2023 Advanced PCB Remake For the 800XL -- Revision 1.1

[olix]

I have made today some crystal tests. Short story:

 

My crystal from the german shop "reichelt.de" - only a grey picture on tube tv.

Scope shows 3.57840 Mhz (to low)

 

same crystal with  22pico on each side: COLOR.

Scope shows 3.57989 (little bit to high)

 

I had also tested a very cheap crystal from China (AliExpress): COLOR (without 22 pico)

Scope: 3.57915 (little bit to low)

 

I have ordered some more crystals for testing.

[+scorpio_ny]

Hi I just want to report back that I received the crystals from Console 5 and I had the exact same issue with a black and white composite display. It also blinks on and off. Without the caps, I am seeing a clock bouncing from 3.577 (low) to 3.579 (high) by reading GTIA pin 29 with my scope (as per the SAM's repair guide.) With the caps in place , I see a reading of 3.579 (low) to 3.582 (high).

 

At this point, I think i am going to finalize my 800XL remake  setup and use the ceramic caps with the crystal in place permanently.  The only thing I am curious if there is a more optimal cap rating  that should be used or is there a better way to do this?

 

I have an 800XL that I keep around for testing, but I have to get an Antic chip so I can do the tests (I took it for my 800XL remake 😄). When I get back together, I will do some more tests with it since this piqued my curiosity.

 

@reifsnyderb: Do you need me to any more testing? 

 

Edited April 2 by scorpio_ny

[reifsnyderb]

31 minutes ago, scorpio_ny said:

Hi I just want to report back that I received the crystals from Console 5 and I had the exact same issue with a black and white composite display. It also blinks on and off. Without the caps, I am seeing a clock bouncing from 3.577 (low) to 3.579 (high) by reading GTIA pin 29 with my scope (as per the SAM's repair guide.) With the caps in place , I see a reading of 3.579 (low) to 3.582 (high).

 

At this point, I think i am going to finalize my 800XL remake  setup and use the ceramic caps with the crystal in place permanently.  The only thing I am curious if there is a more optimal cap rating  that should be used or is there a better way to do this?

 

I have an 800XL that I keep around for testing, but I have to get an Antic chip so I can do the tests (I took it for my 800XL remake 😄). When I get back together, I will do some more tests with it since this piqued my curiosity.

 

@reifsnyderb: Do you need me to any more testing? 

 

Could you post a picture of your crystal installation?  Thanks!

 

 

 

[olix]

This is my installation: I soldered a socket, so i could change the crystal faster without soldering.

[Jfcatari]

Can we start to collect all of the tested crystals with results and needed tweaks?

[+scorpio_ny]

7 hours ago, olix said:

This is my installation: I soldered a socket, so i could change the crystal faster without soldering.

I have been doing something similar for testing. I left a length  leads/legs cut off in the motherboards and I just solder to the extended leads.

[+scorpio_ny]

Warning- wall of text coming.

 

So I have been doing more research on the subject of the crystals. It seems that the clock crystals require an optimal capacitance circuit to clock in correctly. Too low, and they will clock in lower (which we have been seeing). And each crystal has different  requirements. 

 

I decided to use 22PF caps because it was I had on hand that was closest to the 18PF spec of the crystal that

I was testing at the time (based on it’s spec sheet-second one in the list):

https://www.mouser.com/datasheet/2/94/CYxx-355433.pdf

 

Now according to this article, that is not the correct way:

https://www.iqdfrequencyproducts.com/blog/2020/08/03/myths-around-load-capacitance-how-to-choose-the-right-capacitors/

 

Here is an excerpt:

“For the total load capacitance in the circuit, all capacitances need to be considered. Therefore, not only the two capacitors, but also the input and output capacitance of the microcontroller and all stray capacitances must be taken into account. This all together forms the load capacitance.”

 

The reason the Atari crystals are working probably means that its capacitance needs to clock optimally are already met in the current PCB design.

 

In another article, it stated

that it may be better to use a crystal oscillator in some instances:

https://www.jauch.com/blog/en/oscillator-or-quartz-crystal-how-to-find-the-right-component-for-your-application/#:~:text=In%20contrast%20to%20a%20quartz,the%20quartz%20crystal%20to%20oscillate.

 

Here is an excerpt:

“The oscillator is therefore the most space-saving alternative: Everything is compact and built into one component, and the developer does not have to design a circuit around the piezoelectric crystal. The component is particularly suitable for low volumes because it eliminates the need for time-consuming optimization and tuning of circuit components.”

These are the “tin can clocks”.

So, maybe a crystal oscillator may be a better choice to clock the Atari? I am not an expert on this and I am just curious what would be the cons to using this clocking method.

 

 

Edited April 2 by scorpio_ny

[+scorpio_ny]

On 4/2/2024 at 12:27 AM, reifsnyderb said:

Could you post a picture of your crystal installation?  Thanks!

This is how I set it up. May not be the best way, but it works. Just tested it and I get color with composite output.
With this, I am done setting up my 800XL Advanced.

 

Out of curiosity, I tried with two 10PF capacitors on each leg to see if it would work. No go.

Edited April 5 by scorpio_ny

[reifsnyderb]

1 minute ago, scorpio_ny said:

This is how I set it up. May not be the best way, but it works. Just tested it and I get color with composite output. Out of curiosity, I tried with two 10PF capacitors on each leg to see if it would work. No go.

Looks good!  It sounds like maybe the capacitors should be socketed?

[+scorpio_ny]

13 minutes ago, reifsnyderb said:

Looks good!  It sounds like maybe the capacitors should be socketed?

I think that  would be a good idea. Maybe a jumper setting too? One setting could be for Atari  sourced crystals and one for non-Atari sourced crystals? 

 

[reifsnyderb]

1 minute ago, scorpio_ny said:

I think that  would be a good idea. Maybe a jumper setting too? One setting could be for Atari  sourced crystals and one for non-Atari sourced crystals? 

 

Yes, there would have to be a way to switch between the two.

[olix]

I have tested 5 more crystals. All purchased from Mouser:

 

I had the best measured frequency with the ECS-35-30-4X: 3.57934 MHz

Also good results with: LFXTAL016962Bulk and ECS-35-17-4X

 

Only an grey image (no colors): ECS-35-17-1X and LFXTAL000052Bulk

 

(All tests without additional capacitors)

 

LFXTAL016962Bulk
https://www.mouser.de/ProductDetail/449-LFXTAL016962BULK
HC49/4H

Frequency:    3.57945 MHz
Load Capacitance:    30 pF
ESR:    300 Ohms
Tolerance:    30 PPM

Color-Image:    Yes
Measured frequency: 3.57923 MHz

---

ECS-35-17-4X
https://www.mouser.de/ProductDetail/520-HCU357-17X
HC-49/US

Frequency:    3.579545 MHz
Load Capacitance:    17 pF
ESR:    200 Ohms
Tolerance:    30 PPM

Color-Image:    Yes
Measured frequency: 3.57917 MHz

---

ECS-35-17-1X
https://www.mouser.de/ProductDetail/520-HCA357-17X
HC-49

Frequency:    3.579545 MHz
Load Capacitance:    17 pF    
ESR:    120 Ohms
Tolerance:    30 PPM

Color-Image:    NO
Measured frequency: 3.57813 MHz

---

ECS-35-30-4X
https://www.mouser.de/ProductDetail/520-ECS-35-30-4X
HC-49/USX

Frequency:    3.579545 MHz    
Load Capacitance:    30 pF    
ESR:    200 Ohms
Tolerance:    30 PPM

Color-Image:    Yes
Measured frequency: 3.57934 MHz

---

LFXTAL000052Bulk
https://www.mouser.de/ProductDetail/449-LFXTAL000052BULK
HC-49

Frequency:    3.579545 MHz    
Load Capacitance:    20 pF    
Tolerance:    10 PPM
ESR:    150 Ohms

Color-Image:    NO
Measured frequency: 3.57838 MHz
 

[+scorpio_ny]

Thank for the info @olix! So the issue is that some of the crystal are slightly under clocked. One question that comes to mind of the ones that worked were just one offs you just happened to pick the one that worked? 

 

[_The Doctor__]

So long as you get the frequency spot on for the Atari to execute code at full speed and produce color nothing else really matters. Some oscillators are noisy so you may have to clean the signal up a bit else you get more digital noise as well. Always a bit of this and a bit of that. Timing is critical so get it in spec no matter the crystal or oscillator. Especially with add on devices and such tolerances.

[olix]

I have a small suggestion in case there is ever a new board revision:

 

In addition to the connections pads for the console buttons, it would be very practical if there were also connection pads for the signals required by a stereo Pokey extension (PokeyMax):

- A4
- GTIA Bell (GTIA Pin 15)

Even better:
A5, A6 and A7 (for Quad Pokey, Covox or similar)

 

I don't think that would be too much effort, but it would be a great help.

[reifsnyderb]

5 hours ago, olix said:

I have a small suggestion in case there is ever a new board revision:

 

In addition to the connections pads for the console buttons, it would be very practical if there were also connection pads for the signals required by a stereo Pokey extension (PokeyMax):

- A4
- GTIA Bell (GTIA Pin 15)

Even better:
A5, A6 and A7 (for Quad Pokey, Covox or similar)

 

I don't think that would be too much effort, but it would be a great help.

The console pads were mainly added to simplify testing when the board is on the bench.

 

The POKEY area of the board is rather complicated, and if it's possible to add those pads it could easily take several hours to a day to re-route everything just for this change alone.  (I've included a picture of the POKEY chip region.)  Also, if I add pads for one extension, the course is set to add pads for many other extensions.  (There is at least a half dozen.)  An option would be to turn the board into a 4 layer board so as to greatly simplify routing.  The board price would be horrible, though.  So, my suggestion is to use the bread boards to customize the board for whatever connections you need.

 

I do have another board revision under development.  But it's not a high priority.  Here's the current status with additional notes in between lines:

 

 

Changes:

Remove re-settable fuse -- done

    Probably not used and never tested.  Current limit on available re-settable fuses are rather high anyhow.
Add S-Drive simple capability -- done (added breadboard)

    Small breadboard section added to left of SIO connector so connections can be broken out.
Make minor video trace improvements -- done

    Several components moved to simplify trace layout.
Denote composite video only components -- done

    Silkscreen change only
Add +5v and GND input to silksreen -- done
Added optional R1 to cleanup B Phi2 -- done

    Needs tested.  Should remove bounce from B Phi2.
Improved power routing to cartridge -- done

    100nF capacitor is now located on +5v line before cartridge only.
Add optional capacitor footprints for use with clock crystals -- to be done

 

 

Probably the biggest change is the capacitor footprints.  I may just make up custom crystal footprints and use those.

 

 

Here's a picture of the POKEY region:

 

 

 

 

 

 

 

 

 

Edited April 11 by reifsnyderb

[olix]

Ah, I see. The area around the pokey is really very densely covered with routes*. I imagined it would be easier. The idea with the additional pads is also just a "nice to have" and not a "must have". The idea of using a breadboard area for this is completely sufficient.

 

You shouldn't change to a 4-layer PCB just for this reason. That would only increase the costs unnecessarily.

I am happy with your board as it is. 

 

 

*By the way, what is the right word in English for the routes on the pcb : conductor, conductor path or trace? 

Edited April 12 by olix

[Geister]

When I worked in a group that designed circuit boards at Air Products, the Wintek software we used referred to them as traces.

[olix]

I assembled my second 800XL Advanced Remake today. This time as a PAL model. As with the first board, everything worked great. I like this PCB better every time.

Edited April 21 by olix

[reifsnyderb]

1 hour ago, olix said:

I assembled my second 800XL Advanced Remake today. This time as a PAL model. As with the first board, everything worked great. I like this PCB better every time.

Very nice!  You'll want to add that zenor diode, too.

[Jfcatari]

@olix

 

Nice second build...I have two more to do myself, though I have to finish the Atari NUC first!

 

 

[olix]

10 hours ago, reifsnyderb said:

Very nice!  You'll want to add that zenor diode, too.

Oh, I must have missed the diode. I'll make up for it right away this evening. Thanks for the tip.

[olix]

I am happy to share with you some files that I have made for myself to make assembling the board a little easier:

 

Assembly-List.pdf

All components sorted by component name and by component value

 

PCB1_Search.pdf

Searchable graphic of the PCB to make it easier to find the components

 

PCB_Oberseite.tif and PCB_Unterseite.tif

High-resolution, lossless scan of top and bottom of board

 

Maybe this is a little help.

Assembly-List.pdf PCB1_Search.pdf PCB_Oberseite.tif PCB_Unterseite.tif

[olix]

... and here is my shopping list with the order numbers from mouser.com and reichelt.de (German mail order company for electronics).

 

 

BOM-V1d.pdf

[olix]

I have now installed a PoekyMax. I wired the required signal lines on the underside and routed them upwards via the breadboard area.

If you choose an electrolytic capacitor with a low height for C98, the stereo pokey extension fits over the capacitor without any problems and you don't have to bend it over.