred to a 75154, a different beast altogether that isn't in any PET I've seen...

 

[dhoelzer]

I still don't see how a bad cap could possibly cause smoke and sizzling at the Vcc pins of an IC but since you're determined I thought I'd leave it with ya and stay out of it. You might want to check your part numbers though; I've never seen a 74154 in a "skinny" package, but you've now twice referred to a 75154, a different beast altogether that isn't in any PET I've seen...

I do agree with you Mike, but since there's clearly something (likely a capacitor) bridging the Vss and Vcc pins and these pins under this chip are the only link between the UD row and the rail it seems at least plausible that this nearly dead capacitor is allowing a short to exist that gets exhibited here.

The 75154 was a typo. 74154. If you search for it on Jameco or Mouser you'll find the narrow package. The docs mention that it's available in the other package as well but neither seems to carry it.

 

[MikeS]

The 75154 was a typo. 74154. If you search for it on Jameco or Mouser you'll find the narrow package. The docs mention that it's available in the other package as well but neither seems to carry it.

Well, AFAIK an SN74154N is in a .600 PDIP24 package just like the one in the PET, but of course I could be wrong; I'd actually be surprised if you found it in any other package, but strange things happen...

 

[dhoelzer]

Hmm.. ok, thanks for that info about the size! I went back and had another look. The jameco site only uses a stock picture which makes it look like a 0.25" package. Even the data sheet makes it look that way on the first few pages, but you are 100% correct; The dimensions listed on the mechanical data after the addendum for the datasheet specify 0.598"!

Well, AFAIK an SN74154N is in a .600 PDIP24 package just like the one in the PET, but of course I could be wrong; I'd actually be surprised if you found it in any other package, but strange things happen...

 

[dhoelzer]

While I'm waiting for my capacitance tester to arrive, would someone who has a 2001-XXN (non-CRTC) mind checking what the resistance is between the Vss and Vcc rails that power/ground the UD set of chips while the system is powered off?

Thanks!

 

[dave_m]

An important tidbit is that I picked up a 2031LP that I'm planning to connect to this beastie, so I suspect that the original images will not do me. If I remember right, there wasn't built in support for drives until something post 2.0?

Your BASIC 2 ROM set will support floppies. However with BASIC 4 you get a few extra perks like DIRECTORY and DLOAD. With BASIC 2 to see your directory you have to do a LOAD "$", 8 which loads the directory information into RAM and steps on anything residing there. You view the information with a LIST command. Where DIRECTORY just lists to the screen without disturbing RAM. Most people with the old BASIC ROMs use the DOS Support 'wedge' program supplied by Commodore which stays resident in high RAM and provides for some shortcuts like >$ to list the directory.

 

[dhoelzer]

Oh, super! I couldn't remember what it was but I knew that the 4.0 ROM added support. I couldn't recall if that meant that the 2.0 ROMs did not support the disk commands that are common to later commodores.

Your BASIC 2 ROM set will support floppies. However with BASIC 4 you get a few extra perks like DIRECTORY and DLOAD. With BASIC 2 to see your directory you have to do a LOAD "$", 8 which loads the directory information into RAM and steps on anything residing there. You view the information with a LIST command. Where DIRECTORY just lists to the screen without disturbing RAM. Most people with the old BASIC ROMs use the DOS Support 'wedge' program supplied by Commodore which stays resident in high RAM and provides for some shortcuts like >$ to list the directory.