+mytek Posted October 27, 2018 Author Share Posted October 27, 2018 Re-visiting the revised video output section. In order to better ascertain if my simple transistor buffer design would work, I built a little test board to take the place of the UAV on one of my 1088XEL's. I just noticed I stuffed a 1K resistor for R1 instead of the spec'ed 1.5K resistor. Hmm... after checking on a scope, the output looks good, so I think I'll keep it that way. After firing it up I noticed that there were evenly spaced vertical lines in the S-Video display. Argg the dreaded vertical banding problem has reared it's ugly head !!! Example of what those vertical lines looked like (image courtesy of this POST) ... This is caused by chrominance bleeding over into the luminance line. I had thought by adding an independent transistor buffer stage for the composite video output prior to injecting the chrominance signal, that I would avoid this problem. Well I guess the answer to the million dollar question is "No it didn't". So for now I eliminated the entire composite output circuit, but the real culprit is capacitor C2. Just clipping this out of the circuit solved the issue. And now for the comparison test results (These are not video captures. Taken with a camera looking directly at an LCD monitor)... The S-Video results are very good indeed. And for all practical purposes looks identical to the UAV's video output in this regard. This I am happy about . In order to solve the chrominance bleed-over caused by adding the composite video output circuit, would likely take a true video buffer amplifier IC to isolate that part of the circuit from the S-Video outputs. Unfortunately those video buffer ICs have gotten pretty scarce and expensive in a single supply thru-hole device. So I'll need to put on my thinking cap to see what alternative approaches could be used to satisfy the requirements. for the time being I like the S-Video results, and will be keeping that part of the circuit in my 1088XLD design. 4 Quote Link to comment Share on other sites More sharing options...
Kyle22 Posted October 28, 2018 Share Posted October 28, 2018 It looks MUCH better than my original 800 on this 24" Gateway monitor. It has a little bit of vertical banding, but not that bad. 1 Quote Link to comment Share on other sites More sharing options...
+mytek Posted October 28, 2018 Author Share Posted October 28, 2018 The actual onscreen video looked even better then my sample photos. Although after many attempts and fiddling with camera settings, I was finally able to capture something close to what it looked like in reality on my Sharp LCD monitor. 1 Quote Link to comment Share on other sites More sharing options...
Kyle22 Posted October 29, 2018 Share Posted October 29, 2018 (edited) It looks MUCH better than my original 800 on this 24" Gateway monitor. It has a little bit of vertical banding, but not that bad. To be clear, I mean MY 800 has more vertical banding than yours. Edit: typo. Edited October 29, 2018 by Kyle22 1 Quote Link to comment Share on other sites More sharing options...
+mytek Posted October 29, 2018 Author Share Posted October 29, 2018 To be clear, I mean MY 800 has more vertical banding than yours. Yeah I kinda figured that's what you meant . At first I thought the banding I was seeing in my bread boarded prototype was just attributable to the long leads connecting everything together, so that was the main reason I laid out a test board to be sure. Anyway it turned out to not be the cause, and was simply due to the circuit design and the inability to fully isolate the composite color injection from the luma output. So what I need is better isolation in the buffer amplifiers, and I think I found a possible solution that is relatively cheap to implement. ST electronics has the TSH343 triple video buffer chip that might do the trick. It's very inexpensive, selling for less then $2 in quantities as few as one piece, and it runs off a single 5v power supply. The only issue, is that it only comes as a surface mount device, but at least it's in the larger SOIC package which even I with my poor close-up vision can solder. I have some samples of that buffer chip coming. 4 Quote Link to comment Share on other sites More sharing options...
BillC Posted October 30, 2018 Share Posted October 30, 2018 (edited) Yeah I kinda figured that's what you meant . At first I thought the banding I was seeing in my bread boarded prototype was just attributable to the long leads connecting everything together, so that was the main reason I laid out a test board to be sure. Anyway it turned out to not be the cause, and was simply due to the circuit design and the inability to fully isolate the composite color injection from the luma output. So what I need is better isolation in the buffer amplifiers, and I think I found a possible solution that is relatively cheap to implement. ST electronics has the TSH343 triple video buffer chip that might do the trick. It's very inexpensive, selling for less then $2 in quantities as few as one piece, and it runs off a single 5v power supply. The only issue, is that it only comes as a surface mount device, but at least it's in the larger SOIC package which even I with my poor close-up vision can solder. I have some samples of that buffer chip coming. Prior to the introduction of Bryans' UAV mod there was an upgrade that used a video buffer IC, the FMS64XX. The TSH343 is meant for component video http://atariage.com/forums/topic/176257-lets-design-a-new-video-mod The FMS64XX chips buffer chroma/luma, and also create a buffered composite signal. They are discontinued and no longer available from major suppliers like Digikey or Mouser, but are available from eBay at reasonable prices. https://www.ebay.com/sch/Business-Industrial/12576/i.html?_from=R40&_sop=15&_nkw=fms6400&rt=nc&LH_FS=1&_trksid=p2045573.m1684 Correction: The FMS6406 is still available from Digikey and probably other suppliers as well, the price is higher than the FMS6400 through eBay. https://www.digikey.com/product-detail/en/on-semiconductor/FMS6406CSX/FMS6406CSXCT-ND/820964 Edited October 30, 2018 by BillC 4 Quote Link to comment Share on other sites More sharing options...
_The Doctor__ Posted October 30, 2018 Share Posted October 30, 2018 1.98 to clean up a boat load of mess sounds good to me. looks like it's got everything works for PAL and NTSC and with and can deal with stereo as well. Nice 1 to use https://www.fairchildsemi.com/datasheets/FM/FMS6406.pdf Quote Link to comment Share on other sites More sharing options...
+mytek Posted October 30, 2018 Author Share Posted October 30, 2018 Although the FMS64xx looks to be a good chip for this usage, I still like the TSH343, especially considering that it's still in production and I can source it from several vendors. The fact that it was intended for component video doesn't negate it's usage for either S-Video or composite which from an analog standpoint are not all that different. They're both using an SOIC-8 package, so no advantage/disadvantage there. But the far better pricing and availability speaks volumes for using the TSH343. Also I believe the TSH343 is better suited for direct coupling of the inputs, not requiring inline capacitors. Since I'll be doing the chroma/luma mixing for composite video myself, the built-in composite video mixing of the FMS64xx is not required (although that is a nice feature). So I read the entire thread that BillC linked to (thanks for that). And one thing that I saw as an operational concern, was the repeated mention of vertical banding being seen via S-video on LCD displays. This appeared all the way through the rather lengthy development thread, and still seemed to be the case even after the last revision. So that doesn't bode well, since my first attempt at this yielded a fully functional simple design that for all intents and purposes fulfilled the criteria for excellent quality video output, with the one caveat being vertical banding issues which made it a no go in my book. I would not think this would have been an issue with the FMS64xx design, but how to explain all the reports that said vertical banding persisted no matter the revision level of the circuit. Anyway my mission is to create clean S-Video and composite video output in the 1088XLD, as well as do a major cost cut in doing so. So that criteria is going to guide my journey and the methods employed to get there. Not saying I'll end up using the TSH343 in the final design, but for the time being it holds a lot of promise to help create what I am after. Thanks again BillC for pointing me to that alternate video development thread. It made for a very informative and good read . 3 Quote Link to comment Share on other sites More sharing options...
_The Doctor__ Posted October 30, 2018 Share Posted October 30, 2018 to kill the banding didn't wasn't it a matter of attenuating the input signal.... if not see what attenuation of the output does... I do know there were a number of crt's back in the day where they looked great with a weaker signal... Quote Link to comment Share on other sites More sharing options...
+mytek Posted October 30, 2018 Author Share Posted October 30, 2018 to kill the banding didn't wasn't it a matter of attenuating the input signal.... if not see what attenuation of the output does... I do know there were a number of crt's back in the day where they looked great with a weaker signal... The vertical banding occurs when some of the chroma signal bleeds into the luma circuit. Yes attenuation of the chroma will also influence how pronounced the banding is, but that is not the way to fix the problem correctly. Bottom line is you need to prevent the bleed-over in the first place, and the best way to do that is to have complete electrical isolation where the luma and chroma are mixed to create composite video. Essentially you don't want the mixing of the chroma to contaminate the independent luma destined to go to the S-Video output. Edit: CRTs are not affected as much by the chroma leaking into the luma, but for some reason the LCDs really see it. It is primarily an S-Video thing. 2 Quote Link to comment Share on other sites More sharing options...
+mytek Posted November 1, 2018 Author Share Posted November 1, 2018 Chroma Bleed-Over into S-Video Luma Solved on Video Test Design Changed Q2 to NPN darlington transistor (MPSA13) for the composite output, which gave me additional isolation from S-Video luminance circuit. Also adjusted a few values to render a better picture. The MPSA13 is obsolete, although Jameco has over 3,000 of them in stock. However the currently manufactured BC517 should also make for a good substitute. Composite Video The UAV has an ever so slightly better picture than what we see here, but the difference is relatively small. The UAV's composite video is a tiny bit sharper, and the text is slightly whiter in color. S-Video My S-Video results actually look a tiny bit better than the UAV (at least in my 1088XEL). With the UAV I see a very small gray bar on the left hand side just prior to rendering the blue background, whereas my circuit does not have this. I tried adjusting the UAV's trim pot, but I couldn't get rid of the gray bar. However this bar is very small, and not all that noticeable. Overall both the UAV and my S-Video circuit look very good. 6 Quote Link to comment Share on other sites More sharing options...
_The Doctor__ Posted November 2, 2018 Share Posted November 2, 2018 (edited) Chroma Bleed-Over into S-Video Luma Solved on Video Test Design ALT-VB_rev1_1_schema.png Changed Q2 to NPN darlington transistor (MPSA13) for the composite output, which gave me additional isolation from S-Video luminance circuit. Also adjusted a few values to render a better picture. The MPSA13 is obsolete, although Jameco has over 3,000 of them in stock. However the currently manufactured BC517 should also make for a good substitute. Composite Video Composite.JPG The UAV has an ever so slightly better picture than what we see here, but the difference is relatively small. The UAV's composite video is a tiny bit sharper, and the text is slightly whiter in color. S-Video S-Video.JPG My S-Video results actually look a tiny bit better than the UAV (at least in my 1088XEL). With the UAV I see a very small gray bar on the left hand side just prior to rendering the blue background, whereas my circuit does not have this. I tried adjusting the UAV's trim pot, but I couldn't get rid of the gray bar. However this bar is very small, and not all that noticeable. Overall both the UAV and my S-Video circuit look very good. Now this one belongs on all the bases... My kind of mod! +1 MyTek Gratz, _Doc Brown__ Einstein seal of approval ! Edited November 2, 2018 by _The Doctor__ 2 Quote Link to comment Share on other sites More sharing options...
+mytek Posted November 2, 2018 Author Share Posted November 2, 2018 Now this one belongs on all the bases... My kind of mod! +1 MyTek Gratz, _Doc Brown__ Einstein seal of approval ! Thanks Doc . Yeah sometimes persistence pays off. It is in my nature to never give up (quoting Galaxy Quest). Now if I only had that time machine that we all talk about. Because I would travel back to the beginning, and hand this circuit to the creators of the A8 for reference. I know they did wonderful things creating the custom VSLI chips that powered the A8, but I think they were often confused when it came to designing the video output circuits. Just to be clear... Bryan's UAV board is still the winner when it comes to composite video output quality. However when it comes to S-Video, we appear to be in the same league. EDIT: It's also nice to keep it to the through-hole theme I always pursue in my projects, thus making it possible for more DIY types to build my creations. 2 Quote Link to comment Share on other sites More sharing options...
+mytek Posted November 10, 2018 Author Share Posted November 10, 2018 MIDI DIN Interface board revision V1.2 (hopefully this is the final version) Changed to ON-OFF-ON switch (3 position) and routed trace to 3rd position so that you now have these 3 different possibilities... MIDI-OUT only to feed internal Synth (software application can optionally echo MIDI-IN) External Synth only (Internal Synth disconnected) Auto Mode (MIDI-IN or MIDI-OUT feed internal Synth automatically dependent upon activity) Updated manufacturing files and OSH Park link in this thread: http://atariage.com/forums/topic/261147-1088xel-alternative-mother-board-project/page-53?do=findComment&comment=4145545 Note: V1.1 boards are still usable, only requiring one added jumper wire and different switch. 4 Quote Link to comment Share on other sites More sharing options...
+mytek Posted November 30, 2018 Author Share Posted November 30, 2018 1088XLD Main PCB Layout Complete There have been a few changes since my last update. For one, I decided to go with my DIY video output circuitry for both the composite and S-Video aspects. And I've gone with a dual footprint on the SIO connectors, which will accept either the conventional jacks or a D-sub DA-15M wired for SDrive NUXX compatibility. And lastly, I have replaced the single Power ON LED with a bi-color version which will display 3 different states (GREEN = Power ON, RED = CF Activity, YELLOW [RED+GRN] = Disk Swap latched). You'll also notice that the copyright (creative commons actually) is set at 2019. That's because I don't intend to have a completed board for testing in my hands before next year. Lastly for those that haven't been following this thread, this NEW motherboard is designed to drop into a stock 1050 drive case with minimal modifications. Controller ports (joysticks, keyboard, mouse) and a CF drive will be implemented via passive adapter boards accessible from what once was the floppy drive bay utilizing a custom machined bezel. These adapter boards contain no active circuitry, and simply serve as either an IDE to CF connector adapter, or as a fan out of the I/O and status signals into individual port connectors and LEDs. In my first iteration of this design I don't intend to have an actual floppy mechanism, but who's to say that it won't be done in the future by some enterprising individual (I could see the possibility of installing a XF551 style 3.5" drive using one of Dropcheck's custom boards). EasyEDA appears to be the best choice for having this board made, due to it's larger footprint and 4-layer nature. Total cost for a 10 board run will be just under $100, which includes shipping. So about $10 per board. This of course is only the cost of a bare un-stuffed board. Top Side Bottom Side NOTE: PCB images were updated on 12/20/2018 at 7 AM Pads shown without a trace connected, are likely terminated to one of the two inner power planes (4-layer board). Some interesting PCB specs Number of Components: 205 Number of Holes: 1,753 Number of individual Trace Segments: 3,685 15 Quote Link to comment Share on other sites More sharing options...
+Stephen Posted November 30, 2018 Share Posted November 30, 2018 Beautiful. Sure hope I can build one! 2 Quote Link to comment Share on other sites More sharing options...
+DrVenkman Posted December 1, 2018 Share Posted December 1, 2018 Beautiful. Sure hope I can build one! Ditto. I had so much damn fun building my 1088XEL that I’ve been itching to tackle another large build again someday. With luck the money will be available when the design is finalized and the boards are ready. Heck, I’ve even got a dead-beyond-repair 1050 sitting around, too! (Someone has done ugly things to the drive mech to try to “fix” it, years before I ended up with it). 2 Quote Link to comment Share on other sites More sharing options...
+mytek Posted December 1, 2018 Author Share Posted December 1, 2018 Beautiful. Sure hope I can build one! Ditto. I had so much damn fun building my 1088XEL that I’ve been itching to tackle another large build again someday. With luck the money will be available when the design is finalized and the boards are ready. Heck, I’ve even got a dead-beyond-repair 1050 sitting around, too! (Someone has done ugly things to the drive mech to try to “fix” it, years before I ended up with it). You guys will get your chance early next year when I do an official release . Of course I'll be building up the first one myself . And I have some firmware modifications to do in the TK-II chip so that it will allow the keyboard to change mouse ports and initiate a swap disk action. I should be able to have it detect the presence of the diode on the normal TK-II board's 2nd keyboard input, and if it's not present, power up in 1088XLD mode using the 2nd port for switching instead. This way only one firmware version is required for all applications. And then I'll need to create a BOM, and then... well you get the idea, there's still more work to do before this thing sees a public release. 3 Quote Link to comment Share on other sites More sharing options...
+mytek Posted December 3, 2018 Author Share Posted December 3, 2018 DIY Video Output Circuit Change Value change for the emitter to ground resistors. Went from 75 ohms up to 150 ohms to cut back on current requirements. I'll show this on the 1088XEL UAV alternative schematic, although it applies equally well to the 1088XLD project. The original 75 ohm resistors worked, but it caused the driver transistors to run pretty darn warm, although still within their ratings. However with the emitter resistor change it not only runs cooler, but also cuts the current requirements in half, while still providing ample drive into a 75 ohm video load. 4 Quote Link to comment Share on other sites More sharing options...
+mytek Posted December 5, 2018 Author Share Posted December 5, 2018 DIY Video Output Circuit --- Added 5 volt Power Filter Well I couldn't leave well enough alone, and started playing around with ferrite beads. So with the addition of one ferrite bead and a capacitor, I've added an LC filter into the 5 VDC power for the video output circuits. In essence, this is very similar to what Atari used to do in order to isolate the digital switching circuits from the analog sections. The reason I did this, is because I was seeing a tiny bit of repetitive noise occurring in what should have been solid color in a color bar chart. And when it comes to video, I'm a bit picky. The capacitor can be anything from 470uf on up. Since I already use a 1,000uf in the 1088XLD, I'll just use the same value for this new LC filter as well. The capacitor needs to be fairly large due to the video output stages using a fair amount of current, so the extra storage capacity is required to supply that demand. Here's what the picture looks like after adding the LC filter... And the new circuit as shown in the 1088XEL version schematic (same idea will carry across to the 1088XLD)... What I had to do on my 1088XEL video test board to fit this in... Go back to 1088XLD PCB image to see it implemented into that design (L1 and C7 form LC filter in top view upper left corner). 5 Quote Link to comment Share on other sites More sharing options...
+mytek Posted December 22, 2018 Author Share Posted December 22, 2018 I was going to hold off on saying anything until I had a fully assembled working board, but me bad . I just did the first 1088XLD board run, with all of them destined to go to beta testers (sorry ). So far everything appears to be a drop-in fit into a 1050 case ( ). The only case mod required will be to enlarge the former drive select switch hole to allow for plugging into the audio/video DIN-13 connector. BTW that connector spits out Composite, S-Video, Stereo Sound, and if equipped with a Sophia or VBXE, will also pass thru the RGB/Component video. And for those talented at case mods, it looks like Sophia's DVI connector could fit above one of the SIO ports, only requiring a rectangular hole be made to accommodate it. The I/O interface puts the connectors where they make sense. It'll have an extended pin male header, as well as a female header on the main board. And although this I/O interface board would work, I'm seriously contemplating doing a new layout with tabs and holes that capture the front two floppy drive mounting pins. It would likely add a lot more rigidity when plugging stuff into the connectors. Also thinking of making a second stacking board to go above it for the MIDI DIN connectors and the CF2IDE adapter, which would benefit greatly to have the 1st board locked in place. Front View peeking through the original brown bezel. So the idea is to have a 3rd level with the CF slot to the left, and the MIDI DIN connectors to the right. Then a custom machined 4 mm thick panel with engraved labeling will fill that hole left by the original floppy drive mechanism. View without the bezel. I was playing around with Dropcheck's Cartridge extender board plugged into the cart edge card connector. Nice to see it will fit. My initial plans don't include a Cartridge Port, but at least we see that is a feasible option. Yes folks things are definitely happening . 11 Quote Link to comment Share on other sites More sharing options...
ivop Posted December 22, 2018 Share Posted December 22, 2018 That's looking very nice! Update from my side: the VS1053b WaveBlaster prototype boards have been ordered yesterday. I have received all parts (finally) and expect the PCBs before the end of the year. 4 Quote Link to comment Share on other sites More sharing options...
+mytek Posted December 22, 2018 Author Share Posted December 22, 2018 That's looking very nice! Update from my side: the VS1053b WaveBlaster prototype boards have been ordered yesterday. I have received all parts (finally) and expect the PCBs before the end of the year. Yep it's always nice to see a plan come together and see some of what was calculated actually fitting into reality. But at the same time seeing how the pieces relate to each other, brings forth new ideas and/or approaches. So yes as I mentioned there will be changes coming, but so far those seem limited to external plug-in boards and not the main one itself. Good to hear that things are still happening on your board, since that promises to be a low cost option for the MIDI Wave Blaster that hopefully will integrate with the 1088XLD. 3 Quote Link to comment Share on other sites More sharing options...
ivop Posted December 22, 2018 Share Posted December 22, 2018 Good to hear that things are still happening on your board, since that promises to be a low cost option for the MIDI Wave Blaster that hopefully will integrate with the 1088XLD. Yes, here's hoping, too. These VS1053 chips are only $3.40 and don't need an external DAC. The CS9236+CS4333 will set you back around $20. I bought the CS9236 for $9 but that was years ago. They're up to around $15.= now. Quote Link to comment Share on other sites More sharing options...
AtariGeezer Posted December 22, 2018 Share Posted December 22, 2018 I was going to hold off on saying anything until I had a fully assembled working board, but me bad . I just did the first 1088XLD board run, with all of them destined to go to beta testers (sorry ). Yes folks things are definitely happening . Cool, my new eyes are working good. Can't wait to build one 3 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.