Another (new) 130XE boots to red screen

[woj]

I think I should show this for future generations of Atari modders. I socketed the ANTIC on both computers (you'd now from another topic I am getting ready for a VBXE :D). I now perfected desoldering, feel 100% confident with it and no mess-ups of any kind, just time needed. But for those all of you that say "I socketed chip XYZ and nothing works" ;), below the picture of how much crap was left after I soldered the socket in. This is what I scraped off with a pick, AFTER it was cleaned twice with IPA and believe me, it looked clean. There was a giveaway, I watched the solder flow very carefully and I could see the flux leaking out. Undoubtedly, my soldering tin is not the best quality, still...  

 

The other thing - precision sockets! Precision sockets my ass, both ANTICs had contact / seating problems at one point or another. On the 800XL it was fine first, stopped working after flipping the board several times for assembly, had to pushed in again (and it was sitting tight). On the 130XE it did not work on the first power up, again had to be pushed in even harder while sitting in already quite well.

[_The Doctor__]

I don't know what you are using but I hope it isn't etching or acid flux solder.

make sure the room and board are warm for cleaning time.

(((( that chip has some very dirty crusty legs !!!! just zooming in I cringed... clean them with deoxit or a fiber pen, I don't blame the socket at all ))))

dual swipe sockets aren't bad, single swipe sockets should never be used.

 

looks like your using sub par flux / solder and are not getting the chips clean and smooth, it's making your life tougher than it needs to be. Get some good lead based solder for retro electronics work and good flux, make sure you have nice high percentage IPA and none of the rubbing types, make sure you get a good chem cleaner and fiber pen.

 

Take a magnifier and look at your work during and after, in none then phone camera etc macro and then blow up.

I wouldn't be using a metal pick on it that's for sure.

Edited January 29, 2023 by _The Doctor__

[woj]

I did take note of this last feedback  just was not really sure what to reply. In any case, I upgraded my soldering "chemistry" recently, and also made commitment to start using lead-free solder where applicable (= new boards), especially that in my country of residence getting lead based solder is practically impossible. I practiced a bit on this and that and now I am ready for my next mini-project, which is to solder and test the two stereo boards below. I am also moving here from my other thread: 

 

Both boards are essentially a redux version of mytek's stereo board without the PS/2 PIC chip, but have two audio outputs (to kill the cable maze inside the case for my DIN-13 connector and mini-jack one) and the LED is embedded on the board.

 

The left one is quickly hacked together with auto-routing and TH components to check if all this works (and to practice the lead-free soldering a bit more). The right one is the actual thing which is my attempt to go SMD and see how far I can go with the soldering iron on this one. Will take me a couple of days until I get time to fire up the iron, so for now just the bare boards that arrived today:

 

[_The Doctor__]

stay leaded for old lsi and such old style chips, newer stuff is made for the heat so rohs all the way

[woj]

9 hours ago, _The Doctor__ said:

stay leaded for old lsi and such old style chips, newer stuff is made for the heat so rohs all the way

Apart from anything with plastic, so far had horrible experience soldering header strips, the prolonged heat I had to apply to get the tin flowing the right way made the pins slide out of the plastic. Since then I got the temperature setting in better / lower regions, but I still find it a precise game of heat application timing. Apart from that and with thin enough solder wire everything goes very smooth. 

[_The Doctor__]

newer solder that is lead free has a higher melting point temperature, older components, pcb, and connectors can't take the heat as much, lower quality plastics and such melt more easily than higher quality items. Stored heat in the iron is a bit of the timing game. Too low and you have a terrible time joints and damage from longer application of heat melt times, too high and you get almost the same terrible time and damage. So yes heat and application time is a skill/knack that you acquire over time as you learn the correct balance of proper heat and application time, if you do it enough you will develop a rhythm to your work. Ambient room, pcb, and component temperature can change the equation a bit.

Edited February 10, 2023 by _The Doctor__

[woj]

The through-hole board is soldered up and works! It is also now more tidy under the case. Not 100% happy with the soldering (anything connected to a ground plane is a total bitch to solder with lead-free, but I guess everyone know this), but since it works and is generally temporary till I fix the SMD one, I leave it as is.

[J D Potter]

I've got an old 130XE (16 x 4164, NTSC) that was working just fine for months (after it had a >30 year hibernation in my garage). I built my own FujiNet, which worked fine with it for about two days. For no apparent reason it appeared to die on the spot, and only produces the reddish-brown screen described earlier in your posts.

 

I'm an digital electronics engineer with 40+ years of debugging experience, a long time 'scope jockey, and I've got a small collection of spare parts (DRAM, CPU, Antic, GTIA, PIA, MMU) that I believe are good. Since some of you on this topic have already gone through the ringer on problems like this, I'd appreciate a little guidance:

1. The 130XE is generating video: I can synch on the CSYNC (U20 pin 14 or 15) and examine the LUM and COLOR signals, which don't seem to be doing much.
2. All chips are receiving power. No "blowing up" caps exist. All connections to the Clock or Φ2 pins of all chips are live. No chips appear to be constantly holding down RST, HALT, R/W, RAS, or CAS.
3. The color adjust trimpot (R38) definitely affects the output (full-screen dark red, or other color selected), so I'm guessing GTIA isn't dead.
4. I can see Address lines (everywhere) toggling for the most part during the entire video frame.
5. Data lines out of the DRAM (all 16 chips) seem to be toggling for a few seconds after power-up, have no activity except for the 3 H lines before VSync.
6. The CAS signal is high (inactive, after the first 3 seconds of power) during the active video, and until this beginning of the vertical refresh (I'm guessing it's the VBI service via Antic?) when some low pulses regularly occur.
7. Signals on the SIO connector show no signs of requesting boot data from the floppy (or FujiNet) device upon power-up.
8. The CPU feels a little warm, but I can keep my finger on it as long as I care to. Nothing else on the circuit board feels very warm at all.

I'm about ready to buy a de-soldering iron (I've tried the unpowered solder-sucker devices, and I'm not impressed), buy all the sockets I'd need to replace my suspected bad parts, and have at it. Can anyone offer some further debugging steps?

 

Your 8-bit friend,

JDPotter

[Beeblebrox]

@J D PotterAre they MT branded ram chips perchance?   Given sudden death and red screen it still could be a ram chip failing, (even though you have tested drams in situ).

 

Or maybe OS rom too? Possibly a failing ANTIC also. I'd check OS rom and RAM in the first instance.

 

Do you have a Star Raiders cart you can power this up with BTW?  It has diagnostic facilities and can bypass the OS I believe. 

Edited February 11, 2023 by Beeblebrox

[_The Doctor__]

https://8bithardware.wixsite.com/website/post/atari-8-bit-400-800-xl-xl-pictorial-fault-guide

[J D Potter]

(Beeblebrox) I've got a configuration I've read is typical: a single (U16) Toshiba TMM4164AP-20, and 15 MT4264-20. For replacements, I've got nine Samsung KM4164A-15 and four MT4264-15's. I'm trying to test these outside the 130XE before I declare them good or bad. Nothing is socketed on this board

 

My OS ROM is C061598B-29. I found I have a spare that reads C061598B-29; I gather it's from another generation of Atari 8-bit?

 

I didn't save any carts, besides one I removed the ROMs for my own use (the one and only cartridge version of Maze of Agdagon https://wowroms.com/en/roms/atari-800/maze-of-agdagon-atr/70904.html). I did try plugging that in; no change.

 

(The Doctor) Yes, I had found the Pictorial Fault Guide. I had previously suspected C1 because it looked ready to burst open (I had replaced it a month or two before the 130XE exhibited any flaws). Maybe I can test all caps and replace as a first step.

 

Thanks.

[Beeblebrox]

Yeah, stock XE pcb's are jot socketed. XE boards generally aren't as robust and well made as the earlier XL lines. Far easier to lift a trace/via when working on an XE board. 

 

The 28pin OS ROM chip you mention is for the 65XE, 130XE, 800XE, 600XL and 800XL and can be used across those machines so you should be fine. 

 

Mt ram chips are generally considered unreliable and given you have some in the mix my money is on the ram and then the os rom. 

 

Generally you rarely have to recap an A8. I've fixed many A8's now and nearly always it's one or more failing IC's, a passive component such as a transistor, and/or socket/trace issues. 

 

I am not an expert though, plenty more on AA who have far more technical experience, (yourself included), but I am going on familiarity with fixing these wonderful machines. 

 

Check the Mt ram chips in the first instance after removing them. Or perhaps go for socketing /swapping out the os rom chip. 

 

 

 

 

[woj]

Oh, a thread hijack

 

Here is one for you, I tag @flashjazzcat and @mytek, as the first one knows what the U1MB software does here, the second one designed the circuit  

 

I soldered up my SMD version of the redux stereo board, it works, stereo is there, sound is great (so the SMD transistor equivalents of PN2222 I found in my local shop surely work), Yoomp says "stereo detected", seemed perfect.

 

What does not work is U1MB detection of the stereo. I had it on from the previous board, so it starts up with stereo on, the LED flashes the same way on BIOS boot indicating U1MB is checking for the presence of the stereo board, but it grays out the Enable/Disable option for Dual Pokey! So it does not see it, but yet it is still happy to keep the stereo on and tell me that it is enabled, though not giving me a chance to change the setting. (My other through-hole board works just fine here, and it is based on the same schematics).

 

Now, I did triple check everything, I suspected that I mounted the SMD diode in the opposite direction, I measured, it does not seem to be the case (but I did not try to reverse it), I also suspected that the LED controlling transistor is bad leaking current back into M0 line, removed that, no dice, replaced it with a new one, still the same. I checked the on/off voltages on M0 on both boards, they are consistent with each other (0.1V / 3.3V). 

 

I looked at everything and my own conclusion is that the only thing that is different on this board as opposed to my through-hole one or mytek's original, is that I used some generic (CMOS I guess) 4066B chip (due to lack of other options in my shop) instead if the HCT (TTL, was it?) one. Could that be it? (for the moment I have no chance of verifying this claim).

 

Or my soldering went bad somewhere, that will take me ages to trace...

 

 

Edited February 11, 2023 by woj

[Beeblebrox]

@woj can you double check you have conformed connections for a4 and gtia Bell?

A4 pin 13 of cpu and gtia Bell pin 15 of gtia. 

 

Maybe also recheck the mo to mo connection. 

[woj]

Just now, Beeblebrox said:

@woj can you double check you have conformed connections for a4 and gtia Bell?

A4 pin 13 of cpu and gtia Bell pin 15 of gtia. 

 

Maybe also recheck the mo to mo connection. 

Checked all that, GTIA clicks work fine, if A4 was badly connected there would be no stereo, but all stereo programs work just fine. It's just the U1MB that does not detect it (and yes, I triple checked M0 connection end to end, it's solid), but at the same time keeps it on happily (or off, but to turn it off I need to mount the other board, change the setting, and then remount the board).

 

I look at the schematics and the only thing I can still see is that this 4066B can have different timings...

[+mytek]

37 minutes ago, woj said:

I tag @flashjazzcat and @mytek, as the first one knows what the U1MB software does here, the second one designed the circuit  

Well to be truthful I haven't been following this thread, and that's the first time I'm seeing your Stereo Board (looks very nice 👍). If that is indeed based upon the one that I built, then all I can offer at this point is to double check it to my original schematic to insure that the logic and switching are exactly the same on your creation, and that all the components match up (resistor values, ect.). You might also want to double check the stereo detection by putting my board in (assuming you have one) and see if that works properly. if it does, then it becomes a process of elimination between that and yours.

 

On a side note: I know there have been difficulties in the past with reliably detecting that a stereo board is present, but I thought that all got shaken out a long time ago.

 

Sorry I can't be of more help, but I do hope you can get it figured out

[+mytek]

30 minutes ago, woj said:

I look at the schematics and the only thing I can still see is that this 4066B can have different timings...

I just caught this statement. You must use a high speed version of the 4066 (example: CD74HCT4066QM96Q1). This is not required for the analog switching aspects, but is absolutely essential for the parts that are being used for sensing address A4 (Left & Right Pokey selection). So what exactly did you use for the 4066?

[woj]

@mytek Seems to be some generic CD4066B from a local shop, that's the best information I can give you.

 

Your board works just fine, so does mine that is through hole and uses the same DIP 74HCT4066 (it is essentially identical to yours, just no PS/2 stuff), it only slid away when I went SMD (logic is identical) and this suspicious 4066.

 

Well, it seems I need to source proper SMD 4066 😉

[+mytek]

5 minutes ago, woj said:

Well, it seems I need to source proper SMD 4066 😉

Yep that would likely solve all your problems.

 

Actually I'm rather surprised it worked at all with that standard speed CMOS device in use. Although it was probably on the edge of not working, thus why the U1MB stereo detection routines might have failed to see it.

[woj]

On 2/11/2023 at 11:58 PM, mytek said:

Yep that would likely solve all your problems.

 

Actually I'm rather surprised it worked at all with that standard speed CMOS device in use. Although it was probably on the edge of not working, thus why the U1MB stereo detection routines might have failed to see it.

Yep, that did help and it now works fine with a TI 20/30ns 4066 chip. I will need to solder up a whole new board though (I anticipated that), because I made huge mess re-soldering the chip twice. I made a (deliberate) choice of trying to solder it with a soldering paste and hot gun instead, trying it for the first time. It did not work for the first time, got a solder blob under the chip, had to remove it again, in the process lifted one trace slightly, and then it took me three attempts to bodge the chip in place and somewhat make contact on all the pads to make sure that it all indeed works. So it does, but the chip is hanging by the threads, literally, and this needs to be re-done cleanly from scratch.

 

My lesson from paste/gun SMD soldering is - for now I prefer the iron, without a stencil and a hot plate this is totally pointless, not really quicker, and much messier in my case. 

 

Not showing the pictures (did not make any) of the mess I made  

[TZJB]

On 2/11/2023 at 6:57 PM, J D Potter said:

(Beeblebrox) I've got a configuration I've read is typical: a single (U16) Toshiba TMM4164AP-20, and 15 MT4264-20. For replacements, I've got nine Samsung KM4164A-15 and four MT4264-15's. I'm trying to test these outside the 130XE before I declare them good or bad. Nothing is socketed on this board

 

My OS ROM is C061598B-29. I found I have a spare that reads C061598B-29; I gather it's from another generation of Atari 8-bit?

 

I didn't save any carts, besides one I removed the ROMs for my own use (the one and only cartridge version of Maze of Agdagon https://wowroms.com/en/roms/atari-800/maze-of-agdagon-atr/70904.html). I did try plugging that in; no change.

 

(The Doctor) Yes, I had found the Pictorial Fault Guide. I had previously suspected C1 because it looked ready to burst open (I had replaced it a month or two before the 130XE exhibited any flaws). Maybe I can test all caps and replace as a first step.

 

Thanks.

The U16 4164 DRAM is often a different (better) manufacturer. The Micron Technology 4164 DRAM installed in the XE range often fail and complete replacement is generally recommended. I normally use 41256, if you can get them cheaper, and especially for the second bank as this can the be used as 256K expanded DRAM with a simple logic chip addition for a 320XE!

 

In your case, you have enough Samsung 4164 for the primary 64K bank so I would look to upgrade the secondary bank with 41256 DRAM, as you probably need to change that too, and recommend that you install a Peterson 320K upgrade after you have confidence that all the DRAM work at 128K.

 

The OS ROM can be replaced with a 16K EPROM once it is socketed, but your fault symptom is most likely to be caused by DRAM failure at this point. I have had success in safely removing the DRAM chips with a sprung solder sucker, but have upgraded more recently to a desoldering station as I had lots to do.

 

I have had some faulty capacitors lately with the main 470 uF normally bulging. There were also two in a 800XLF that I had to change, and I unusually had one go in an 800XL also, but only had a 680 uF replacement with the correct pitch. It obviously works fine.

 

Best of luck!

[+mytek]

3 hours ago, woj said:

My lesson from paste/gun SMD soldering is - for now I prefer the iron, without a stencil and a hot plate this is totally pointless, not really quicker, and much messier in my case.

Yeah my feelings also. I have a special heat gun with various tips, but I never found it to be better than just using a soldering iron with a small tip. Before setting the SMD chip in place, I use a tiny bit of liquid flux applied to the pads with one of those flux pens. I also try to stick with SOIC devices in my designs which are the largest SMDs.

[woj]

Done the thing right this time, all works, and I can close this chapter. In the end I got four different stereo boards, one mytek's, one own through-hole design, and two copies of the SMD own design. I went a bit overboard I think just to get the stereo in there  Still on a lookout for a PokeyMax, as I am short one chip, my 800XL sits without one  

 

Now I can go back to my Popeye game writing  

[woj]

Out for a bit of advice / feedback. My long awaited Sophia 2 is waiting at home for my return. It will go into my 800XL that so far has no upgrades to speak of (just UAV and some cabling for U1MB eventually) and the RF modulator already removed.

 

Now, I know there are two possible ways to mount the DVI connector, I've seen both methods on @flashjazzcat's videos, one (a) with the connector bolted to the case (rect. hole plus two drill holes) and one (b) with the connector PCB soldered to the modulator vias and one rather large rect. hole in the case.

 

For (a) I know it needs to come off every time I'd need to do anything on the board, while it will be easier to make it beautiful, but also sticking out more. For (b) it's more faffing about to position it right and make it look nice, while it would be a permanent installation and not sticking out of the case.

 

Anything else I should know / consider? What are you preferred ways of doing this? The PBI blanking plate method is out of question for me BTW. At the same time, I do not exclude the possibility of having the VBXE in this one too at some point, and I know that it will be difficult to fit the DIN13 socket regardless of going (a) or (b) way.

[flashjazzcat]

Note that I don't present methods A and B as equivalent. Once it dawned on me that Simius had designed the DVI connector daughterboard such that it mounted neatly in the vias vacated by the RF modulator, it was method B all the way. The only reason method A appears on any of my videos at all is because of my ignorance regarding the design at the time.

 

I would suggest that method A is actually harder to 'make beautiful' than method B anyway, since all method B requires is a perfectly rectangular aperture and nothing else. Attaching the connector directly to the case, on the other hand, necessitated (when I did it, at least) a rectangle with angled sides and two carefully positioned drill-holes.

 

Sophia and VBXE in the same 800XL should present few if any logistical issues, anyway, since after fitting the DVI connector in place of the RF modulator, there is still plently of space between it and the legacy monitor jack for a DIN13 connector (all you'll have to do there is situate the DIN13 so that it avoids traces on the underside of the board or patch any traces it does happen to collide with, and cut a perfectly circular hole in the back of the case). Not that DIN13 is mandatory by any means (it simply makes sense from an aesthetic perspective and happens to propagate something of a pinout standard given the existence of at least one commercially available RGB video cable); one alternative video jack solution currently the subject of absolutely relentless self-promotion in almost every conversational sphere and whose name I dare not utter lest it functions as an incantation might also work were it not for the fact it also contends for the space vacated by the RF modulator.

 

My solution for VBXE and Sophia in the same machine turned out to be my 1088XEL, but largely because I wanted to fill the pre-existing DVI-connector-shaped hole in the back of the case, having already fitted VBXE.

Edited March 19, 2023 by flashjazzcat