The Worlds Smallest Atari 8-Bit?

[+mytek]

24 minutes ago, MisterOz said:

Just read the first page again, missed that - oops.. You've inspired me to try and create my own A8 mini. This one is going to be called the micro. It'll only be for my own use at this point. I've got the layout completed and have started connecting traces. As I'm only proficient in expresspcb/schematic, manually placing these traces and vias is becoming quite the exercise in patience. I've already re-ordered board components about 8 times as I stumble upon different components that I want to try out. I'm probably a bit optimistic in how tight I have everything spaced, as I'm sure you've encountered.  Thanks again for your other designs - and of course to all the collaborators. They've been very fun and rewarding to build so far.  

You'll probably want to wait for the final schematics (which will get released soon) - there's been changes .

 

[+mytek]

Picking up where I left off in that previous post...

 

In the first 3 prototype versions, I had been feeding the csync directly out of the GTIA into the luma DAC circuit via a blocking diode. Now although this appeared to be working, it had an unfortunate side effect of overheating some of the internal circuits of the GTIA, thus causing the eventual destruction of the GTIA. Sometimes this took a day of continuous running, and other times it would kill the GTIA in a matter of minutes. So buffering this signal was needed.

 

For the V1.1 prototypes that the beta testers were building, this buffering came in the form of gluing down a SN74LS07 SMD, and then doing some point to point wiring from one of its 6 buffers. Since the output was open drain, the blocking diode was not required, but a 75 ohm limiting resistor was needed on it's output feeding into the resistor DAC circuit.

 

For the next board version (V1.2) the fix was incorporated using two inverters from a SN74HCT14 schmitt trigger hex inverter, with this same chip to be shared by the crystal oscillator circuit. However with this not being an open drain device required the reinstatement of the blocking diode, and because of the voltage drop of the diode only a small resistance was needed before tying it into the DAC output circuit.

 

Here is the Schematic of the new video circuit.

Along with this change, the former SOT-23-6 analog switch was replaced by a SOIC equivalent, thus making all SMD components into SOIC packages which are much easier to handle and solder due to their larger size.

 

Another change was a suggestion by @tf_hh to replace the former DIN-5 audio/video jack with a DIN-8 which still allows a DIN-5 plug to be used, while giving us 3 additional pins for routing the audio signals out of a stereo capable upgrade (this would require a DIN-8 plug and custom cable).

 

[+DrVenkman]

4 minutes ago, mytek said:

Now although this appeared to be working, it had an unfortunate side effect of overheating some of the internal circuits of the GTIA, causing the eventual destruction of the GTIA. Sometimes this took a day of continuous running, and other times it would kill the GTIA in a matter of minutes.

Well, my board - modified with the buffer chip + resistor fix - has now run about 20 hours total without any GTIA issues. Over the four-day holiday weekend I'll do some 12 hour-plus runs as well.  

[Mr Robot]

I lost three GTIA's while we were working that out and then the beta board and the dead chips got stolen from a mailbox so I totally lost them. 

 

If you ever see a yellow NUC+ for sale not from TBA, you'll know where it came from and that it kills GTIA chips

 

[+Stephen]

Just now, Mr Robot said:

I lost three GTIA's while we were working that out and then the beta board and the dead chips got stolen from a mailbox so I totally lost them. 

 

If you ever see a yellow NUC+ for sale not from TBA, you'll know where it came from and that it kills GTIA chips

 

That sucks.

[+mytek]

22 minutes ago, Mr Robot said:

I lost three GTIA's while we were working that out and then the beta board and the dead chips got stolen from a mailbox so I totally lost them. 

 

If you ever see a yellow NUC+ for sale not from TBA, you'll know where it came from and that it kills GTIA chips

The saddest part about this, is that the thief probably chucked it into the garbage after he opened the box and saw what it was. That's nearly $125 down the tubes. Total piece of shit .

 

Anyway we got Mr Robot fixed up with another board, so his development can go on .

[+slx]

5 hours ago, mytek said:

Another change was a suggestion by @tf_hh to replace the former DIN-5 audio/video jack with a DIN-8 which still allows a DIN-5 plug to be used, while giving us 3 additional pins for routing the audio signals out of a stereo capable upgrade (this would require a DIN-8 plug and custom cable).

 

Mostly amazed about what I don't know and can't do when reading this....

 

Would the 8-pin jack introduce any additional danger of e.g. a Commodore PSU being plugged into the video port? Other than that I think it's a great idea as it would solve the problem of how to elegantly route stereo audio out.

[+mytek]

5 hours ago, slx said:

Would the 8-pin jack introduce any additional danger of e.g. a Commodore PSU being plugged into the video port?

Potentially yes. However no where near as bad as if you plugged that C64 PSU into the power jack of an XL or XE . Worst case scenario is that you might blow out the video buffer chip on the NUC, and whatever was connected to pins 6 and 7 on the stereo output side of things (those pins are 9 VAC on the C64). There is even potential of getting into trouble if you plug in your XL/XE PSU into the NUC's DIN-8 jack, although the voltage is only 5 VDC, and it has to transition through 75 ohm resistors in order to get to the buffer. Chances are that this would not be destructive because of the current limiting imposed by those resistors. But definitely something I'll test for, since there is a high probability of someone having one of these Atari PSUs close to their NUC.

 

[+MrFish]

Why are people trying to plug C64 power supplies into Ataris at all in the first place? I mean, if people just grab anything lying around and plug it into their machine (especially the video port), they deserve to have it fried.

Edited November 25, 2020 by MrFish

[+mytek]

2 hours ago, MrFish said:

Why are people trying to plug C64 power supplies into Ataris at all in the first place? I mean, if people just grab anything lying around and plug it into their machine (especially the video port), they deserve to have it fried.

The simplest answer would be "Stupid is as Stupid Does" according to Forrest Gump.

[+slx]

5 hours ago, MrFish said:

Why are people trying to plug C64 power supplies into Ataris at all in the first place? I mean, if people just grab anything lying around and plug it into their machine (especially the video port), they deserve to have it fried.

Probably people not really in the know who find this stuff in an attic and think "if it was stored close together and fits together it might work together", not contradicting mytek's explanation.

[+MrFish]

1 hour ago, slx said:

Probably people not really in the know who find this stuff in an attic and think "if it was stored close together and fits together it might work together", not contradicting mytek's explanation.

 

Yeah, still pretty stupid, but probably the kindest explanation one could give for such folly.

 

[+Stephen]

2 hours ago, MrFish said:

 

Yeah, still pretty stupid, but probably the kindest explanation one could give for such folly.

 

Just because it fits doesn't mean you should stick it in there.  I'll get my coat.

[+Stephen]

Tough repair with the board populated!  Took me an hour but I am VERY happy with how it turned out.  Ignore the chip in the socket (from cutting a DIP in half to remove the support bars) and the hot melt glue "hairs" - they were cleaned when I removed any flux.

 

For scale:

[+DrVenkman]

9 minutes ago, Stephen said:

Tough repair with the board populated!

Yep. Took me an hour too. Definitely would be easier on an unpopulated board. Of course, the next version of the board incorporates the fix as part of the base layout. 

 

On my board, I used a tiny piece of double-sided foam tape to mount the LS07 buffer - the only hot glue I have around here doesn’t seem to adhere to PCB’s very well. I also left the bypass cap installed and soldered the power and ground leads to the legs of the cap. 

[+DrVenkman]

This is the first of @mytek‘s bespoke Atari A8-based machines which don’t incorporate one of Byan’s UAV boards for video, instead designing the video circuitry from scratch.

 

At least Y/c output through my usual C1702 monitor is completely on par with any other Atari machine. 
 

 

[CPUWIZ]

On 11/25/2020 at 7:07 PM, Stephen said:

Just because it fits doesn't mean you should stick it in there.

 

That's what she said.

[ajacocks]

1 hour ago, DrVenkman said:

This is the first of @mytek‘s bespoke Atari A8-based machines which don’t incorporate one of Byan’s UAV boards for video, instead designing the video circuitry from scratch.

 

At least Y/c output through my usual C1702 monitor is completely on par with any other Atari machine. 
 

 

Looks fantastic! I’m really excited for this.

 

- Alex

[+Stephen]

Good news - my 576 is working perfectly.  There were 2 things preventing it from booting.  1 - the donor CPU I used was defective.  2 - the 64kB SRAM I received was defective.  At least I feel better now about my assembly skills.  Also - this is the first time I have ever diagnosed faulty hardware by myself.  Probing the CPU with my o-scope I found the bad CPU and the next thing I tried was testing the SRAM in my new TL866Plus.  Boy did I get lucky.  The original design calls for the SRAM to be soldered in, but I just cannot bring myself to solder a 32-pin DIP into a 4-layer PCB so I socketed it as well.

 

I must say - the video output on this is amazing.  Definitely on par with the UAV so great work there.  I have so far only tested on my 1084S which needs some work.  I can't way to fire up the various Sony PVMs.  I'll take my SDriveMaxx to the basement tomorrow and finally do some solid testing.

[MessalineApghar]

On 11/8/2020 at 10:50 AM, Mr Robot said:

To be frank, some of the support requests for the XEL/XLD builds were pretty challenging, not everyone taking the project on is really qualified to do so and that has made supporting them pretty hard work. This build includes SMD devices for the first time and everyone on the team agrees that providing a kit would create a support nightmare for us all. 

 

I have argued that we could make the blank PCB available with just a BOM and a notification there there will be no support for it, that if you're not 100% certain you can do it, you shouldn't start. Something along the lines of "This board has been well tested and is known working. Any issues you may have will be with the components or your assembly, once you have thoroughly checked your assembly for issues, please contact your component supplier for support" but even that isn't strong enough for some of the team so it's not likely that position will change for at least a couple of years.

what about the idea of a partial assembly , with just the SMDs mounted ? 

personally not for me, I want an assembled one, but an option for the DIYers ? 

[+mytek]

7 hours ago, MessalineApghar said:

what about the idea of a partial assembly , with just the SMDs mounted ? 

personally not for me, I want an assembled one, but an option for the DIYers ? 

Normally I would agree that this would be a good way to go, but it doesn't allow for the SMD assembly to be completely tested, thus leaving the distributor subject to harassment if the board fails to work for the DIY assembler.

 

Bottom line is it's best to only sell completely assembled, fully tested boards.

[ajacocks]

IMO, selling fully assembled boards and blank PCBs is the way to go. That removes the support questions altogether.

 

- Alex

[+mytek]

OK so after my last very recent post in reply to the idea of any kind of DIY version being sold. I had some additional thoughts that popped into my head that deserve to be written for others to read. So here it is...

 

To set the record straight as far as things I invent in the world of Atari, I am not obligated to release the finished form of this (e.g., gerbers, firmware) to the general public. And although I have made many public releases of my technology, it must be understood that I won't always do so. Maybe it's best to think of this in the same way we do for Candle (the creator) and Lotharek (the producer). Speaking of which when was the last time you saw either one of these guys release a gerber of their products? In this sense MacRorie and myself have formed a similar relationship, with the caveat that I have always at least shared the schematics of whatever comes out of my development lab.

 

Speaking of schematics, I will always release the schematics of whatever makes it to a production stage, and many times as it's developing to that point. This is just like the days of vacuum tube radios where you could count on a schematic to be pasted inside the cabinet. I too feel the need to do so. Because when MacRorie and I are long gone in a galaxy far, far away, someone will still be able to diagnose and repair these boards in the year 2525 and beyond . For me having access to schematics on the original Atari line of 8-bit computers, is what truly inspired and enabled me to do what I do today. I can only hope that others find my schematics as inspiring to their dreams and efforts.

[+DrVenkman]

Another factor to consider for folks who may wish to buy a bare board at some point down the line ... the design requires programming 5 different chips using two different programmers: the custom EEPROM for the OS and game/language slots, the MMU, the EMMU, the TKII PIC chip and the VGATE PIC chip. The first three can be programmed with the current-generation TL866 II Plus programmers (but the EMMU cannot be programmed with the older TL866A/CS models). The PIC chips can be programmed with either a PICkit programmer under Windows, or a JOY2PIC programmer built from the boards and schematics provided by Mytek on his website. One PIC (the VGATE chip) is a DIP and it’s socketed, so you can program it separately. The TKII PIC is an SMD chip and is programmed via the ICSP header on the NUC+ board. 

 

So in addition to buying a board and the parts, you also need to buy or acquire 2 entirely separate programming devices.  It’s not a trivial effort for someone who doesn’t routinely build, test and repair these devices. 

[+ZuluGula]

@DrVenkman I don't think you will see here many newcomers that will jump straight in to assembling 576NUC+. I think most of the people who would want to build one, alterady build 1088XLE or 1088XLD (or both) and succeeded, they will have the required tools and skills. We also have some people who attempted it and failed, they will probably understand that 576NUC+ is not for them.

1 hour ago, mytek said:

To set the record straight as far as things I invent in the world of Atari, I am not obligated to release the finished form of this (e.g., gerbers, firmware) to the general public. And although I have made many public releases of my technology, it must be understood that I won't always do so. Maybe it's best to think of this in the same way we do for Candle (the creator) and Lotharek (the producer). Speaking of which when was the last time you saw either one of these guys release a gerber of their products? In this sense MacRorie and myself have formed a similar relationship, with the caveat that I have always at least shared the schematics of whatever comes out of my development lab.

The difference between you guys is that Candle/Lotharek partnership was business oriented from the beginning (with exception for SIC!), and your project started as fully free to anyone and became commercialized with time (1088XEL - gerbers, 1088XLD - pcb and 576NUC+ - fully assembled only). I don't say that it's anything wrong with that, only the comparison to Candle/Lotharek is wrong.