Sunday, 19 July 2020

Mini PET Update

The Mini PET seems to have been received well. I am seeing lots of pictures of very neatly assembled boards (and also the one that TFW8b built) and revived PETs up and running again.
That's great to see, and very few problems building them, other than a few people who had to built it without reference to the manual (the person packing the orders has of course been fired). So far, the only issue to report is a slight error that seems to have slipped though all of our other testers (who have, of course, all now been fired).
When first putting together the design for the prototypes of the Mini PET, I came to the userport and was pleased to find I had already drawn a component for the PET userport. Perfect I thought and added it to the design. I didn't consider at the time that I had drawn that component for things which plug into the userport on a PET, not for the host side of the PET userport, which is the mirror of that pinout. (I have, of course, also been fired)
And so, the pinout is reversed. Pins that should be 1,2,3,4... are 12,11,10,9... During testing, no one it seems noticed that, as no one it seems has anything that would plug into the PET userport. The only things I had were userport sounders and composite video output. All of which were already provided on the Mini PET. I think this will only affect a handful of people making userport serial interfaces, as the PET seems to have hardly any commercial userport peripherals.
I have made some boards which reverse the pinout. An interesting routing problem, 12 wires what all need to cross over each other and the same on the other side of the board.
I'm fairly happy with the solution I came up with, although I wonder if there is a neater way to reverse the connections? Those are being sent out to anyone with an affected board (V1.42, V1.44), although no one has actually asked for one, and only one user has reported a problem. The ones going out have white soldermask.
I build a little LED tester plug to make sure that was working correctly.
Because of the userport problem, I have moved more quickly than planned to the next board designs, as correcting the pinout on the userport required quite a bit of rerouting. The updated versions of the 'Kit A' boards is V1.45.
Other than the correction to the userport pinout, and a couple of minor tweaks, this is pretty much the same as before.
The LED tester confirms the pinout on these boards is now correct, so the twister boards are not required for V1.45 and later boards. The new 'kit A' versions (for use stand alone with a keyboard) are shipping now.
The Mini PET boards had been designed to be the same size as the keyboard PCB, which had already been designed as a direct replacement for the chiclet keyboard on the 2001 PET, so the Mini PET was designed to be the same width as that.
That meant it wasn't quite wide enough to pick up on the two mounting pillars at the back of the PET case.
I had tried out a green soldermask version of the Mini PET board and that had looked better inside the PET case. It seemed a good opportunity to split this into two versions, one white soldermask board, the same size as the keyboard, for stand alone use, Kit A. The second, a green soldermask board which fits inside the PET, kit B.
The first thing to do was to make the B boards larger, so they are now the same width as the top end of the original PET boards.
So those now fit into the case and pick up on both rear mounting pillars.
It also seems a good idea to remove things not required for this version.
I've removed the DC jack, and boxed off the composite video section, which can be fitted if you need it, but otherwise can be left unpopulated.
You can see here I prefer to fit the smaller logic chips without sockets. They are still supplied in the kits if you want to use them, I just think it looks better without. TI uses the same moulding for 14 and 16 pin chips, so when aligned correctly, you don't see the difference between 14 pin and 16 pin chips.
The wider kit B boards now pick up on both rear mounting pillars, and the power supply interface board has been integrated to make a single board solution.
As before, this can be used with the 9 way power connector found in most PETs or the 5 way connector from the original 2001s. You could cut down the pin header before soldering if you are only going to be using the 5 pin connector. But it's not a problem if you don't, one missing pin is an extra ground, the other two are for a second transformer winding which is not used.  
There is now a side datasette port, for datasette 2, positioned as on later PETs, so the board gives access to both rear and side datasette ports.
This now works better for the 8032-SK case where the board is mounted at 90 degrees, and the side port becomes the rear port.
The board picks up on the same two mounting points and the extension cables connect as normal inside the 8032-SK case.
The board can alternatively be mounted further into the PET case, which allows an SD2PET to be installed inside the case. This can be useful in unattended events to keep it away from prying hands, or for those short of desk space, or wanting to access the SD card by lifting the case rather than reaching around the back.
The second datasette is still accessible through the holes in the case in this configuration.
You will see I have moved over to using jumper settings, rather than DIP switches on the B boards (thinking it would be adjusted less frequently once mounted inside a PET case). 
I'm not sure about this, and may go to a single 8 way DIP switch on future boards as it adds quite a few extra parts to be sorted out and soldered in, but keep the onboard labelling of options.
These kits are being distributed by The Future Was 8 bit. My Tindie store remains closed due to the ongoing problems with shipping. 

If you want to support this blog, you can donate via Patreon or Paypal, or buy something from my store (if / when it reopens) - TFW8b.com is still open, so go and buy something from there instead.