Atari Schematics Lost?

[+kheller2]

@Dropcheck  How are things going?

[Dropcheck]

Sorry for the silence.  Kinda skidded to a stop until the XF551 pcb issue gets sorted out.  I'm hoping that the new pcbs that came in today will fix that problem.  After assembly and testing and re-posting, should be back on track in a week or two. 

[Dropcheck]

Update: 

 

After the experience with the XF551_B I decided to recheck my 1450XL schematics for the current prototype and discovered some errors against the loaned board truth.  And I'm not finished yet.  Some are mine.  Others appear to be coming from the original batch of schematics.  Not sure if Atari engineers made some changes on the loaned board version and they didn't get into the version of schematics I have or what.  So assembly of the prototype has stopped.  It isn't going to work anyway. 

 

Not a complete waste of time and money.  I was going to have to check part fit and make some footprint changes anyway.  For now I am focusing on trying to verify my schematic against the board truth.  All several thousand part holes.......🙃

[reifsnyderb]

13 minutes ago, Dropcheck said:

Update: 

 

After the experience with the XF551_B I decided to recheck my 1450XL schematics for the current prototype and discovered some errors against the loaned board truth.  And I'm not finished yet.  Some are mine.  Others appear to be coming from the original batch of schematics.  Not sure if Atari engineers made some changes on the loaned board version and they didn't get into the version of schematics I have or what.  So assembly of the prototype has stopped.  It isn't going to work anyway. 

 

Not a complete waste of time and money.  I was going to have to check part fit and make some footprint changes anyway.  For now I am focusing on trying to verify my schematic against the board truth.  All several thousand part holes.......🙃

Is there any way to hack the board to get it working?  Then you could, hopefully, only need 2 board revisions. 

[Dropcheck]

9 hours ago, reifsnyderb said:

Is there any way to hack the board to get it working?  Then you could, hopefully, only need 2 board revisions. 

Some major footprint revision has to be done in order to get a few critical parts installed and I'm at 20 rewires already with still have about 600 connections left to verify.  It might be possible, but I'm not holding my breath.  I'll know more in a few days. 

[Dropcheck]

I just ran into something that is freaking me out.  On the loaned 1450 board I have I have been rectifying my schematic with the board truth and currently working on the GND and +5V connections.  While verifying the power and GND connections on the ram chips, I discovered they seem to be swapped.  ie normally +5V on pin 16 and GND on pin 8 of each chip, but on this board it's opposite.  All the bypass caps are tied to pin 8 as well.  On the board I have the silkscreen chip image has them facing towards the left.   I don't remember anything being said about the silkscreen being backwards for the ram chips.  

 

Can anyone else verify this?

[tep392]

3 hours ago, Dropcheck said:

I just ran into something that is freaking me out.  On the loaned 1450 board I have I have been rectifying my schematic with the board truth and currently working on the GND and +5V connections.  While verifying the power and GND connections on the ram chips, I discovered they seem to be swapped.  ie normally +5V on pin 16 and GND on pin 8 of each chip, but on this board it's opposite.  All the bypass caps are tied to pin 8 as well.  On the board I have the silkscreen chip image has them facing towards the left.   I don't remember anything being said about the silkscreen being backwards for the ram chips.  

 

Can anyone else verify this?

I think for DRAM's that pin 16 is normally Vss, GND.

[AtariGeezer]

4 hours ago, Dropcheck said:

I just ran into something that is freaking me out.  On the loaned 1450 board I have I have been rectifying my schematic with the board truth and currently working on the GND and +5V connections.  While verifying the power and GND connections on the ram chips, I discovered they seem to be swapped.  ie normally +5V on pin 16 and GND on pin 8 of each chip, but on this board it's opposite.  All the bypass caps are tied to pin 8 as well.  On the board I have the silkscreen chip image has them facing towards the left.   I don't remember anything being said about the silkscreen being backwards for the ram chips.  

 

 

Can anyone else verify this?

 

Looking at this pic from @bob1200xl, you can see the ram chips orientation...  I just checked the the 1450XL schematics and they show pin 8 going to +5v and pin 16 to ground which is wrong...

 

[+mytek]

56 minutes ago, AtariGeezer said:

 

Looking at this pic from @bob1200xl, you can see the ram chips orientation...  I just checked the the 1450XL schematics and they show pin 8 going to +5v and pin 16 to ground which is wrong...

Wow that's gonna make for a sad day indeed. @Dropcheck I feel your pain when you get this news 😥

[Dropcheck]

Both 1400XL and 1450XL schematics I have for source have correctly identified pin 8 as GND and pin 16 as the power pin on all 8 ram chips.  Which is correct.  That reflects the ram chips data sheet as well.  At least for this the schematics appear to be fine. 

 

I am trying to reconcile my translated version of Atari's schematics with the actual connections on the 1450 bare board on loan from @kheller2.  That board was one of the few boards purchased from Best Electronics about 10 years ago now.  On the back there is a part # of CO24355-001 REV X2.  

This is the closeup picture of the ram chip area. 

 

 

As you can see the chips are orientated so that pin 8 is to the right bottom and pin 16 is left top on all ram chips according to the silkscreen.  You can clearly see the bypass caps all connect to pin 8 of each chip.  Right away this is not looking right.  Then I ohm out the GND(pin 8?) and Power (16 ?)pins on all ram chips.  I can confirm that pin 8 is going to +5V and pin 16 is going to GND.  The only way I can square this is that the silkscreen is flipped by a -90 degrees.

 

Does anyone else have this board and can confirm this? 

[_The Doctor__]

Just look at the chip data sheet, vcc is 8, vss is 16, just as the real board looks like, remember the cap goes between vcc and vss so it does not matter if it is near one pin or the other just so long as it is between the two.

Edited May 15 by _The Doctor__

[AtariGeezer]

31 minutes ago, Dropcheck said:

Both 1400XL and 1450XL schematics I have for source have correctly identified pin 8 as GND and pin 16 as the power pin on all 8 ram chips.  Which is correct.  That reflects the ram chips data sheet as well.  At least for this the schematics appear to be fine. 

 

I am trying to reconcile my translated version of Atari's schematics with the actual connections on the 1450 bare board on loan from @kheller2.  That board was one of the few boards purchased from Best Electronics about 10 years ago now.  On the back there is a part # of CO24355-001 REV X2.  

This is the closeup picture of the ram chip area. 

 

 

As you can see the chips are orientated so that pin 8 is to the right bottom and pin 16 is left top on all ram chips according to the silkscreen.  You can clearly see the bypass caps all connect to pin 8 of each chip.  Right away this is not looking right.  Then I ohm out the GND(pin 8?) and Power (16 ?)pins on all ram chips.  I can confirm that pin 8 is going to +5V and pin 16 is going to GND.  The only way I can square this is that the silkscreen is flipped by a -90 degrees.

 

Does anyone else have this board and can confirm this? 

I'll open up mine tonight and recheck, but I installed mine the way the silkscreens show.  It works, so the orientation has to be correct...

Doesn't pin 16 of the ram chips show continuity with the +5v with the LSI's???

 

[tep392]

6 minutes ago, _The Doctor__ said:

Just look at the chip data sheet, vcc is 8, vss is 16, just as the real board looks like, perhaps the schematic redo has a typo mixup.

Yes to this.  The only OKI M3764 datasheet I can find has a typo. Both pins 8 and 16 are marked Vcc.  Every other DRAM datasheet I look at has pin 16 as Vss.  Also, looking at the PCB pictured above, the address and data pins are connected correctly when the chip is oriented with the silkscreen.

[Dropcheck]

Okay I must have been using the wrong data sheet.  Looking at this datasheet, it puts GND on pin 16 and VCC on pin8.  So that would confirm the silkscreen is orientated correctly.  Jeeze, I've obviously tried to give myself a heart attack, long before I need to.  🤕 

TMS4164-datasheet-texas-instruments.pdf

[AtariGeezer]

I had that wrong too   The schematics I was looking at is correct

TMS4164-15 Dynamic Random Access Memory (DRAM) DIP IC

Edited May 15 by AtariGeezer

[+bob1200xl]

Yes... pin 8 and 16 are swapped on DRAM chips. Pin 16 is ground and pin 8 is +5v.

 

Bob

 

[+kheller2]

12 hours ago, Dropcheck said:

Okay I must have been using the wrong data sheet.  Looking at this datasheet, it puts GND on pin 16 and VCC on pin8.  So that would confirm the silkscreen is orientated correctly.  Jeeze, I've obviously tried to give myself a heart attack, long before I need to.  🤕 

TMS4164-datasheet-texas-instruments.pdf 462.43 kB · 2 downloads

Put down the motherboard, and slowly back away from the multimeter.  Go look outside for a few minutes.  Then enjoy burning the Atari schematics in a violent rage of anger.  

[ivop]

Here's weird 4164 datasheet with yet another pinout: https://www.farnell.com/datasheets/1905614.pdf

 

 

The datasheet talks about A0-A7, but where's A7 in the pinout?

[TZJB]

Samsung have it correct. 😀.

 

These were in my 800XLF, sorry to say 6 had expired but they were wired correctly.

 

[Dropcheck]

10 hours ago, kheller2 said:

Put down the motherboard, and slowly back away from the multimeter.  Go look outside for a few minutes.  Then enjoy burning the Atari schematics in a violent rage of anger.  

Unfortunately the schematics are digital pdfs.  So no bond fire possible.  🙃

 

But I did turn off the computer for awhile today.  Had a few medical appts.  No heart attack thank goodness.

[Dropcheck]

FYI:

 

While I cannot guarantee I have found all errors, I do feel somewhat confident in making a second attempt at a functioning board.  PCB order went out Monday.  One thing I did discover was that the proto board I have on loan has at least three different GND planes.  GND, Shield/Earth GND and another isolated GND around the Pokey chip and the joystick ports.  So at least for this version of the board I decided to keep Atari's routing.  In the future I may try combining all in one GND. 

 

I expect to receive this new batch sometime next week or so. 

[_The Doctor__]

If something is marked as shield depending where and what it is, even if you tie it to ground, please keep the note that it is a shield so we don't treat it as a ground making loops to other peripherals etc. Ground loops aren't fun.

[+mytek]

Well just my 2 cents - the Atari engineers didn't always know what they were doing, and as such made some design decisions that either made things worse or were totally unnecessary when moving from one model to the next.

 

If the 1450XL(D) board was a 4-layer type, then you are far better to dedicate one plane completely for the GND bus and another for the +5V. What does matter more than independent GND and +5V regions, is to simply isolate analog from digital circuits. So for instance if at all possible, all analog should be located grouped together and as far away from high-speed digital clocking circuits.

 

When I was developing the 576NUC+ I didn't adhere to this, and it bit me in the butt, taking 4 revisions of the board and the circuits therein to have clean analog video coming out of it. Yes the board was 4-layer, and yes it did have separate power planes, but because of the density, and perhaps some poor decisions on my part I had the analog Luma resister DAC nested in an area where it was surrounded by high speed clocking circuits. As I went through revisions on the board and slowly started to reroute clocking circuits farther away from the video, things began to clean up. However in the end it wasn't quite enough, and the circuit density made it impossible to completely isolate the DAC, so I took a different route and created a high current low impedance DAC that wasn't susceptible to being affected by close by digital clocking circuits. This was the best solution by far, and the video output of the 576NUC+ is now quite excellent.

 

So how was that new DAC created?

 

Well unlike the stock Atari 8-bit which uses a 4050 CMOS buffer having only 5 ma of drive capability, I used a 74HCT08 instead which has a 25 ma drive capability. This also served as an electronic switch for the Luma bits, so that the VGATE feature could work.

 

Anyway if you look at the resistors being used on the output of the 74HCT08 as part of creating the resistor DAC, you'll see that they are of a fairly low level of resistance. This low resistance absolutely requires the extra drive current being supplied by the 74HCT08 AND gates. So in order to move the needle in the DAC's summed analog output, you need that high drive. And since the digital noise floating around is comparatively low when trying to couple into the DAC, it can't compete and virtually disappears - being no longer a problem.

 

576NUC+ Video

 

 

 

Here's the stock Atari Luma DAC circuit. Notice the much higher resistance values in use (24 times higher than the 576NUC+). Digital noise can easily couple into the output signal in this design.

 

If I was recreating the 1450XL(D) you can bet I wouldn't necessarily stick to its original design in all regards, and instead would fix areas that needed improvement. However that's me, and I know the purists wouldn't like that 

[_The Doctor__]

nah the video circuit was plagued with the keystone cops routine, it is in the Atari tradition to change it with a model revision, so I'd go with MyTeks refinement. Though the 4050 does have another purpose than just video, no reason not to just add the 74HCT08 version he's shown and keep the 4050's other use separate. That way we have nice video and no upgrades get all uppity down the road. That's the spirit of a purist wanting the best.

Edited May 22 by _The Doctor__

[+mytek]

I think the UGV did it even better, by slightly increasing the low impedance resistor values, thus staying more inline with the total drive capability of the 74HCT08. And the overall video output spec is much closer to 1 V, whereas the 576NUC+ hits almost 1.2V. This design also replaces the two inverters that got used in the 576NUC+ with a smaller CD40107 dual open drain NAND gate package (no diode required). Works for both PAL and NTSC with automatic detection where the VGATE feature is involved.

 

And if you wish to stick mainly to THT devices, there is a THT equivalent for all except the FMS6400 Video Buffer/Combiner. That chip gives very fine quality composite video output as well as separate Luma and Chroma for S-Video, without creating any bleed issues between the signals. However it's obsolete, but not a problem to purchase as NOS from eBay or AliExpress, and really not subject to being faked like some other popular chips.

 

 

Realistically for an embedded design, switch SW1 and resistors R6 and R7 can be left out, thus yielding 1 V video output which after all is the standard.

 

I also think this design has a few dedicated fans that really praise the quality of the video output.

 

UGV VGATE Firmware Download