Another gargled video thread

[gmc]

I got a old Ti99 4A EU Version from a friend and trying to fixup a video issue. Read through quite a few threads and it looks like a common issue where if there is a problem with the 4116 chips then you could have some garbled video.

 

I've looked through the troubleshooting pics and in my case the whole screen looks faulty. 

 

It looks like its booting up and there is some resemble to what the video should look like.

 

Could it be I have more that one 4116 faulty? Maybe all of them?

 

I've checked the usual things like Vcc, no warm chips, clock signals etc.

 

I've seen some threads about possibly the 9929A being faulty.

 

Any ideas where I should start with, start replacing the 4116's (I've got a 4116 RAM tester so can can remove, test and replace them) or does this look  like the 9929A?

 

Edited December 12, 2021 by gmc

[Hans23]

This looks like a RAM problem to me.  Does the machine go into TI Basic or is it totally dead?  You might want to try piggy backing a 4116 onto each of the other chips to see if you can locate the problem without desoldering all of them.  

 

I find it rather useful to solder a momentary switch between RESET (TMS9900 pin 6) and ground to be able to quickly reset while probing around.

 

Keep us posted if you can't figure it out.

 

-Hans

[+arcadeshopper]

Definitely ram

Sent from my Pixel 6 Pro using Tapatalk

[Stuart]

This bit is stuck at 1: xxxx x1xx. That looks like bit 5 on https://www.ninerpedia.org/index.php?title=Troubleshooting (which only shows bits stuck at 0).

[+mizapf]

Maybe I should prepare some stuck-at-1 pictures.

[gmc]

1 hour ago, Stuart said:

This bit is stuck at 1: xxxx x1xx. That looks like bit 5 on https://www.ninerpedia.org/index.php?title=Troubleshooting (which only shows bits stuck at 0).

Ahh brilliant. You guys are amazing.

 

Replaced the 4116 for bit 5 and its working perfectly.

Edited December 12, 2021 by gmc

[Stuart]

21 minutes ago, mizapf said:

Maybe I should prepare some stuck-at-1 pictures.

I was going to suggest it might be a good idea, now we've had a stuck-at-1 'in the field'. 

[+Lee Stewart]

2 hours ago, Stuart said:

I was going to suggest it might be a good idea, now we've had a stuck-at-1 'in the field'. 

 

Stuck-at-one did happen at least once before, not that long ago.

 

...lee

[+Lee Stewart]

I have written an fbForth 2.0 program to display the TI Title Screen with singly stuck-ON or stuck-OFF VRAM bits 0 – 7 at the user’s whim. I will post it tomorrow.

 

...lee

[GDMike]

That's cool

[+Lee Stewart]

19 hours ago, Lee Stewart said:

I have written an fbForth 2.0 program to display the TI Title Screen with singly stuck-ON or stuck-OFF VRAM bits 0 – 7 at the user’s whim. I will post it tomorrow.    ...lee

 

Here is the Blocks file (TI Files format) with the program for showing the TI Title Screen with one bit stuck as the user dictates. Initially, it shows a screen with all bits OK:

 

STUCKBIT

 

To use it with the fbForth 2.0:13 cartridge, copy it to DSK1 (or any disk or disk image) and type the following:

USEBFL DSK1.STUCKBIT   \ or DSKn, where n is its location
1 LOAD

For those interested in the code, see the spoiler below:

Spoiler

\ fbForth 2.0 Program to display TI Title Screen with VRAM stuck
\ bit of user's choosing. It is stored in blocks file, STUCKBIT,
\ in Blocks 1-5. The directions are located in Block 10 and are
\ shown below as they display on the screen. The lower 4 KiB of
\ VRAM as it should exist at power-up of the TI-99/4A are stored
\ in Blocks 11-14. The stuck bits are numbered from left to right,
\ i.e., MSb to LSb.
\ 
\ Directions display as:
\ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
\ \\ Stuck bit display program...            \\
\ \\                                         \\
\ \\ After program starts and TI Title       \\
\ \\ screen displays, press one of the       \\
\ \\ following keys to change how stuck      \\
\ \\ bits display:                           \\
\ \\                                         \\
\ \\     O (or o)...to show bit stuck ON     \\
\ \\     Z (or z)...to show bit stuck OFF    \\
\ \\     0-7........to set bit to show stuck \\
\ \\     <break>....to exit                  \\
\ \\                                         \\
\ \\ Each time a digit 0-7 is pressed,       \\
\ \\ the screen will re-display with the     \\
\ \\ indicated bit stuck on or off. The      \\
\ \\ program beeps when display complete.    \\
\ \\                ...tap <BREAK> to begin  \\
\ \\                                         \\
\ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
                                                                                 
HEX

: WITHIN  ( N LO HI -- TRUE|FALSE ) OVER - >R - R> U< ;

0000 CONSTANT ORFLAG    \ AND|OR flag..initially 0 for OR
0000 CONSTANT BITSTUCK  \ bit (0-7) that is stuck

\ Parameter Field of ANDOR is modified by SETOP based on ORFLAG
: ANDOR OR ;            \ synonym for OR|AND

: SETOP  \ set ANDOR based on ORFLAG
   ORFLAG IF    \ flag is TRUE for ANDing
      ' AND CFA ' ANDOR !  \ ANDOR will now do AND
   ELSE  \ flag is false for ORing
      ' OR CFA ' ANDOR !   \ ANDOR will now do OR
   THEN  ;

: BLKS2BUF   \ copy blocks 11-14 to >E000->EFFF
   E000           \ beginning buffer address
   0F 0B DO       \ blocks 11 - 14
      DUP         \ DUP current RAM address
      I BLOCK SWAP   \ src,dst in proper order
      0200 MOVE      \ copy current block to RAM buffer
      0400 +         \ next block
   LOOP
   DROP  ;           \ clean up stack

: BUF2VRAM   \ copy 4 KiB RAM to VRAM
   E000 0 1000 VMBW  ;  \ copy 4 KiB bytes from RAM to VRAM

: STUCKBIT
   SETOP                \ set ANDOR operation
   8080 BITSTUCK SRC    \ adjust stuck bit
   ORFLAG XOR           \ will toggle bits if ANDing
   F000 E000 DO   \ 4 KiB in byte-pairs
      DUP               \ DUP bit pattern
      I @ ANDOR         \ get 2 bytes and AND/OR with pattern
      I !               \ put modified byte-pair back
   2 +LOOP     \ next pair of bytes
   DROP  ;              \ DROP bit pattern

\ Show Title Screen with stuck bit
: SHOWSTUCKBIT   
   BLKS2BUF
   STUCKBIT
   BUF2VRAM  ;

\ Show normal Title Screen
: TITLESCRN
   GRAPHICS
   7 SCREEN
   BLKS2BUF
   BUF2VRAM  ;

: DIRECTIONS
   TEXT        \ Text mode
   0A LOAD     \ show directions
   BEGIN
      PAUSE
   UNTIL  ;    \ wait for <break> to exit

: DOIT
   VDPMDE @    \ push screen mode
   DIRECTIONS
   TITLESCRN
   BEGIN
      ?KEY
      CASE
         2 OF 1 ENDOF   \ break
         ASCII O OF 0000 ' ORFLAG ! 0 ENDOF   \ stuck ON (1)
         ASCII o OF 0000 ' ORFLAG ! 0 ENDOF   \ stuck ON (1)
         ASCII Z OF FFFF ' ORFLAG ! 0 ENDOF   \ stuck OFF (0)
         ASCII z OF FFFF ' ORFLAG ! 0 ENDOF   \ stuck OFF (0)
         ELSEOF
            8375 C@              \ get key
            DUP 030 038 WITHIN   \ 0-7?
            IF
               07 AND            \ make an actual number
               ' BITSTUCK !      \ store in BITSTUCK
               SHOWSTUCKBIT BEEP \ show stuck bit & beep
            ELSE
               DROP              \ don't need key
            THEN
            0 ENDOF
      ENDCASE
   UNTIL
   VMODE  ;     \ restore screen mode

CR ." Type DOIT to start program." CR CR

 

 

Later, I will re-post this over in the fbForth thread for more Forth discussion.

 

...lee

[+Lee Stewart]

Here are the screen images of the TI Title Screen with single bits 0 – 7 stuck ON from the Forth program in the last post:

 

 

...lee

[+Lee Stewart]

Just for balance, here are the screen images of the TI Title Screen with single bits 0 – 7 stuck OFF from the Forth program two posts back:

 

 

...lee

[+arcadeshopper]

Thanks Lee!

Sent from my Pixel 6 Pro using Tapatalk

[Duewester]

Wow, what a great resource this community continues to be. 

I just purchased (ebay) a beige unit assembled 24th wk 1984. Looks the same as my other two older units except for exterior. So not a QI.

Anyway, i knew the unit was suspected dead and it turns out that video ram is indeed toasted (somewhere). From this thread and some other ones it looks like the Bit 5 is stuck issue (actually bit 1 stuck off but you get the idea).

So,....I've decided (since this is a T'sU unit) to go the upgrade route. I've ordered the adapter boards from the Arcade Shopper, the 4164s from a chip dealer (hope they are reputable) and a dip kit with a variety of sizes.

I have a FA18 and thought briefly about putting it in but decided that since I just have the one, to leave it in my working baseline unit.

My only concern, getting the old 4116's out without damaging them (they might come in handy someday).

Edited August 2, 2022 by Duewester

[matthew180]

15 hours ago, Duewester said:

My only concern, getting the old 4116's out without damaging them (they might come in handy someday).

Desoldering is at least twice as hard as soldering, and if you want to save the parts, it can be even harder.  Make sure you keep the parts and PCB clean with 99.9% alcohol (MG Chemicals: https://www.amazon.com/MG-Chemicals-Isopropyl-Alcohol-Cleaner/dp/B078C8JYVV ), use a good no-clean flux, use a quality LEADED solder (lower temperature, easier to work with, and all the retro system are leaded solder), and ideally use a powered vacuum tool.

 

Alternatively use ChipQuik to change the solder's melting point, then use low-temp hot air to remove the parts.  Once the solder is mixed with ChipQuik it does become a little harder to clean up, but it does stay liquid with ~150c temperature, and a Q-tip with alcohol can usually be used to wipe up the solder.  Once the chip is out, you can reflow each hole with clean solder, then clear each hole with solder-wick.  Again, always cleaning, always using flux to help the solder flow.

 

The old PCBs are fragile, and it is *very* easy to pull a pad or trace, so minimal heat and time on each joint is critical.  I usually clean all the joints top and bottom, reflow with new solder top and bottom, clean again, flux the bottom, and use my Hakko FR-301 desoldering tool (or ChipQuik before I got the FR-301).

 

Also, old PCBs don't always have solder mask on all traces (especially thin traces between pads), so be very careful you do not create any accidental short circuits.  This happened to me and it took a long time to find the problem.  See the photo, the short is only 0.002 inches wide and 0.016 inches long (was not visible to the naked eye).  That is a trace between VBB and CAS pins on a 99/4A DRAM.

 

Take it slow, control your heat, keep it clean, flux is your friend.

 

 

[Duewester]

Thank you @mathew180.

That's a very good checklist, Very similar (without as much detail on brand of tools) as what I had put together. I'm going to look into the Hakko FR-301. I'm used to just using the braided wire and my manual sucker. Been soldering and wire wrapping since the early 70's but, never been much about de-soldering. The newer surface mount stuff has been hard for me to get my skills improved.

 

Just a side note.

 I was comparing the new board with one of my older boards and the quality of construction was better (IMHO) on the older model than the newer model. The new(er) board has several chips where the solder is good on the back side but not present on the component side.

 

Again, thank you for the response and the advice. I'm always open to advice.

[Stuart]

As 4116s are neither rare nor expensive, better to sacrifice the chips to avoid damaging the PCB?

[AwkwardPotato]

To second what Stuart said, I think you may find salvaging 4116s to not be worthwhile when they can be found inexpensively online. Removing the DRAM on these boards with solder wick/a manual vacuum tool is a right pain in the neck; the less desoldering you have to do with these, the better. The joints for the power pins especially sink a lot of heat, and it's easy to accidentally destroy the through-hole plating. An oft-recommended approach for the DRAM is clipping each of the IC legs and pulling them out one-by-one with the soldering iron.

[GDMike]

Try cutting the chip pins and solder onto those existing pins.

[Duewester]

@Stuart, @AwkwardPotato, You have excellent points as do you @GDMike.

I was not looking forward to unsoldering the chips and I wasn't aware of the availability status of the 4116 chips.

It's snipping time.

Thanks for the advice. Have a great day and be safe.

[Duewester]

Update on my Bit 1 off issue.

Boards and chips have arrived.

Soldered headers and sockets to boards.

issue arises - boards are too wide to go in side by side on board.

stuck some headers in to see if that will give me some clearance. Nothing soldered yet and old 4116's still in place. 

Picture illustrates what I'm dealing with.

[Stuart]

48 minutes ago, Duewester said:

Update on my Bit 1 off issue.

Boards and chips have arrived.

Soldered headers and sockets to boards.

issue arises - boards are too wide to go in side by side on board.

stuck some headers in to see if that will give me some clearance. Nothing soldered yet and old 4116's still in place. 

Picture illustrates what I'm dealing with.

That looks like an awful lot of work considering that a single known IC is defective and a direct replacement is easily obtainable ... ;-/

(You're not a C64 user come to tease us? )

Edited August 17, 2022 by Stuart

[Duewester]

23 minutes ago, Stuart said:

That looks like an awful lot of work considering that a single known IC is defective and a direct replacement is easily obtainable ... ;-/

(You're not a C64 user come to tease us? )

Nope not a C64 or any other C? user. Went from TRS 80 Model 1 to Ti994a back in 83'.

It is an aweful lot of work. I'm gonna replace just the one defective (suspected) chip first and then see how it looks.

I have a electric solder sucker on order. My manual one isn't cutting it for the full job. I can make it work for this one chip though (I hope).

[Duewester]

One week later and finally replaced the bit 1 chip with new 4164 adapter and chip. Had to cut a hole in the rf shield.

Well good news is, the screen looks awesome. Even crisper than my other units.

The bad news, the keyboard doesn't work except the number 6 key. It's got one of the cheaper membrane kbs. Found several membrane areas pinched. Separated the layers and found the top to bottom junction connection trash. Might try to hard solder the junction. Can't hurt to try. In the meantime I'll look for a replacement kb.