Another (new) 130XE boots to red screen

[woj]

Right, thanks Jon! Yes, the factory video traces were my worry, I did figure that I would have to bother them to get the DIN13 mounted in either case.

 

In the meantime, I came up with a very dangerous solution C, which is to design a custom dvi to hdmi board. Dangerous, because it will cost me and it will further delay my PopEye project, all with the assumption that I will not pass on also having the audio hooked up to the hdmi, which, as far as I understand so far, is the only complicating part of this.

[Beeblebrox]

On the move so skim reading but my pennies worth. 

 

The din13 jack addition for VBXE using the cool novelties DIN13 to scart is one mod I've done in both an 800xl and an XE. 

 

The 800xl is easier. The XE mod is less so but you can do what I did and mount to the right side of the board down from the joystick ports. Means I could still keep the rf modulator in order to keep legacy video. 

 

As fjc said aesthetically the din13 installed in an 800xl in place of the modulator looks great and it carries audio too. Unlike an XE when it comes to the XL the  modulator is not required as you still get to keep legacy video with it removed. 

 

There is also the savo board in conjunction with VBXE as an option. 

 

 

Edited March 19, 2023 by Beeblebrox

[flashjazzcat]

23 hours ago, woj said:

In the meantime, I came up with a very dangerous solution C, which is to design a custom dvi to hdmi board. Dangerous, because it will cost me and it will further delay my PopEye project, all with the assumption that I will not pass on also having the audio hooked up to the hdmi, which, as far as I understand so far, is the only complicating part of this.

Sounds great. Related to this: the only problem I've experienced when using passive DVI->HDMI adapter cables is persuading the TV/monitor to accept auxiliary audio in (on the 3.5mm jack) when HDMI input is selected (since we do not yet have digital audio from the Atari).

23 hours ago, Beeblebrox said:

The 800xl is easier. The XE mod is less so but you can do what I did and mount to the right side of the board down from the joystick ports. Means I could still keep the rf modulator in order to keep legacy video.

General note: only composite is killed on the XE (not XEGS) when the RF can is removed (s-video still works), and if you're lucky enough to have UAV as well, composite doesn't require the RF modulator. There are no traces on the underside of the XE board anywhere near the holes for the DIN13 connector (while you can easily hit some on the XL if you don't position the connector properly).

23 hours ago, Beeblebrox said:

There is also the savo board in conjunction with VBXE as an option. 

Oh God... you said it out loud.

[woj]

Feeding audio to the HDMI signal seems to be quite difficult, I am definitely putting this project on the side for the moment. The Sophia works, the picture is great, even s-video is much better. I do, however, have some huge stability issues, quite often the computer does not want to boot at all, just goes into a blank screen (video not initialized). I had some problems of this sort before on this machine along with some GTIA socket sitting issues, it was fixed, but it seems Sophia brought it back, have to investigate a bit more before I drill any holes in the case.

[woj]

Something definitely wrong with the XL, the typical pattern is that the machine boots fine, cycling it with RESET is fine, but as soon as I turn it off and on again (BTW, I am rewatching IT Crowd at the moment, could not help it to mention this) it boots to a blank screen. It did have symptoms like this before I installed Sophia 2 (in fact, I recall such things when it arrived at my door during first tries with it), but at some point I think I got rid of them with the stock GTIA, though totally not sure what was the thing that cured it. It is clearly some timing and/or holding the state by one of the chips issue, because once I also got the machine to boot directly into Basic instead of going blank. The other thing that I know for sure is that it does not like the stock OS rom chip, memory test runs with errors there, while the OS rom chip seems to be perfectly fine outside of the XL habitat, and the XL works perfectly fine with a self burned EPROM. Also, if I recall correctly, the MMU was socketed from factory (nothing else was in sockets), so I presume they replaced it post assembly line. That's all diagnosis things I have for it.

 

I already tried going back to 74LS08 from 74F08 that I installed, nothing changes (though taking the chip out and in seems to make the machine boot again, but it does not matter which chip). I am yet to do some swapping around with the 130XE to try to narrow it down a bit more, and also I know I disturbed one of the 33 Ohm (I think) resistors on the memory lines close to GTIA during some work a while back, it is a bit cracked, I have a replacement which I now ought to try just to eliminate this as a possibility. Otherwise I am not planning to go too deep into this as I have better things to do. But this also means that the Sophia 2 will not be fully installed before this is fixed.

[_The Doctor__]

The first thing you do is replace the physically damaged resistor. No brainer on that one. Punching out...

[woj]

I am not going to skip on that, only that there is logistics involved with taking apart the computer, cleaning up the desk for soldering, etc. (not to mention finding out where I stuck the new resistor) so I thought I would quickly check some easy to do things before that. What I just did is I put back the factory MT RAM in, I ran the thing on KM4164B-10-s, these are considerably faster than MTs, 100ns vs 150ns. It seems to have made some difference, though it still hangs on boot, but now it is more likely to eventually boot, properly without going straight into Basic, and also it is possible to help the unbooted machine by pressing RESET, though this does not always work either. 

 

The resistor will be replaced along with another bigger soldering work I have to do, in a few days I hope.

[woj]

So it seems it might be the Sophia after all, I get the same thing, though less frequent, when I put the chip into the 130XE. Blank screen and a silent high pitch buzz on the audio channel. So I guess I have to talk to Simus.

[woj]

So now that I got my 800XL to work with the SRAM module and learning a bit about O2/BO2 intricacies I can move on the actual thing I was up to, let's do it here.

 

The challenge presented itself a couple of posts up, when @flashjazzcat suggested it would be possible to fit a DIN13 connector for VBXE on the side of the DVI one in an 800XL, but that would require disturbing the factory video signal traces. One side of me accepted this and I was slowly getting my drill ready, the other side said "hell no!". Some research later - Sophia also outputs RGB signal and does so through the DVI connector, so why not hook up the VBXE RGB lines through there and make an appropriate DVI to SCART cable? Simple enough, but what about keeping the Sophia HDMI signals at the same time, and also still allowing Sophia to output RGB instead of VBXE when needed (for whatever reason)? So, ideally to be able to switch the RGB ouput between Sophia and VBXE with a jumper or a small slider switch, or even in software from U1MB plugin menu...?

 

That all called for designing a small board with an analog switch IC that would plug-in in between the Sophia's DVI board and Sophia and take VBXE signals on the side. The biggest challenge was to fit all this so that it would stay within the DVI board frame. One evening of designing the circuit, one evening for the board, couple of weeks of wait for getting the board, and here it is:

 

 

This is a first quick solder up just to see if it works, it will be the only one for now though, I found out yesterday I am short on 1K resistors, not sure how I missed that 😕. First tests show that it works, but I still have to wire up the VBXE to this. I also did implement an extension to Jon's U1MB plugin to control the output, with this there is a problem though, this setting in software cannot be fully "persistent". The U1MB firmware pulls the outputs low during boot and BIOS entry, so during that time it always has to be Sophia VBXE RGB output, it can only become the selected one once it gets to the actual OS boot, or forced by a change from the menu while in the BIOS menu:

 

 

(The only thing is I am very afraid to hook up the wire to U1MB knowing from @mytek how easy it is to fry it 😕, even though I checked with him what should be safe).

 

Finally, I also designed a twin board for my 130XE that does a similar thing, but there it is to allow Sophia to output it's RGB signal through DIN13 alternative to VBXE. This has to wait until I get my 1K resistors though. Oh, and I am soldering all this on a budget homemade hot-plate   which make a whole lot of a difference:

 

More to come once the wiring is finished and more tests made.

 

Edited October 4, 2023 by woj

[woj]

Absolutely sweet, everything works as expected, U1MB is still alive, cables found their route and nice connectors. I knew this can be done and it is not too difficult, but I never suspected it would take off on the first attempt  The only rose torn is that U1MB keeps the M1 pin down during its "thing" as I described above, so in particular the BIOS menu will always use the VBXE output initially, no matter what the setting. I actually did try to be more aggressive in setting the pin state manually during the plugin init phases, yet, there are still times when this is out of the plug-in programmer control, and I know that because I checked with Altirra that the corresponding register gets 0 written to it outside of the plugin code. 

 

On the video, Sophia output is considerably darker, and it also has the VGATE option on, you can see the picture being clipped on the sides.

 

Also, if you look carefully at the picture you will notice that I took the csync signal out of the composite output pin for the VBXE connection, while Spectre has a separate pin for this (just next to it). I measured the voltage on this pin and it seems (I just used the simple voltometer) Spectre provides TTL level sync signal (it perhaps is exactly like it should be and up to corresponding specs), which is not entirely OK with RetroTINK that I have, or at least its documentation warns about this. So to be on the safe side I took the composite one instead. Interestingly, the csync signal of Sophia2 seems to much lower level, but that was also checked with just the voltometer. (Why I switch between the two csync signals, while the one from Sophia would be probably sufficient for both devices, that I actually do not know, I thought it would be a good idea, just in case they are meaningfully different). 

 

 

 

[woj]

For some time I was doubtful if this exercise with switching RGB outputs is worth the effort, but now I know it definitely was. I connected the computer to my living room TV that has a built in, rather poor, up-scaler. On this one the superiority of Sophia's RGB output is very visible. With VBXE there is a visible checker board patter highlighting half-pixels on blue screens, not so much on the attached picture (look on the left-hand side, on the right what you see is camera artifacts), but very pronounced in reality. Nothing like this with Sophia, the blue is solid. Ghosting is the same on both, so adding that Sophia has the VGATE option to clean up the sides in some programs, Sophia's output is my choice here for non VBXE graphics. At the same time, I can't use the Sophia's HDMI mode on this TV, first it does not want to sync (last time I checked, it does not like the almost but not exactly 50Hz signal), and there is no way to convince the TV to use alternative audio source. So Sophia RGB is the best I can do on this one. 

 

I also spent the whole evening making sure all the extensions clear the keyboard to be able to close the case gracefully. That required making a new keyboard connector and cable (the factory one got really messed up when I started bending it many ways to clear the stereo board), forgot to take pictures, and it also turned out that the front chip extensions (SRAM, VBXE) lift them high enough to prohibit to use the bottom RF-shield at the same time (yes, I know, it is not needed, but until this points there was no problem having it, and I like the fact that it adds to the weight of the machine and provides stiffness to the motherboard).

 

Anyhow, this one awaits a small distance screw for my Sophia extension board (it is wobbly a bit with just relying on the red connector to keep its position) and this one can be called done and closed. A similar RGB board for my 130XE is awaiting parts to be soldered up too, now that I see there is a point in having this functionality.

[flashjazzcat]

On 10/4/2023 at 9:08 AM, woj said:

I also did implement an extension to Jon's U1MB plugin to control the output, with this there is a problem though, this setting in software cannot be fully "persistent". The U1MB firmware pulls the outputs low during boot and BIOS entry, so during that time it always has to be Sophia VBXE RGB output, it can only become the selected one once it gets to the actual OS boot, or forced by a change from the menu while in the BIOS menu

Yes. This kind of control method was attempted before with the AntiX device and it proved to be a miserable waste of time. The device being controlled needs to latch itself to the desired state if you want things configured a certain way from the point the power is turned on, since the aux config isn't set up until the OS is booted.

[woj]

An interesting next finding. I acquired two Sally CPUs from eBay from a Slovakian seller (perhaps someone on here?), these are refurbished used ones, not sure which make though, there is no original print and a sticker instead. Anyhow, none of the two work with the SRAM module and SIDE3 in my 800XL, they both crash exactly the same way as the NCR one. (Without SIDE3 they work just fine.) So now the statistics is that out of the four CPUs I have, only the Mexican one is fine for this setup.

 

Some other things arrived too, they will be some updates about my RGB switching boards 😉

[woj]

A lot of cable crimping, soldering up, and my RGB boards, both different, are cleanly mounted in both computers. I also took the opportunity to make a new, slightly longer MMU cable for the 130XE to route it more nicely and not to lean on the UAV. My RGB board in the 130XE sits on top of Sophia and it's kind of difficult to mount any distance, even though I prepared a hole for it. I'd need to make a custom one out of something non-conductive not to short Sophia's pins. What is also visible is the self made keyboard cable and connector in the 800XL.

 

Definitive next project is the Decent XE keyboard, just that I still have not made my mind as to vintage vs. modern.

[woj]

This might be some entertainment for some of you, though honestly probably rather boring videos to watch (they are both at double speed to make them more bearable), I was fine tuning my hot plate parameters in preparation for Ultimate Cart and PokeyMax V4 soldering, still had some boards and parts left from previous projects, so decided to film the process. It did allow me to see where the small solders balls are coming from, it might be totally destructive when doing BGA components. I had that problem since day one, and changing the temperature curve either direction did not help much. It is clear from the videos that having less paste would do the trick most probably. With these kinds of components this is easy to clean up afterwards, pictures of that attached too. Well, let's see how the next project goes that has much smaller pitch components...

 

 

 

[_The Doctor__]

Ironing board soldering 101

might be a little hard on the PCB, maybe infrared heat detecting gun used to set temp?

[woj]

12 minutes ago, _The Doctor__ said:

Ironing board soldering 101

might be a little hard on the PCB, maybe infrared heat detecting gun used to set temp?

The iron plate has two temperature sensors screwed in, and the whole contraption is PWM controlled by Arduino, and it follows the curve advised by people "out there". The only problem is the inertia, especially downwards, to cool the board down I need to lift it by hand, but keep it in the vicinity for a while to get the right effect.

[woj]

Another totally exciting clip, even more so, because it's slightly out of focus , all of the three bridges that formed that I mentioned in the Pokey thread are not in the visible spots (of course):

 

 

 

[TZJB]

58 minutes ago, woj said:

Another totally exciting clip, even more so, because it's slightly out of focus , all of the three bridges that formed that I mentioned in the Pokey thread are not in the visible spots (of course):

 

 

 

 

Thanks for sharing this @woj. It is quite exciting if you like watching solder melt, and I do. But I couldn't hear any frying sound which would make it much better 😁!

 

Your soldering method looks quite well managed with hardly any smoke, although I would like to see your preparation method for orienting the chips and solder paste.

 

Just an observation regarding this cartridge's edge connector. For the cartridge to have a long life without contact troubles, I thought best practice was to have gold plated contacts and a bevelled edge?

 

[woj]

I will try to be more considerate next time and film some more, no sound on video because the only thing you would hear was my burping and my my kids shouting in the background. But some close up photos (and only now I see I have some more solder balls to remove), the back came out nice too, even though the paste was applied by hand with a toothpick and soldered with hot air gun, the first board I did the back with soldering iron and regular wire tin. BTW, this is all fully lead free. Anyhow, this board is now alive too, if someone in the EU zone (really, I am not going to fiddle with custom declarations and such, sorry) is interested in getting one of them cheap-ish (my parts costs plus shipping) without the case, let me know through PM. 

 

Regarding the connector - I did the bevel by hand / sand paper, the gold plating is something that escaped me, but this was also supposed to be an exercise in FPGA board building (to prep for the PokeyMax 4 assembly) rather than producing a shiny thing, so I cut costs.

[woj]

A half honest question, cleaning up my desk I just saw the measure of my addiction - how from a person "I will be fine with one versatile cart and an SDrive" did I end up with this lot?:

 

 

[TZJB]

49 minutes ago, woj said:

A half honest question, cleaning up my desk I just saw the measure of my addiction - how from a person "I will be fine with one versatile cart and an SDrive" did I end up with this lot?:

 

 

 

I did originally think I would only need an S-DriveMAX as it was so cheap to make, but then the Fujinet came out for networking, then I needed a cartridge emulator, so AVGCart. I stopped there.

 

You probably really only need one cartridge emulator if it's an AVGCart, so you have an obsession with owning multi-carts?

[woj]

2 hours ago, TZJB said:

You probably really only need one cartridge emulator if it's an AVGCart, so you have an obsession with owning multi-carts?

I hope not, for the actual multi-carts it was first SIDE2 and then later AVGCart (mostly for the reason of "native" support for SIO, SDriveMax is great, but a bit cumbersome to use, neither I like the touch screen, nor running the SDrive application and rebooting every time). All others are the aftermath of working on Popeye and/or doing soldering project.