Cloning a Commodore PET-2001

[kizmit99]

Wow! - that cassette interface is certainly more complicated than the one I'm using for my TRS-80 clone... 
It's not a direct copy of the original.  It's similar, but should be significantly less finicky than the original.



I'm amused that I ended up in a similar place to you on the small voice-recorder to replace the old cassette deck.  (Yes, I'm easily amused  )
Anyway, always fun to see more progress.  Thanks for sharing!

[GK]

Wow! - that cassette interface is certainly more complicated than the one I'm using for my TRS-80 clone... 

Well,  like I said; "this is the "deluxe" cassette port analogue interface" 

When I was prototyping my PET for expediency I soldered together a cassette port audio interface just as simple as yours, but it wasn't entirely reliable. And a mic.-level write signal won't suffice for a line-level input and vice versa. And it wouldn't load at all, let alone with complete reliability, from playback signals getting as bad as this: 

[kizmit99]

Eek - I didn't notice first time around that you were trying to clean up Mic level inputs.  Not much signal to work with there to begin with!
Out of curiosity, did the original have to deal with Mic levels?  That seems like a pretty formidable task, I would have expected Line level inputs at worst.

Question for you -- are you able to successfully replay data that's been mp3 compressed?  I've been mostly playing with wav encoded files to this point, but having a 5Meg audio file encoding 15K of data seems like a bit of over-kill

[GK]

Eek - I didn't notice first time around that you were trying to clean up Mic level inputs.  Not much signal to work with there to begin with!
Out of curiosity, did the original have to deal with Mic levels?  That seems like a pretty formidable task, I would have expected Line level inputs at worst.

Question for you -- are you able to successfully replay data that's been mp3 compressed?  I've been mostly playing with wav encoded files to this point, but having a 5Meg audio file encoding 15K of data seems like a bit of over-kill

Writing to both mic.-level and line-level inputs; hence the switched attenuator in the write signal chain giving three different write signal levels.
I wanted the read input to work without complaint from a couple of volts down to 100mV input with a adequate safety factor. Bits start dropping out at around 10mV rms, (depending on the signal-to-noise ratio) so that's that spec. met.

The various models of the Commodore "Datasette" all communicate with the computer via TTL-logic signals. All of my recordings so far are in mp3 format - no problems at all.

I finished the unit off and started putting it to use this afternoon . The banana posts on the rear panel connect to the motor-control relay contacts and can be used if desired to auto-switch the motor of an external tape recorder. The "sense" switch tells the computer that play/record have been pressed. This switch is normally just left on to skip "PRESS PLAY ON TAPE" (load) and "PRESS PLAY AND REC ON TAPE" (save).

[GK]

http://youtu.be/e4uILTKAoj8

[MK14]

Thanks for the video, and very interesting project.
I'm amazed at how neat, the internal construction of your project is.
It seems to be working very well, and has brought back, found memories to me, by watching the video.
Well done!   

One quick question. What is the difference, between the two versions of basic, apparently selectable, by the switches on your unit ?

[GK]

The upgraded BASIC gets a big list of bug and limitation fixes, the Bill Gates Easter egg and "Tiny Mon" (TIM), a simple machine language monitor.
Some software written with work-arounds for the original BASIC bugs and limitations might conceivably not work properly with the fixed, "upgraded" BASIC, so I added being able to run both as a feature. Also the character ROM bundled with the original BASIC machines swapped upper and lower case A-Z in the second page of the character set - that being the "alternate" page called the "business" set, as opposed to the "graphics" set that the machine boots up in by default.

 

[GK]

Finally, the project web page is not yet complete, but all of the technical documentation and design files are posted:

http://www.glensstuff.com/pet2001/pet2001.htm
 

[SiliconWizard]

Impressive 

[MK14]

The upgraded BASIC gets a big list of bug and limitation fixes, the Bill Gates Easter egg and "Tiny Mon" (TIM), a simple machine language monitor.
Some software written with work-arounds for the original BASIC bugs and limitations might conceivably not work properly with the fixed, "upgraded" BASIC, so I added being able to run both as a feature. Also the character ROM bundled with the original BASIC machines swapped upper and lower case A-Z in the second page of the character set - that being the "alternate" page called the "business" set, as opposed to the "graphics" set that the machine boots up in by default.

Thanks for that really nice explanation.
It must be real fun, playing around with those different versions, of the PETs Basic roms and things.

You're tempting me, to build your project   

Thanks for making it open source and/or available online and here. It makes a very interesting read and/or project to build.

One thing/feature you forgot.
When an early PET was switched off, you could see a white dot, that faded over a number of tens of seconds, in the middle of the (crt) screen. Unfortunately, you seem to have forgotten about it. (If I remember correctly).
Only joking, no you have done a great reproduction job!

[GK]

Hey, thanks for noticing 

As a general rule, once the design files are turned out into the public domain, that's the last you hear from any interested correspondent.

I tipped the project over at Hackaday in a fit of shameless self-promotion. Dunno if it will be worthy though.
 


   

[nad007007]

Hi GK

I have emailed JLCPCB.com about my error "Can not identify the board outline" and they suggest to manually enter the sizes of the boards. Now do you have the dimensions for the 3x boards?

Thanks

Regards
Conrad   


 Conrad Larsen
Hi Glen, very cool project. I have up loaded all three Gerber files for pet2001 to JLCPCB.com and i get errors "Can not identify the board outline" this puts the drill holes out. Hope you can help with editing the Gerber? Kind Regards Conrad
Glen Kleinschmidt
Hi. I normally use PCBcart or Elecrow for personal projects, but have not used JLCPCB and have never received a similar complaint from another manufacturer, so am not sure what the problem might be. Might be worth contacting JLCPCB technical support for an explanation?
Conrad Larsen
Hi Glen, do you have the three boards dimension please?

Regards Conrad

[GK]

In thou:

Motherboard - 10200 x 6800
Video board - 7100 x 4500
PS2/RS232 board - 5950 x 2100

[BrianHG]

Hi GK

I have emailed JLCPCB.com about my error "Can not identify the board outline" and they suggest to manually enter the sizes of the boards. Now do you have the dimensions for the 3x boards?

Thanks

Regards
Conrad   

@GK, in the future, when making gerbers in Protel99SE, to make the PCBs gerbers be auto-recognized by JLPCB.com, you need a trace around your PCB on the 'Keep Out' layer, and, also a copy of those traces on the main 'Mechanical Layer 1'.  When generating the gerbers, make sure that the 'Mechanical layer 1' is applied to all greber layers.

With this, many online PCB shops, like some in China who use an all automated online proofing of gerbers before manufacturing like JLPCB will auto-detect the outline of your PCB without issue and auto cut the PCB to that shape.

[GK]

In each case I have the board outline defined in the mechanical layer (GM1). From here:

https://support.jlcpcb.com/article/42-how-to-export-altium-pcb-to-gerber-files

Please make sure you have the clear outline in mechanical layer.

 


In almost 20 years of Protel now an hundreds of boards manufactured, I've never had a request for anything beyond that. Can you define a board shape in 99SE like you can in AD13? I don't know because I've never had to do it before, and right now I have to go out.

[BrianHG]

In each case I have the board outline defined in the mechanical layer (GM1). From here:

https://support.jlcpcb.com/article/42-how-to-export-altium-pcb-to-gerber-files

Please make sure you have the clear outline in mechanical layer.

 


In almost 20 years of Protel now an hundreds of boards manufactured, I've never had a request for anything beyond that. Can you define a board shape in 99SE like you can in AD13? I don't know because I've never had to do it before, and right now I have to go out.

You are correct.  Yes, I've done PCBs since the early 90s.  However, over the past 27 years, all my gerber files were inspected by a human operator loading the files and seeing the board outline right there on their gerber viewer.  (At the time, my prototype PCBs also cost hundreds of dollars, over a thousand or two for a 4 layer, all made here in North America unlike today's highly all automated shops which would end up being under 100$.)

JLCPCB has no human in the loop when you upload a .zip of gerber files on their website.  Withing seconds, it just shows you a 3D quality rendering of your final PCB, that is if it can determine the PCB outline.  Otherwise, a human would need to do a degree of work.

I'm not sure if it's a quirk with P99se, but all I have done is tell you how I define my PCB outlines in P99se & when I send my P99se .zip gerbers to JLPCB, it instantly reports back my board dimensions and show me a preview of my PCB.


Note than in P99se, it is usual practice that the board outline is defined on the 'Keep Out Layer'.  Using this layer, you draw the shape you like, and you can even create multiple PCB on 1 cad document this way.  I just draw a rectangular trace around the PCB.

The reason I copy and paste my board outline from that layer to the mechanical layer is to have the board outline visible on every gerber file layer.  This makes it obvious to the automatic detection software which JLCPCB uses to find the PCB outline.

I usually send my .zip to JLCPCB as a test before I make a design public since the render they give you on the web page is instant and represent what they and others will give you if others have lazy inspection engineers and just hit the 'AUTO' button on their gerber viewers.

[BrianHG]

Here is an example upload I've done on this site which you can play with / look at the protel PCB / send the .zip to https://jlcpcb.com/quote 's website to see what happens:

Protel project file to inspect / or copy my source PCB in protel, you may play with it and inspect the board outline:
https://www.eevblog.com/forum/projects/free-subwoofer-audio-processor-project-for-forum-members/?action=dlattach;attach=403539
Gerbers (upload these at https://jlcpcb.com/quote ):
https://www.eevblog.com/forum/projects/free-subwoofer-audio-processor-project-for-forum-members/?action=dlattach;attach=403538
Click on 'ADD your gerber file', send the .zip, and all JLCPCB's fields should automatically be filled in.
You should read:
"Detected 2 layer board of 70x100mm(2.76x3.94 inches) ."
and everything else on the web page should have the correct default values...

You can click on 'Gerber Viewer', and, on the new page, use you mouse wheel to zoom in and out.  At the top, you can select which layer of the PCB to view as well as PCB mechanical statistics.  You can also turn ON/OFF each layer, IE silk, drill, outline, copper, which make the PCB.

[GK]

You are correct.  Yes, I've done PCBs since the early 90s.  However, over the past 27 years, all my gerber files were inspected by a human operator loading the files and seeing the board outline right there on their gerber viewer.  (At the time, my prototype PCBs also cost hundreds of dollars, over a thousand or two for a 4 layer, all made here in North America unlike today's highly all automated shops which would end up being under 100$.)

Well, I've had these boards (the exact same Gerber files) and a ~dozen other designs made by PCB cart and Elecrow now, and this is the first time there has been an issue.
I just uploaded the Gerber files for the motherboard to JLPCB's viewer. It's having a fit over the alignment/outline of the drill guide.

[BrianHG]

Ahhh, ok, just do / try this (I've seen this before with my friend's board miller.) :

Follow my settings in the attached screenshots:

For the Gerbers Films, Select Imperial or Metric, with the highest resolution of 2:5.  For the film coordinates, use the relative origin.  (This means the coordinates used in the gerber file places 0.0 X 0.0 at the position of your set origin marker in your PCB.)

I usually use the bottom left corner of the PCB.

Now for the NC Drill settings. 
Use the same units and resolution as the Gerber File settings, and
!!! Use " Reference to relative origin " just like the Gerber Films so that your drill hole has the exact same coordinates as the pads on your Gerber Film which is also now set to the same reference relative origin.

This used to be aligned manually by the tech at you PCB manufacturing shop, so, the NC drill used to never have to match the coordinates of the gerber film.

See my PCB screenshot for relative origin position...

[GK]

This used to be aligned manually by the tech at you PCB manufacturing shop,

It still is I'd guess, it's probably just that JLPCB here are trying to be too smart in their on-line automated quoting system.

I made sure to follow the Elecrow guide here (as this is who made my boards):

https://www.elecrow.com/wiki/index.php?title=How_to_export_gerber_files_from_Protel

Note that they demand neither a keep out nor a mechanical layer and say nothing about the NC Drill file origin having to match the copper layers.

Also 2:4 format is specified for the copper layers and 2:3 for the NC Drill. 2:5 is beyond the manufacturing capabilities of a lot of these cheap fabs. Do you think it is wise to default to 2.5 for both? A manufacturer might balk if the format resolution of the submitted Gerber files exceeds their manufacturing abilities.


I think I'll just re-upload my Gerber *.zip folder with a .txt "readme" giving the board dimensions. That should be enough if required to get anyone out of trouble and I can't magically and simultaneously accommodate the possible quirks of every manufacturer out there.
 

[BrianHG]

Ok, nothing wrong with their instructions except they are missing the 'Advanced Tab' in the Gerber Film settings.

To change your current gerbers settings without changing anything else, all you need to do in protel99se is open your .cam file: ' CAM Outputs for ************.cam ' file.  (Not the CAM directory...)

Once open, double click on 'Gerber output 1'.
Now you can make changes to any of your gerber settings you like.
Same for 'NC Drill output 1'.

(you may also right click in the middle of the white blank window and 'Insert BOM' file, or 'Insert Pick and Place' file which will be included when you generate your gerbers.)

Once you make the relative origin change, press 'F9' to re-generate your cad files with the new settings.

This may be important as some users may wish to mill your PCB in one of those cheap desktop Chinese PCB miller which require the film and NC drill to have the same coordinates, otherwise the user has to edit the gerber or NC drill files.  Also, JLCPCB may have a 100% automated setup, with direct optically exposed PCBs, meaning, those purchasing your PCB from them will get PCBs which look like the web photo if there isn't a knowledgeable person in the loop who inspect the files and recognizes the error before manufacturing.

     As for the film resolution, I choose 2:5 for both since today, some of my footprints are designed in metric, others designed in imperial.  Making both the gerber film and NC drill high res, with equal precision just ensures on the PCB manufacturer's side, when their software/hardware converts your gerber/drill file's units to their machine's internal units, I just want to cover the finest possible accumulated rounding errors as units may be converted back and forth a number of times since my foot print and it's placement in protel.  But, yes, 2:5 is at least guaranteed above their machine's final resolution... except, on a fine BGA board with 1 mil micro-vias I made a few years back, there was a difference where 2:3 has a few vias off center.  Since then, I've always used 2:5 for both and never had a problem since.
     We don't need to save processing time and shrink file size today to the 2:3 resolution like we did for slower hardware back in the late 80s, early 90s.

[RJSV]

   AH Come-on, folks,  name Jeri Ellsworth I'm pretty sure,  (sorry to be so abrupt, exxcuse,) Well from what I read, she implemented an entire Commodore C64 using FPGA logic.  I'm not up to date,  now, but Jeri produced a multi-game clone, around 2004, of a bunch of C-54 classics.   Also included a TV set driver, in the hand-held, so it was a stand-alone system.
   I love working with the 6502, register rich, primative some ways, but with some really inspiring indexing and pointer power !
A C-64 with 256 bit colors, would be nice.
   please look up Jeri Ellsworth for her Engineering projects, if you care to,  thanks

[GK]

   AH Come-on, folks,  name Jeri Ellsworth I'm pretty sure,  (sorry to be so abrupt, exxcuse,) Well from what I read, she implemented an entire Commodore C64 using FPGA logic.  I'm not up to date,  now, but Jeri produced a multi-game clone, around 2004, of a bunch of C-54 classics.   Also included a TV set driver, in the hand-held, so it was a stand-alone system.
   I love working with the 6502, register rich, primative some ways, but with some really inspiring indexing and pointer power !
A C-64 with 256 bit colors, would be nice.
   please look up Jeri Ellsworth for her Engineering projects, if you care to,  thanks

The point?
I think that there would be few people reading this thread out of an interest in vintage computing who wouldn't already be aware of Jeri Ellsworth and/or the C64 Direct-to-TV product.

[RJSV]

Sorry about that post, ignorant / impulsive...  I saw those proto boards, on page 1, (love 6502!) and did not realize so much material was beyond. Apologies, Rick-Jack

[soldar]

Impressive project. I had a PET 2001 and also a Rockwell AIM-65 with which I did several projects and I still have a keypad access system I designed and built.

I still have a bunch of HN482764  (8192-word x 8-bit UV Erasable and Programmable ROM) and TMS2516 (2K x 8 bit UV ROM) which I will gladly give away to anybody who can use them.