G_P_T-0_1

WTF did this guy use to solder with? An arc welder? Acetylene torch? As for that line that goes to the sound chip... That's the LSB of the data bus, which actually matches the function of the pin. So, my theory, now that I know what it's actually jumpered to is that this guy blew some traces off the board with his soldering technique. So, presumably you'll need to restore this wire, and tying to that point on the board is actually a perfectly fine thing to do. I'm guessing whoever replaced this chip lifted a trace and replaced it with a jumper wire. If they lifted one trace, they may have mucked up others. Pin 12, BTW, is VSS. This should just be pulled to ground. Can you try restoring the pin 17 patch and giving a nice clean ground to pin 12? "WTF did this guy use to solder with? An arc welder? Acetylene torch?" THANK YOU!!! This is exactly the sentiment I have been expressing since I opened this beast. :roll: intvnut, if you ever find yourself in the Seattle, WA area I owe you a beer for all you've helped me out with so far. Figuring I was about to be in for a helluva uphill climb, I stopped by Radio Shack after work for one of the few things they're still actually useful for - some IC sockets. I picked up enough for my DRAM chips and one 40-pin socket for the VDP. The contact resistance on those is 20 mΩ in case you were interested. I think my game plan will now consist of: * Clean the massive amounts of burned flux off from the bottom of the VDP area of the PCB with some rubbing alcohol and a little TLC. * Bridge the jumper wire, then attempt to ground pin 12 properly. * Test the CV to see if it works correctly. * If so, woot! * If not (which is EXTREMELY likely in my mind) then I think I'm going to desolder the VDP and examine the traces in and around the VDP in great depth. * If traces are busted, I'll jumper as needed and then solder in my new 40-pin IC socket and then test the VDP. * If no additional traces are busted, I'll solder in the IC socket and test the VDP. * If the VDP works correctly from now on, woot! * If not, I'll re-examine the traces for damage. * If additional traces show damage, I'll repair and repeat the above couple of steps. * If no additional traces show damage upon subsequent inspection, I will solder in new DRAM chips. * If the new DRAM chips correct the issue, then woot! * If the new DRAM chips fail to resolve the issue, I will obtain a new VDP and install it. * If additional traces show damage, I'll repair and repeat the above couple of steps. * If new VDP exhibits same behavior, I think I'll chuck the thing out the window. Or examine my other options, such as rebuilding on a new PCB, or replacing other components. I'll let you know my results as I go along. Those other points with the fantastic solder job... If I'm not mistaken, they form a route from the RF modulator, through the NTSC mixer circuit back to the VDP. I'll lay odds this guy got hit by lightning. Or rather, the antenna on the TV that this was connected to did. Either that, or one heck of a static electricity snap. Interesting observation. This was received from people who live in a very lightning-prone area of the world - Minnesota, USA. Presumably this CV never left the Midwest. I'd definitely be willing to go with the lightning explanation as the reason all this work had to be done. Let this be a lesson, kids. Always ground your antennas and ALWAYS have surge protection / uninterruptable power supplies for the electronics precious to you! For now though, my head is pounding from my lack of sleep and day of work. Sleep is unfortunately required occasionally. I'll see you cats tomorrow!

What in the...? That would short the LSB of the CPU data bus to one of the read-data lines on the VRAM bus. The TMS9928A is an NMOS device, which means that its outputs consist (conceptually) of a resistor to +5v and a transistor to ground. (Really, the "resistor" is another transistor wired up to be a dummy load, but work with me here.) Tying these two pins together theoretically shouldn't hurt the chip, since they're both just taking turns switching to ground, but it would definitely cause some weird behavior. That the title screen comes up with the bridge, but not without it makes me wonder if the internal pullup resistor is bad on one of the two lines. If you have, say, a 10K resistor laying around (+/- a couple K), do you mind trying this experiment? Disconnect the bridge wire. Take the resistor and connect one end to +5v, and then connect the other alternately to either pin 17 or to pin 27 (or 28--which ever was the other end of the bridge). My hunch is that when connected to one of the two, it'll make the display correct, and for the other it'll do nothing. Even then, though, it seems like an iffy proposition. Most of the time, pins 27 and 28 should just be in a "high-Z" state, being pulled up by the normal float-high tendency of NMOS and TTL. Those two pins are on the read-data bus for the VRAMs, and only would get pulled low during an actual memory read. In fact, the only time it needs to go low is when a 0's getting read from the RAM. In contrast, the CD7 pin should be getting pulled high and low by the CPU and ROMs and so on, in addition to the VDP. Even if pin 17's pullup was bad, there should be enough other float-high tendency on that bus that if the VDP's pullup was missing you should still be able to read a '1' reliably on that pin. So, that brings me to a different question: Which pin was it that had the crummy solder job, such that it looked like it was "floating in the hole"? Wonderful... my first thought was that this may have been a botched composite video mod, but it didn't seem like a typical botched-mod job to run the bridge wire from pin 17 UNDER the VDP to reach the pins on the top row of the chip. (Top as in above the text printed on the chip, key on the left side.) The wire is packed in tight enough under there that the whole chip had to have been desoldered at one point to accomodate it... how odd. I think what I'll do this afternoon when I get home from work is to take pictures of my CV mainboard so you can all get a view of what I'm working with. I'll also take photos of the garbled CV title screen and the weird (but clear) characters that display when the bridge is severed. I'll definitely try your experiment, I've got tons of resistors laying around in my various component stashes that seem to infuriate the better half. Even if a correct bridge enables a clear display, do I want to keep this chip? Would the re-routing kill other components on the mainboard of I keep the VDP as is, with a revised bridge? I'm not sure right off which pin it is that had the crummiest solder job - I'll post pics when I get home. Here's my massive update - literally. First off, it turns out that the red wire doesn't bridge to another pin on the VDP, it bridges to a solder spot UNDER the VDP. I have noted where everything is in the included photos. When I alligator clip a 10K resistor to the wire coming off pin 17 on the VDP and test the other end against all the other pins, I get various results including more of my weird characters on about 1/3 of the pins, a black screen on about 5 pins, rapid flashing of video lines for several more pins, and a bright orange screen on one pin. None of the connections to the other pins re-created the CV title screen I get when the red wire is intact, so I decided to have a closer look at the bottom of the board.There's a nice shiny newer solder spot on the bottom of the board directly underneath the VDP. When I touch the other end of the resistored alligator clip to this shinier-than-the-rest solder spot, I get my garbled title screen! It appears that this trace leads to a pin on the SN76489AN chip. (???) Second, even though it was definitely noticeable in real life, the scan I made of the underside of the PCB really exemplifies the crummy (IMO) soldering job done on some of the components. I have taken the liberty of highlighting the wonderful jobs in red boxes in the photos. The pin that looks like it is 'floating' in the hole is pin 12 on the VDP, viewable in the photos. And yes, I definitely to agree that there appears to be moisture damage on the larger traces of the PCB underside, but everything save the VDP appears to function correctly. I can put in a Donkey Kong cart and attach a controller and play it with perfect sound and control. Thoughts? Apologies for the large image size - I wanted as much detail as possible on these images.

What in the...? That would short the LSB of the CPU data bus to one of the read-data lines on the VRAM bus. The TMS9928A is an NMOS device, which means that its outputs consist (conceptually) of a resistor to +5v and a transistor to ground. (Really, the "resistor" is another transistor wired up to be a dummy load, but work with me here.) Tying these two pins together theoretically shouldn't hurt the chip, since they're both just taking turns switching to ground, but it would definitely cause some weird behavior. That the title screen comes up with the bridge, but not without it makes me wonder if the internal pullup resistor is bad on one of the two lines. If you have, say, a 10K resistor laying around (+/- a couple K), do you mind trying this experiment? Disconnect the bridge wire. Take the resistor and connect one end to +5v, and then connect the other alternately to either pin 17 or to pin 27 (or 28--which ever was the other end of the bridge). My hunch is that when connected to one of the two, it'll make the display correct, and for the other it'll do nothing. Even then, though, it seems like an iffy proposition. Most of the time, pins 27 and 28 should just be in a "high-Z" state, being pulled up by the normal float-high tendency of NMOS and TTL. Those two pins are on the read-data bus for the VRAMs, and only would get pulled low during an actual memory read. In fact, the only time it needs to go low is when a 0's getting read from the RAM. In contrast, the CD7 pin should be getting pulled high and low by the CPU and ROMs and so on, in addition to the VDP. Even if pin 17's pullup was bad, there should be enough other float-high tendency on that bus that if the VDP's pullup was missing you should still be able to read a '1' reliably on that pin. So, that brings me to a different question: Which pin was it that had the crummy solder job, such that it looked like it was "floating in the hole"? Wonderful... my first thought was that this may have been a botched composite video mod, but it didn't seem like a typical botched-mod job to run the bridge wire from pin 17 UNDER the VDP to reach the pins on the top row of the chip. (Top as in above the text printed on the chip, key on the left side.) The wire is packed in tight enough under there that the whole chip had to have been desoldered at one point to accomodate it... how odd. I think what I'll do this afternoon when I get home from work is to take pictures of my CV mainboard so you can all get a view of what I'm working with. I'll also take photos of the garbled CV title screen and the weird (but clear) characters that display when the bridge is severed. I'll definitely try your experiment, I've got tons of resistors laying around in my various component stashes that seem to infuriate the better half. Even if a correct bridge enables a clear display, do I want to keep this chip? Would the re-routing kill other components on the mainboard of I keep the VDP as is, with a revised bridge? I'm not sure right off which pin it is that had the crummiest solder job - I'll post pics when I get home. I'd tend to agree with you here as well. It definitely looks like some extensive work was done to try to make the CV output anything but I really wonder what the original cause of damage was that they opted to bridge pins on the VDP to get any output? Thanks for the feedback so far guys. EDIT: I may be buying one of doubledown's VDPs much sooner than I thought I would!

Oh boy... I have a whole new question now for you guys. Before I started soldering out a DRAM chip, I decided that I might as well correct the shoddy soldering job on the VDP before anything else. I popped the heat sink off and found a short length of red wire leading from pin 17 up under the body of the chip to pin 27 or 28 it looks like. It was previously hidden from view by the overlap of the heat sink. The wire-pin 17 has had some horrible soldering done to it on the underside of the board. This lead me to believe that my garbled output may have been due to the wire. After searching for a good hour to find other mentions of the wire and finding none, I decided to experiment with it. Severing the bridge created by the wire seems to clear the video garbling issue up entirely, with the sole exception that it's not the COLECOVISION title screen, it's a black screen with two groups of identical characters - one group being in the upper left corner of the screen, one being almost in the center. The characters look like the number "1" and an odd hash sign sort of like a "#" but with an extra line on the top and bottom. When I reconnect the wire, the garbled COLECOVISION title screen comes back up, corrupt as always. Does anyone else have this red wire bridging two pins of the VDP in their CV? If so, what purpose does it serve? My VDP is model # TMS9928AN if this helps at all.

Wow... I've been gaming almost since birth and emulating systems nearly as long. Yet I'd still never heard of emu-docs.org. All aboard the S.S. Failboat, captain G_P_T-0_1. Still, it'd be nice to have some of those docs searchable and such. Thanks for the link 5-11under!

Wait, do you mean VGA as in 5 signals for R,G,B, H-sync, and V-sync? How do you get all that from the VDP? MAGIC! The VDP's 256 x 192 YUV signals run through an upscaler which accepts and transforms them into a true 640 x 480 VGA output signal useable by any tv/monitor with a standard HD-15 VGA input connection. The display is truly amazingly clear. Really some of my best work. Holy crap, Batman! Three things I want to know: 1) Do you have schematics for this mod? 2) If not, would you be willing to mod a CV? 3) Is the upscaler of your own design, or is it purchasable through a vendor? Because seriously, this looks just gorgeous.

I don't think they will make a noticeable difference. IIRC, the DRAMs were socketed in my TI-99/4A. I know the VDP itself was. I do know from one of the members of the VDP design team that the VDP did really push the DRAM timing specs, so if you do have a flaky DRAM, I imagine the VDP will notice it sooner than anything else. (I was hired into Texas Instruments by a guy who was on the TMS9918A VDP design team. He's the guy that added the bitmap mode to the TMS9918A. It cost 50 transistors, in case you're curious. :-) ) He also mentioned that since it was pushing the process technology of the time (the 10.73MHz was pushing it a tad), it was sensitive to its power supply level. I imagine this explains the recurring "Clean your power switch!" suggestions for CV owners. I just got my 4116s delivered today at work, so I'm going to be working ASAP to solder in some replacements. I think based on this I'll go ahead and socket them. Then at least if a kill a chip during a Mod-Gone-Wrong® in the future, I don't have to resolder them in. I think if I end up having to replace the VDP in the future, I'll seat it in a socket but for now I'll avoid it. That's actually good to know... I'll hold off on the heat sinks for now and use the money to buy my IC sockets. I will measure the surface temperature of the DRAM chips after I get this beast working though, just to make sure. I also never thought about games causing issues with a system's components, but this makes total sense - especially with how sensitive the CV seems to be. I think I'll be doing a preliminary interview with the internals of any new CV cart I buy to ensure my lovely little problem doesn't happen again. On another note, is there anything I can do in the AC adapter to ensure power flow is even and uniform? (e.g. replace aging components with more modern, efficient models?) I know the standard answer is "all power supplies are different, and different CV systems like some power supplies more than others" but there must be some way to make the output a little more even. After all, if we go to these pains to keep our CV systems in working order, why wouldn't we do it for their power supplies? I don't know about you all, but I HATE the "throw it away" culture that seems to be prevalent these days! :x Thanks for all the feedback so far on this guys, I really do appreciate it!

Awesome! I've spot checked it and it looks pretty good! I'll see about scanning those appendices this weekend. I also have this Data Manual here. If I scan that, do you mind OCRing it? Not at all. You scan 'em, I'll OCR 'em.

If you do, would you mind terribly sending me a copy for my own archives? Also, if there's interest in scanning the code examples and the sample font, please let me know. I might get around to it this weekend if there's interest. Not at all. Heck, the main effort is scanning the stuff in to begin with! I'll let you know when I'm done with it. intvnut, here's the OCR'd PDF of the guide thus far: RapidShare of VDP Guide PDF Please keep in mind that this was run through an OCR engine, which while pretty accurate, isn't going to capture every word 100% correctly. I did a brief quality control check on it, and it should be good, though. The PDF is fully searchable, and all text is selectable. If you decide to scan the other appendices, I'll toss up another version of the PDF. Thanks again!

If you do, would you mind terribly sending me a copy for my own archives? Also, if there's interest in scanning the code examples and the sample font, please let me know. I might get around to it this weekend if there's interest. Not at all. Heck, the main effort is scanning the stuff in to begin with! I'll let you know when I'm done with it.

Absolutely awesome, thanks! If you'd like, I can OCR these images and convert them to a single PDF document with searchable text in it, which may prove helpful to some people. Any interest in me doing this?

doubledown, I may actually be interested in obtaining one of your VDPs even if mine does seem to be in working order right now. I'm sure one of my harebrained schemes down the road will necessitate having a spare on hand. *mourns the loss of a Stella chip who valiantly died in the quest to see if I could indeed watch Blake's 7 while soldering one night...* I'll let you know when I'm in a position to buy one from you!

This is awesome, thank you for this. I only hope there's something I can offer you in the future to help you out. If you would be inclined someday, I'd sure be glad to get my hands on scans of your VDP documentation. I'm sure at least a few others would be pleased to have them as well. By the by, it turns out most likely (I'm 98% sure) that I have some bad DRAM chips, so I ordered a few of them (plus spares) which I expect in any day. I'd like to seat these in some IC sockets, but I want to ensure there's no drawbacks which anyone may know of in doing so. Sockets shouldn't make ANY change in timing, latency, static issues, etc... as long as everything is properly seated and soldered, right? Also, would it be worth it to attach a heat sink to all of the replacement chips? I've read here and there that heat can be one of the possible factors in these chips going bad from time to time... Do sinks really decrease the heat enough to avoid failure for any meaningful amount of time? Is DRAM failure usually caused by heat, or do other factors contribute? If so, what are these factors? Thanks for your time and answers!

Blasphemy! If it ever happens again, why don't you just sell me the broken CV at a nice discount?

Thanks for the capacitor link - this looks like it fits the bill nicely. Interesting information on the chips, however. What difference would it make to the operation of the ColecoVision to use TMS4116-15NL versus the TMS4116-20NL? (I ask because there's a lot of 15NL on eBay right now for a semi-reasonable price...) I also think you may have the right idea as far as the TMS9928ANL goes, unfortunately. It looks like at some point someone tried to solder a new chip in, or perhaps just to resolder the existing chip as there's quite a bit of leftover burned looking flux(?!?!?!) on the underside of the board and a few really amateurish looking solder spots midway down the chip (including one pin that looks as if it has next to no solder on it at all - just floating in the pin hole). I don't have a logic probe or the knowledge of how to use one at the moment, but I'm a quick study at these types of things and think I could learn readily enough. It's not as if I have to get the CV working in the next 12 minutes, after all... :-P I think my path will be the following: 1. Reseat the TMS9928ANL in it's socket and hope I don't have to replace it. * Do you know of a nice cheap source for replacement VDP chips, or what an average price for one should be? 2. If the reseating doesn't work, I'll buy a logic probe and troubleshoot the DRAM chips. I'll replace as necessary. * Is there a chart perhaps that lists the internal ColecoVision boot screen's character map's memory addresses? The same letters are constantly corrupt on this screen, so I'm hoping maybe I can use this information to my troubleshooting advantage. * Is it best to not rely on a map and go with the logic probe, if possible? 3. If the DRAM chips come out reading correctly, do you have any other suggestions on what I can check? * The AC adapter is supplying the correct voltages; I've cleaned all contacts between the AC plug and the AC socket on the CV. I've tried a second AC adapter to confirm it's not an AC problem. * The original awesome switch has been replaced with a much more reliable recent model. Thanks again for your advice!