Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


ralphb last won the day on May 2 2017

ralphb had the most liked content!

Community Reputation

1,113 Excellent


About ralphb

  • Rank

Contact / Social Media

Profile Information

  • Gender
  • Location
  • Interests
    Developers, developers, developers, developers

Recent Profile Visitors

8,112 profile views
  1. ralphb

    SDD 99

    Ha, it seems like it did work -- IC soldered, no shorts. Got rid of the paste, though, and instead pre-soldered all pads. Much easier, and also much quicker. 👍
  2. ralphb

    SDD 99

    So far I didn't have problems, probably because I reduced the air flow. But I'll look that up!
  3. ralphb

    SDD 99

    SMD parts not a problem; almost all SDD ICs are SMD (see the main controller to the left and down). What is a problem are BGAs and company. I do have a small reflow oven from China, but its temperature is too high for the step-down regulator. Flux is a must for non-BGA SMD ICs, since you need it to cover the pins evenly with solder. I doubt that it will help for hot hair, especially since it took me more than 10 minutes to heat everything up, which would make the flux evaporate. There are low temperature solder pastes containing Bismuth, but they produce brittle connections. Anyway, I think I just need more practice. The boards aren't lost with a bad solder job, just the regulator. And Digikey has thousands of them ... 🙂
  4. ralphb

    SDD 99

    The solder paste should already contain flux, but I guess an extra droplet doesn't hurt. But what I image will happen is that the flux evaporates first, before the solder becomes liquid. You should be able to see the capillary force align the part, but my lighting with the heat gun right on top isn't that great (EDIT).
  5. Ah, I misspelled FORTIFY, but it worked on the second try: cassiopeia /vol/src/mame225 > time make -j 24 ... Linking mame64... 15366.948u 980.779s 14:04.90 1934.8% 0+0k 688+9040208io 2pf+0w I was basically burning-in my new system, and it passed the test. 😀 The -j 24 hints at the processor, and 14:04 is the elapsed time in minutes and seconds (of which 980 seconds were spent inside the kernel).
  6. It's a compiler problem (according to MAME devs), and there's a workaround by allowing unsafe functions with _FORTIY_SOURCE. I'm trying that now.
  7. I've tried to compile version 225, but I get an error: In function ‘char* strcat(char*, const char*)’, inlined from ‘void nubus_image_device::file_cmd_w(uint32_t)’ at ../../../../../src/devices/bus/nubus/nubus_image.cpp:262:10: /usr/include/x86_64-linux-gnu/bits/string_fortified.h:128:33: error: ‘char* __builtin___strcat_chk(char*, const char*, long unsigned int)’ accessing 129 or more bytes at offsets 988 and 860 may overlap 1 byte at offset 988 [-Werror=restrict] 128 | return __builtin___strcat_chk (__dest, __src, __bos (__dest)); | ~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ I guess I need to change that to a memcpy, but I don't know exactly where and how. Have there already been official patches for this? EDIT: I found where, but not how. 😅 EDIT²: I found the mamedev note about _FORTIY_SOURCE.
  8. ralphb

    SDD 99

    That wouldn't do, since it's the IC that is so small (and not available any bigger). Also note the (covered) copper areas around the IC; their size and geometry is set in the datasheet. Any other layout, and the power supply might not work.
  9. ralphb

    SDD 99

    I just had a go at the new boards. Suffice to say that my first board is junk, because of one new IC. On my previous boards, I used a linear power regulator, because they're pretty simple to use. But with upto 1.2 A (after recalculating the power usage) and 1.7 V voltage difference, that's more than 2 W and a lot of waste heat. (It'd be even worse for the PEB card, since we have 8 V to regulate down to 3.3 V.) So I decided to switch to a so-called step-down regulator, which does not produce that much heat. You'll see the problem once you see the IC: The large pad on the bottom has no side connectors, thus cannot be soldered conventionally. Either you put large vias below the pad that you can solder it "from below", or you use a heat blower and solder paste to solder this IC. Either way is not made easier by the actual size of the part: Anyway, in my first attempt I used way too much solder paste, which did solder the IC, but also produced a viable short -- any pin is connected to all other pins. 😝 I have 4 or 5 more of these ICs, let's hope I'm more successful with the next board! 🙈
  10. ralphb

    SDD 99

    Yeah, I don't think that will be necessary ... There are plenty of "official" cards that don't have a case! Besides, ordering a PCB that fits the PEB box will be more expensive than a sidecar, since you have the bus connector at the bottom and the peripheral connectors at the top. My go-to producer even has a 10x10 cm2 limit that probably wouldn't work for the PEB, so some creative solution is needed.
  11. ralphb

    SDD 99

    Beery, I agree on both counts. Early on, someone contacted me about 3d-printed cases (he also did cases for the FinalGROM). You could also buy a right-sized plastic box as case, but you have to be careful to not put too much stress on the edge connector -- the PCB has to be at exactly the right height. This is probably easier with a printed case. I also will start on documentation soon. The SDD is quite complex, and I think it would be nice if the beta testers will have some reference material as well. I plan to do online but also a PDF version.
  12. ralphb

    SDD 99

    No, it still takes some time. Next week I will build the sidecar, then test it, and work on the firmware a little. Then I can send out a few boards for beta testing. If everything is OK, I can order some assembled boards, which are then for you. If the built sidecar works, I can also start designing the PEB board, which keeps the main logic intact, but needs a new power supply and a reordering of connectors. Maybe some connectors have to be removed. Once the design is ready, the process will repeat. Thus, the sidecar will be ready before the PEB card.
  13. ralphb

    SDD 99

    Today I received my final boards, where final means that I won't change them again, unless there are any errors. The layout is two centimeters narrower, but I cannot shrink it further, or the connectors won't fit anymore. Over the next days I should receive some more parts I'll need, and then I can build a sidecart. With that, I'll be able to test the self-update I implemented over the last weeks, and the parallel port and general I/O also need some work. Keep your fingers crossed that this one works! 🙂
  14. Somewhat off topic, but my upcoming SDD 99 is able to play PCM files, which can be extracted from WAV files. The SDD produces 8 bit, 44,1 kHz sound, which is half CD quality. Alas, there are no good demos, since anything noticeable would be killed by YouTube. 😔
  15. Oh, you're right -- I can reproduce this with the latest version of xdt99. I guess I didn't test it on the FinalGROM. Either you wait for a fix, or you use this older xdt99 version: https://github.com/endlos99/xdt99/releases/tag/2.0.1 EDIT: Actually, the reason is that make.py is still written for Python 2, but xdt99 is using Python 3 now. If you replace make.py with the attached version, everything should work. make.py
  • Create New...