F18A DIY?

[Hans23]

2 minutes ago, ti99iuc said:

I have found these two images

Thanks!  Now I need to find a source for these nice pins.  It looks as if they're individually made and not on strips, which makes searching harder.

 

Cheers, Hans

[Jimhearne]

I think the pins Matthew used must have been single pins rather than a header.

Probably press them into a turned pin IC socket to keep them aligned while soldering them to the PCB

 

Jim

[Jimhearne]

How about these ?

 

https://www.digikey.co.uk/product-detail/en/mill-max-manufacturing-corp/3169-0-61-15-00-00-03-0/54-3169-0-61-15-00-00-03-0CT-ND/7931982

 

Or lots of others

https://www.digikey.co.uk/products/en/connectors-interconnects/terminals-pc-pin-single-post-connectors/323/page/1

 

Jim

 

Edited February 13, 2021 by Jimhearne
added extra link

[Hans23]

Those Mill-Max pins look great - They considerably add to the overall BOM price, but I guess I'll just take all the encouragement that I've got here and order 500 of them. That'll give me enough material for 21 boards, and I'm pretty sure that once I've built those, I need to pass the ball on to someone else

 

Cheers, Hans

[wierd_w]

As a matter of interest--  the socket-pins I ordered as strips, easily slide out of the plastic strip.

 

https://www.amazon.com/gp/product/B0778TFL39/ref=ppx_yo_dt_b_asin_title_o09_s00?ie=UTF8&psc=1

 

If you want, I can easily demonstrate this.  It is actually rather annoying when trying to use them as strips, but if you want them as free socket pins, it might be a boon.

[Hans23]

What I need is double sided pins, and they need to have different diameters on the PCB and the socket side. Even the pin strips that I have used so far are not perfect because the diameter of the PCB holes is too large for them.  Also, build time is a real consideration now that I need to build a whole bunch of boards. I would much rather not have to mess around with repurposing parts.

 

I think that I will get the pins next week and with them, the boards are going to look great

[wierd_w]

I get what you mean.  But the price difference looks pretty significant.  There are a LOT of socket pins in that package for 8$.

 

But really, getting them out of the strip is not difficult at all. (Pressing gently on a hard surface will dislodge the entire strip) The metal 'cup' of the socket is about the same size as a large via.  I have a pair of calipers, and can mic them out if you want.

 

 

[Hans23]

@wierd_w Thanks, but I had already placed my order at digikey.  For myself, I'd go with a less professional version, but as I'm building for others, I'd like the result to look as good as possible.

[wierd_w]

Well, Just for general information then:

 

The profile of the de-stripped sockets in that pack is like this:

 

[Hans23]

How would you mount these pins in the holes of the F18A?  They're 0.7mm in diameter.

[wierd_w]

I said "the size of a large via"

 

If that is a standard via, then no go.  This is why I mic'd them out.

[matthew180]

Too much to say, not enough time.

 

* Sorry for being absent, but life, pandemic, new job, etc. have all been keeping me away from anything hobby related for the last few months.

 

* Sorry for not having any F18A (of any kind) available for, uh, er, um, years...    This was not my intent.  But a lot of the reason for the MK2 design was exactly the problem Hans23 mentioned, i.e. the availability of the Spartan 3E FPGA, etc.

 

* The schematic on the CircuitMaker website was not really forgotten, I just never bothered to log back in to delete it.  I was going to use C.M. for the MK2 design, and to rework the original F18A for general open-hardware release (which has still not actually happened how I would like).  So that board layout was an import of the original F18A design I did in Eagle.  It is probably based on the final board layout for the F18A, but I never built any boards from that C.M. import, and I use KiCAD now.

 

* Board version V1.3 was the last F18A revision, and also the point at which I realized I should have used revision "letters" for the board, so I could keep it separate from firmware "versions".

 

* The extra unpopulated footprints on the board around the flash IC were added when I was going to try to include composite video output.  I needed a MOSFET and a few passive components, but I did not finish or test the design, and I did not have room in the FPGA for the logic, so I just left the parts off of the board.

 

* Yes, PCB pins are a pain in the ass, and I literally spent tens of hours searching for solutions.  The size of the F18A is also a problem, which is another reason for doing the MK2.  The pins are mill-max, both short and tall-pin versions.  The short pins are press fit, but the hole I used was just a little larger (I did not want to buy the $1000 tool to press the pins), so the pins fit tightly and are soldered from the top.

 

Short pins: DigiKey: ED90328-ND, Mill-Max: 3157-0-00-15-00-00-03-0, Press Fit

Tall-pins: DigiKey: 575-641591, Mill-Max: 342-10-164-00-591000 SIP header, break away

 

I put the pins into a DIP-40 machine-pin socket (one by one), then press the F18A board down and solder.

 

The MK2 will use Flip-Pins: http://oshchip.org/products/Flip-Pins_Product

 

The socket hole drill size: 27.559mil ~= 0.7mm, 41.338mil ~= 1.05mm diameter

 

* The F18A V1.3 board will run the V1.9 firmware.  Yes, you should run the latest firmware, and yes, it breaks a few of the demos since sprite-linking was removed in V1.9 (to make room for features that were actually useful).

 

* I don't know how I would feel if someone started producing the F18A in bulk.  Not sure.  No one has me asked about it either, though.

 

* The HDL in the Phoenix project *is* the F18A HDL.  I gave the project permission to use the F18A core, and ultimately got heavily involved in reworking and writing a lot of the HDL for the project (which ended up being everything except the Z80 core and SNES + Keyboard HDL, IIRC).  It is not my official public open-source release, but yes, the HDL is sitting there in the Phoenix repo for anyone to discover (if anything, we don't want to be hypocrites).

 

* I believe the F18A should not draw more than about 150mA in my testing and calculations, which is about 100mA less than the 9918A is spec'd for max power.

 

* Yes, the JTAG header is 2.0mm with a ground pin between every signal pin.  You can blame Xilinx for that...  Very big header for a small board.

 

Here is the original F18A V1.3 BOM for reference if it helps with individual builds:

F18A V1.3           Mouser/Digikey          Manufacturer Part       Package     Qty
FPGA                122-1525-ND             XC3S250E-4VQG100C       VQFP100                             1
Serial Flash        M25P80-VMW6TGCT-ND      M25P80-VMW6TG           SOIC-8                              1
Level Shifter       863-MC74LCX16245DTG     MC74LCX16245DTG         TSSOP-48                            1
Single AND gate     512-NC7SZ08M5X          NC7SZ08M5X              SOT23                               1
REG1v2              576-2836-1-ND           MIC5268-1.2YM5 TR       SOT-23-5 150mA                      1
REG3v3/2v5          576-1927-1-ND           MIC5320-SJYD6 TR        SOT-23-6 150mA/150mA                1
OSC50m0             535-10087-1-ND          ASV-50.000MHZ-E-T       DFN4 5mm x 7mm                      1
CAP 0805 1.0uF      GRM21BR71C105KA01L      81-GRM40X105K16L        0805 1.0uf, 16 volt, X7R, 10%       5
CAP 0603 0.1uF      GRM188R71C104KA01D      81-GRM39X104K16         0603 0.1uf, 16 volt, X7R, 10%       20
Resistor            71-CRCW0805-4.7K-E3     CRCW08054K70FKEA        0805 4.7K                           2
Resistor            71-CRCW08051K20FKEA     CRCW08051K20FKEA        0805 1.2K                           1
Resistor            311-100HRCT-ND          RC0603FR-07100RL        0603 100ohm                         3
Precision resistor  311-100CRCT-ND          RC0805FR-07100RL        0805 100ohm, 0.1% Horz, Vert        2
Precision resistor  311-523CRCT-ND          RC0805FR-07523RL        0805 523ohm, 0.1%                   3
Precision resistor  311-1.05KCRCT-ND        RC0805FR-071K05L        0805 1.05K, 0.1%                    3
Precision resistor  311-2.10KCRCT-ND        RC0805FR-072K1L         0805 2.1K, 0.1%                     3
Precision resistor  311-4.22KCRCT-ND        RC0805FR-074K22L        0805 4.22K, 0.1%                    3
LED                 755-SML-211UTT86        SML-211UTT86            0805 2ma                            1
JTAG Header         WM17469-ND              Molex 87831-1420        2x7, 2mm, straight, header          1
VGA DIP Header      609-3337-ND             68021-116HLF            2x16, right 0.1 header, gold        1
User Header         855-M22-2520405         M22-2520405             DIL 4x2 2MM pin header              1
User Jumper         952-1305-ND             M22-1900005             Jumper, gold, black                 1
User Jumper         952-1307-ND             M22-1910005             Jumper, gold, blue                  1
User Jumper         952-1309-ND             M22-1920005             Jumper, gold, red                   1
User Jumper         952-2199-ND             M22-1930005             Jumper, gold, gray                  1
PCB PINS            ED90328-ND              3157-0-00-15-00-00-03-0 Press Fit                           23
Alt PCB PINS        575-641591              342-10-164-00-591000    SIP header, break away 64-pins      1
VGA Cable           http://www.pccables.com #07129                  VGA HD15F to IDC16                  1

Exact cap and resistor part numbers do not matter as long as the specs are correct, i.e. capacitance, ESR (X7R or X5R), voltage spec, resistance, wattage, etc.

 

I always figured the original F18A might become a DIY kind of project, and after the MK2 is done I was planning on reworking the schematic in KiCAD and releasing it as open hardware/software.  However, the FPGA is still end-of-life, so I don't know how viable it will be as a DIY project.  I thought about using the Spartan-6, but all the small packages are BGA and I figured that was not doable for most hobbyists.  So, if someone (I'm not going to take time away from the MK2 right now) wants to find a suitable QFP FPGA, it would really help move the original F18A design towards the open-hardware goal.

Edited February 15, 2021 by matthew180

[Hans23]

Thank you for the writeup, Matthew!  The BOM is particularly helpful.

 

The flip pins look neat, and they're certainly priced better than the Mill-Max pins.

 

Thanks for the pointer to the Phoenix repository, I will enjoy finding out what the F18A actually does

 

Cheers,

Hans

[GDMike]

Thx dudes for taking time ?

[matthew180]

On 2/10/2021 at 5:55 PM, Hans23 said:

When the boards arrived, I found that a track was missing in the layout, visibly.  The track was to one of the enable pins of a voltage regulator and the schematics called for it to be there, so I went for a patch:

That track should not be missing, and could be an error in the project when it was imported into Circuit Maker.  When I get some time that won't take away from getting the MK2 out, I will get the original gerbers around and post them on my github.

 

Quote

... I will enjoy finding out what the F18A actually does

 

You can also ask questions.  I'm happy to discuss any details about the design and such, and there is a lot of technical information posted in the forums.

Edited February 16, 2021 by matthew180

[Asmusr]

With all this interest in F18A I feel inspired to do something new for the F18A.

 

[GDMike]

5 minutes ago, Asmusr said:

With all this interest in F18A I feel inspired to do something new for the F18A.

 

Dual monitors..

[Asmusr]

22 minutes ago, GDMike said:

Dual monitors..

New software is what I mean. ?

[wierd_w]

(muses impotently about a "fully-discrete, but humongous" F18A implementation)

 

There's a surprising amount of room..  Especially if you route it out the sideport like the oldskool 80col card. A fully discrete F18A-as-sidecar (and thus, remove the dependence on the EOL FPGA) would be interesting, if hyper niche.

[+Ksarul]

8 hours ago, wierd_w said:

(muses impotently about a "fully-discrete, but humongous" F18A implementation)

 

There's a surprising amount of room..  Especially if you route it out the sideport like the oldskool 80col card. A fully discrete F18A-as-sidecar (and thus, remove the dependence on the EOL FPGA) would be interesting, if hyper niche.

In that mode, you could just resurrect the design for the TIM. It works as an Analog RGB 80-column card using a Yamaha V9958 and gives you up to 192K of video memory. It wouldn't give you VGA output, which is the real advantage of the F18A, but it would be a discrete component device that would easily fit into the keyboard case.

[RickyDean]

41 minutes ago, Ksarul said:

In that mode, you could just resurrect the design for the TIM. It works as an Analog RGB 80-column card using a Yamaha V9958 and gives you up to 192K of video memory. It wouldn't give you VGA output, which is the real advantage of the F18A, but it would be a discrete component device that would easily fit into the keyboard case.

That's pretty much what @fabrice montupetdid with his Tiny99.

[matthew180]

If TI had designed the 99/4A hardware ports to be specified via a table, like the MSX, then you would add a new video device without having to write new software.

 

The problem with making a "new video device" is, that's a lot of effort to have zero support.  You might as well just make your own custom computer at that point.  But this is a typical conversation for the retro community.  These days it is never a question of "what *can* we do", but rather a questions of "what *should* we do"?

 

I have thought many times about making an external video card for the 99/4A, so you could have a dual-monitor 99/4A, etc.  But without software, what's the use?  And it has been proven that I'm really bad at getting around to writing software.  So, it will just be something I think about, and I can enjoy it virtually.

 

On 2/17/2021 at 4:56 PM, Asmusr said:

With all this interest in F18A I feel inspired to do something new for the F18A.

 

It is always really fun to see what you come up with!

[Jimhearne]

I thought i would update my F18A to V1.9 firmware as it's only V1.4 at the moment.

I've got a Xilinx programmer from a few year back and tried to use the virtual machine version of Xilinx ISE 14.7 that came with it.

That doesn't work, as soon as i do scan chain in Impact it just closes Impact.

Tried updating Oraclebox to the latest versions but didn't help.

 

So i thought i would try the non VM versions of ISE14.7 which can be made to work with Windows 10.

But the Xilinx download site has been down for maintenance for the last week or more now.

 

Does anybody have a copy that i can download,  ISE 14.7 but not the VM version.

 

Many thanks,

 

Jim

 

 

 

 

[Tursi]

You could run the software updater. It's in the 1.9 zip file from Matt's page: https://dnotq.io/f18a/f18a.html

 

You can run it on the TI from the CF card (NanoPeb, CF7, etc) with the DSK image, FinalGROM or burn-your-own cart with the 256k or 512k ROM images, or ColecoVision via the Atarimax cart or burn-your-own Megacart with the Coleco.ROM image.

 

 

 

[Jimhearne]

There is one rather fatal flaw with that plan, i don't have any of those machines, my F18A is in a Powertran Cortex.

 

Jim