Panther Posted June 1 Share Posted June 1 Well, I should probably just wait until I'm ready to open up the NUC to install a NUCplus4... 1 Quote Link to comment Share on other sites More sharing options...
+mytek Posted June 18 Author Share Posted June 18 On 5/31/2024 at 11:49 PM, Panther said: Well, I should probably just wait until I'm ready to open up the NUC to install a NUCplus4... Perhaps it's a good thing you waited because new information has come to light about the primary oscillator duty-cycle issue, and what specifically caused it to be off. So first a little bit of history... Back when the 576NUC+ was in development I had started out with a 74HC14 for both the primary crystal oscillator and the buffered CSYNC. And as part of the discussions nearing the end, someone raised a concern about the CSYNC buffer aspect not switching fast enough and suggested changing over to a 74AHCT14 instead. Well not being true to myself, I gave in and accepted this advice, even though my spidey sense started tingling . So long story short the production version boards went out with SN74AHCT14DR chips specified, and even had this on the silkscreen for U19. So thanks to someone wishing to have the gerbers for the 1088XLD, I started perusing the multitude of pages for that topic, and discovered some scope pictures of the oscillator output for that unit. Guess what? Yep the duty cycle for the 1088XLD was a perfect 50/50. And it employed almost an identical circuit as was used in the 576NUC+, albeit having 2 such oscillators one for PAL and one for NTSC. In fact that's where the circuit in the 576NUC+ was originally derived from. So what was different? The XLD was using a 74HC14 (specifically a SN74HC14DR). So I grabbed my modded 576NUC+ and ripped out the previous duty-cycle mods and replaced the 74AHCT14 with a 74HC14 instead. Here's what that looked like... Primary 3.579Mhz oscillator with an almost perfect 50/50 duty-cycle, being very clean and with minimal under or over shoot! Buffered CSYNC also looks good! And 2 Stupid Dogs demo also looking good! In the future I gotta stick to my guns, and do what feels right from the get go. So from this point forward substitute a SN74HC14DR for the 74AHCT14. And probably over the next week or so I'll be fixing the schematics, BOM, and the gerber silkscreen to reflect this. If the 74AHCT14 is already soldered in place, then go ahead and do the resistor/capacitor mod to correct its duty-cycle (end result is the same). EDIT: Well I spoke too soon. Although the substitution of the 74HC14 for the 74AHCT14 works very well to regulate the duty-cycle, I noticed that there was still a tad bit of color fringing on some edge transitions. So I started thinking that the harshness of the square wave output may need a bit of smoothing out, so back in with the 150pF capacitor from the output of the oscillator to GND, and that did the trick. Here's what the oscillator output looks like now. Notice the rounded edges. Edit2: It appears that 300pF is even better at neutralizing any color fringing, while making the oscillator output a bit more like an analog sine wave in nature. Adding the capacitance across the oscillator output and GND doesn't seem to have much if any affect on the actual color tint or saturation which suggests that the frequency is unaffected, but it definitely corrects the color fringing that was present without the capacitor in circuit. So it appears that the GTIA oscillator input prefers a sine wave over a square wave, which is what the original 2 transistor crystal oscillator would have done. That of course would be too radical of a change to implement in the 576NUC+, so I'm glad that the capacitor smoothing seems to do the trick. Edit3: making the waveform more of a sine wave throws off my scope's built-in frequency counter, notice the slightly lower frequency reading. 3 Quote Link to comment Share on other sites More sharing options...
Panther Posted June 18 Share Posted June 18 I'm glad you researched this and came up with a workable solution. Perhaps I'll wait a bit longer until you're certain about the best course. Actually, I'm still going to wait for that NUCplus4 anyway. =B> It doesn't make sense that a semi-sine wave would upset the frequency counter. The peaks are still the same. Quote Link to comment Share on other sites More sharing options...
_The Doctor__ Posted June 18 Share Posted June 18 interval trigger might just barely miss seeing the last peak in it's sample frame, which certainly would be peculiar. The only other thing would be the resonant frequency being changed by load and capacitance. Try another counter maybe. If it's the same then some more tuning might be in order/ or not. Quote Link to comment Share on other sites More sharing options...
+mytek Posted June 18 Author Share Posted June 18 1 hour ago, Panther said: It doesn't make sense that a semi-sine wave would upset the frequency counter. The peaks are still the same. I'm not the first to see this effect, and my particular scope is but one of several that at times render an inaccurate frequency reading. But do keep in mind that my scope is not a high dollar instrument. However I do have one that is, an older Tektronix 500 Mhz storage scope that I can verify against. BITD it cost several thousand dollars and was considered top of the line. However I'm not too worried about it, because I can likely rely on the fact that the color hue did not change by smoothing out the waveform, so the chance of any frequency deviation is unlikely. Somebody once described on their YT channel why some of the cheaper scope's frequency counters weren't always accurate. I just don't recall what that reason was, nor am I going to lose much sleep over it either. Quote Link to comment Share on other sites More sharing options...
+mytek Posted June 18 Author Share Posted June 18 And now for a distraction from our regularly scheduled programming. Back when this all began... this was the first mocked up concept image I created of what was known as the iTari 576XE This eventually morphed into the 576NUC+ of present day. For the full history go HERE. 8 Quote Link to comment Share on other sites More sharing options...
+mytek Posted June 18 Author Share Posted June 18 This one's for you @Panther I finally got everything working properly on my prototype, so now it's just a matter of finishing up docs such as the BOM. This diagram has been updated to reflect changes that were required to have a happy system. You can see the "4" different aspects in brown numerals that make up the plus4 system. If there is no immediate interest in FujiNet, then that board can be left off or installed down the road when it's needed. Without FujiNet it defaults to the SDrive for SD card storage and retrieval, and still incorporates disk rotate functionality. With the FujiNet board installed, you can switch back and forth between either SDrive or FujiNet via the keyboard function keys F8 and F9. The boot-up will always default to the A8 Pico Cart, but to access the other drives simply type "X" at the Pico menu to exit (booting into the Pico Cart upon power-up is extremely fast). This is going to be released later this Summer of 2024 4 3 Quote Link to comment Share on other sites More sharing options...
Panther Posted June 18 Share Posted June 18 Awesome! This looks great, and will enable everything I could want from the NUC. Great photo by the way. Summer 2024... oh, so this Thursday! Sorry, I had to. 1 Quote Link to comment Share on other sites More sharing options...
+mytek Posted June 18 Author Share Posted June 18 4 minutes ago, Panther said: Awesome! This looks great, and will enable everything I could want from the NUC. Great photo by the way. Summer 2024... oh, so this Thursday! Sorry, I had to. You're too generous - 2 whole days 🙄 And don't go thinking you're peskiness had anything to do with the timeline, I just had a very lucky troubleshooting session and discovered a capacitor that was about 5 times the value it needed to be on the FujiNet auto program mode circuit. So that makes for a simple component change, with the PCB design unaffected (the best kind of problem). 1 Quote Link to comment Share on other sites More sharing options...
Panther Posted June 19 Share Posted June 19 7 hours ago, mytek said: And don't go thinking you're peskiness had anything to do with the timeline, I just had a very lucky troubleshooting session and discovered a capacitor that was about 5 times the value it needed to be on the FujiNet auto program mode circuit. So that makes for a simple component change, with the PCB design unaffected (the best kind of problem). No, no, no, of course not. But still, I appreciate it's coming in the near future. I have my workroom torn apart right now anyway, so no access to my workbench with my soldering equipment. It sounds like your project is begging to be finished. 1 Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 6 Author Share Posted July 6 576NUCplus4 Custom Case Complete Just a quick look at the latest case design for the NUCplus4 system. Sorry that it's only in SD, but my GoPro batteries were dead. Down the road I'll do an HD version. I just got a new Bambu Lab A1 mini Combo printer (came with AMS Lite - the 4-color filament station), and gave it a go. I really like how the text and the Atari logo pop!!! Matte Black for the body also works very well in this Darth Vader version. I did try a 2-color print process for the lower front panel where the joystick and some button graphics are, but that didn't work all that well, and it was a tremendous waste of filament during the color changes. I think I'll just get a silver Sharpie or paint pen and go back in and highlight that stuff. Thank you @Mr Robot for your original shared 576NUC+ case STLs 👍 Check out my Blog Post for more info on this 9 Quote Link to comment Share on other sites More sharing options...
Brentarian Posted July 7 Share Posted July 7 That looks amazing! 1 Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 7 Author Share Posted July 7 4 hours ago, Brentarian said: That looks amazing! Thank you Brent Yeah I'm quite pleased with the end result. Of course Mr Robot's base design gave me something solid to start off with. It was just a matter of building upon that, and adding my own personal touches. I also decided to veer away from the use of nutserts to fasten the two halves together, and opted to go for long black oxide coated screws and nuts instead. With both being recessed into the case so they are hardly noticible and in my opinion offer a much easier solution to implement, while still being very strong. And moving to the Bambu printer was the icing on the cake. Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 7 Author Share Posted July 7 Some technical printing info... What I call the BADGE is kept as a separate piece, being printed face down. The text and Atari logo are recessed 0.03 mm into the face, and then I used the color fill option in Bambu Studio to individually select each recessed object to be filled with a different filament color. This ends up with the text and logo being given 4 passes of the alternative color filament creating a very solid color transition. The BADGE is aligned with the BASE via 3 pins, and there is a recessed place in the front face of the BASE to accept it. Then it gets locked in place with a dab of hot glue on the backside of each pin. Conceivably the BADGE could have been integrated into the case and then been colorized within Bambu Studio. But that would have required too many filament changes to render the different colors, and been both slow and wasteful. Also by printing the BADGE face down we end up with a much nicer finish. The LID's text and logo were also recessed 0.03 mm into the face and then color filled within Bambu Studio and printed face down. The hole above the Atari logo is for a fresnel LED lens to pop in that shows FujiNet status on a board mounted 3-color RGB LED from below. I used TinkerCad to do all the CAD manipulation and redesign of the original STLs provided by Mr Robot. Printing Surface: Stock A1 mini flexible steel PEI magnetically held textured plate. 7 Quote Link to comment Share on other sites More sharing options...
+Stephen Posted July 7 Share Posted July 7 You really make me love the fact that my printer bed is maybe 30mm too small to do a print (resin). Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 8 Author Share Posted July 8 2 hours ago, Stephen said: You really make me love the fact that my printer bed is maybe 30mm too small to do a print (resin). Maybe it's time to get an FDM printer. The Bambu Lab A1 mini Combo is on sale for $350, $110 off the regular price. This gives you 4 color printing. 1 Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 23 Author Share Posted July 23 Latest Update on the NUCplus4 Project Because of the sheer density of all the stuff that's being integrated into this 576NUC+ expansion system, I was getting concerned about heat. So I found an inexpensive, small, and very quiet fan to try to mediate that problem. At first I just tried free floating it in the case, and even made a simple bracket to hold it, but without some sort of shrouding it wasn't very efficient. The shrouding does improve the cooling effect quite a bit, because with a fan just suspended in mid air it tends to draw air from all around and not just past the objects of interest. So I decided to design a 3D printed shroud for it and while I was on that journey I made it look like a finned copper heat sink for the heck of it (of course it's totally non-functional in this regard). It sits directly above the Antic and GTIA chips on the 576NUC+ mother board below, thus cooling them, while also drawing air past the other chips on that first deck. Then it flows over the upper components before finally exiting the LID vents on the upper left side. While testing in an 81°F environment, I saw a reduction of the case internal temperature of nearly 18°F with the fan installed. So I essentially took it from 100+°F down to approximately 83-84°F. A worthwhile reduction. Later I did another test with a fully enclosed unit while also running the FujiNet module with the ESP32 now adding its own heat. And this was at a much higher ambient temperature (87.4°F). The inside temperature of the case (92.8°F) was only 5.4°F higher than the outside air. Not bad 👍 This particular unit will be going to @MacRorie for beta testing this week. And I have nearly completed an 8 page datasheet going over all the pertinent, and perhaps not so pertinent details of its operation. The BOM is done. Firmware appears to be done. So this is getting ever closer to a project release. EDIT: I just did a 3D print of the Fan Shroud with Silver graphics and text. Filaments used: Overture SILK PLA Color = Copper, Bambu Lab PLA Basic Color = Silver 3D Printer: Bambu Lab A1 mini 3D Printer with AMS Lite and stock PEI textured bed plate And yes this is PLA not metal. I thought I should make that perfectly clear since a few people that I showed this to on a Zoom Meeting tonight thought it was made out of metal. So as a heat sink it would be totally useless. basically it's all for show, no go from a heat sink point of view But I think it really adds something to the overall look of the board set, and it does serve a practical purpose as a fan shroud for cooling. 6 Quote Link to comment Share on other sites More sharing options...
Panther Posted July 24 Share Posted July 24 Good idea on the shroud. That's a 20mm fan? What's the largest that would fit? I'd prefer to use a larger fan at a slower speed. Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 24 Author Share Posted July 24 10 minutes ago, Panther said: Good idea on the shroud. That's a 20mm fan? What's the largest that would fit? I'd prefer to use a larger fan at a slower speed. Yep 20x20x10 high. You can go bigger, but if sound is the concern, this one is perfectly quiet. As for efficiency in this application, it appears to be quite good, especially now that it has a shroud. Since the STL file will be released (I'll do a blank one as well without the fan or the graphics), you can make it work for whatever fan you prefer. EDIT: here you go 576NUCplus4 Fan Shroud Blank.stl 3 Quote Link to comment Share on other sites More sharing options...
Panther Posted July 24 Share Posted July 24 2 hours ago, mytek said: Yep 20x20x10 high. You can go bigger, but if sound is the concern, this one is perfectly quiet. As for efficiency in this application, it appears to be quite good, especially now that it has a shroud. Since the STL file will be released (I'll do a blank one as well without the fan or the graphics), you can make it work for whatever fan you prefer. EDIT: here you go 576NUCplus4 Fan Shroud Blank.stl 31.14 kB · 1 download I just find that tiny fans have a tendency to make an annoying high pitched whine, but I'll play with some and see. Thanks! I'll bet one of these would go great in a 576NUCplus4... Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 24 Author Share Posted July 24 7 hours ago, Panther said: I just find that tiny fans have a tendency to make an annoying high pitched whine, but I'll play with some and see. Thanks! I'll bet one of these would go great in a 576NUCplus4... Let me know what you discover, and if you discover something better let me know and I can give it a go over here as well. I'm not opposed to changing the shroud to fit a new fan. But I will have to build myself a new board set, since I'm sending off my one and only today. Quote Link to comment Share on other sites More sharing options...
+remowilliams Posted July 24 Share Posted July 24 On 7/6/2024 at 6:11 PM, mytek said: I just got a new Bambu Lab A1 mini Combo printer (came with AMS Lite - the 4-color filament station), and gave it a go. I really like how the text and the Atari logo pop!!! That case looks amazing! I'll add another huge recommendation for Bambu here as my printer (the Carbon X1) has been nothing short of stunning in just about every aspect. Forgive the appearance of the Atari adversary in the thread 😄- but I just went through much the same process as you did to create this beautiful multicolor case. 2 1 Quote Link to comment Share on other sites More sharing options...
+MrFish Posted July 24 Share Posted July 24 3 hours ago, remowilliams said: Atari adversary Most people don't look at the Amiga that way, in spite of it being a Commodore product. Quote Link to comment Share on other sites More sharing options...
_The Doctor__ Posted July 24 Share Posted July 24 1 hour ago, MrFish said: Most people don't look at the Amiga that way, in spite of it being a Commodore product. Because it pretty much is an Atari product and Atari 8 bit successor. 3 Quote Link to comment Share on other sites More sharing options...
+mytek Posted July 25 Author Share Posted July 25 On 7/24/2024 at 10:56 AM, remowilliams said: That case looks amazing! I'll add another huge recommendation for Bambu here as my printer (the Carbon X1) has been nothing short of stunning in just about every aspect. Thank you, and yours looks great as well! On 7/24/2024 at 10:56 AM, remowilliams said: Forgive the appearance of the Atari adversary in the thread 😄- but I just went through much the same process as you did to create this beautiful multicolor case. When it comes to retro-computers there are no adversary's as far as I'm concerned. Having passed the 6 decade mark myself, I really can't fathom how there would still be someone silly enough to worry about such things When it comes to computers, I create based on where ever the wind moves me... Here are two recent projects, one for the C64 and the other as a ColecoVision alternative much like the 576NUC+ is to the Atari. 5 Quote Link to comment Share on other sites More sharing options...
Like us on Facebook
Follow us on Twitter
AtariAge on Instagram
Watch us on YouTube
Chat on Discord
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.