tabm0de wrote on 2024-10-14, 06:21:

Nice project, it it possible to purchase board? 😉

Possibly, I sent you a message.

Super preliminary, but I got the testing platform for the ES1869F chips working! Super exciting! I need to perform some testing for sound quality but the reference design was easy to follow and it sounds pretty good so far. I will do a full test of stuff like the joystick port when I get some time.

Of course, I also received the next prototype PCB. I can't wait to start building one. Will probably have time this weekend. 😁 Things are looking good!

Pretty cool!

Behold!! Backwards V0.4!

There are a few minor issues that will need a V0.5 revision, but it's 95% functioning.

The first issue was the use of the opamp as an output buffer. I should have tested that better in hindsight but the fix was possible with a bodge. Because I'm powering with +/-12V in unity gain non-inverting, I ended up with an unacceptable amount of DC offset. I'm sure this was a consequence of the layout and routing, but essentially there was no ground reference for the opamp in the signal path. This led to a rather large DC offset on the positive inputs even after the rather large decoupling caps inline from the ES1869F. The simple fix was a 10k from the positive input to analog ground after the decoupling caps.

The second issue is a mistake on the joystick layout! I didn't test this but found the miswiring of the X and Y components for both joystick ports. Basically, I have the inline resistor on the wrong side of the cap/resistor filter according to the reference design. I might be able to bodge this but a fix is needed in the layout since the traces are on the inner copper layers. D'oh!🤬

I also experienced some crackles on the auto. The fix was to reflow the ES1869F and everything is happy.

The clearance of things like the stereo jack and vertical wavetable header turned out as expected. The Wavetale PI thingy does look weird hanging off the back, but it performs as expected.

Along with the vintage 7-seg drivers, I sourced some vintage red 7-seg LEDs! The overall aesthetic is a bit strange, but I like how it turned out. What I really wanted was a full custom solder mask color for these. Think bright orange! That would be awesome. I think JLC or PCBWAY offers short-run custom solder mask colors with some kind of inkjet-type machines.

Getting there!
How much are the prototype boards to make btw?

I have some ideas I want to explore but it looks like I'll be failing a lot before I learn how to make it work.

I should be doing a better job keeping track of the cost but I'm guessing the total parts cost to make one prototype is $80-100. I'm not really designing for cost though. I'm only ordering 5 PCBs at a time and they cost about $15 each. I'm sure it could be much cheaper if I slimmed down the design and ordered in larger quantities but there's no real plan here. I'm mostly doing this for fun... 😁 Since I'm making these by hand it took about 6 hours to build!

Backwards V0.5

After some issues with the V0.4 board, I've decided on a complete re-route. This is mainly due to some instability experienced with communicating with some ISA cards.

After some probing, I can see that my poor routing on the ISA lines is the most probable cause. In comparison, the PCI lines were stable.

The result of my poor routing to integrate the sound chip resulted in many of the ISA lines being forked as the ISA slot.

Here's an example of ISA ADDR5. You can see it's forked in the middle of the board.

The oscilloscope probe shows a rather severe overshoot in the negative voltage. This is while probing the Allen Bradley card on ISA DATA0.

Another thing I'm discovering is the different SBCs have different IO drive strengths. This is the probe on ISA DATA0 on a PCISA C400 vs one of the Allen Bradley socket 7 boards. There's a difference here and I believe my routing is causing the instability in some of my SBCs combined with the difference in drive strength. However, this is an assumption on my part.

I initially tried a voltage divider termination which was outlined here. https://www.smbaker.com/diagnosing-xi-8088-stability-issues. My crud addition actually fixed the instability in communicating with my ISA sound cards. The results from this caused me to rethink and rewire everything in favor of longer trace runs and reduce any forking as much as possible.

With a complete rewire, some of the traces are quite long. The ISA IOW in the new design is nearly 300mm for example! Because of this, I've decided to add both series and pullup termination on all the ISA lines. The series termination happens between the SBC slot and the ISA slot. I will need to test more once I get the new boards, but I feel much better removing all the forked lines along with the termination that solved the communication issues. All the termination resisters will be located on the back side of the PCB.

I also made two changes to the design. I added the supported IDE CD-ROM channel to the ES1869F chip. I had to increase the length of the board by 7mm, but I was able to fit the connector as well as the buffer chip for the 8 ISA DATA lines.

I've also redesigned the PCI mapper card and moved it clear of the PCI slots on the far right side. The design now uses a single reversible card that has both configurations! This uses a PCIE 1x slot and should be about the same total cost vs the pin header and socket version.

I should be able to order this design soon, but it might be a while before I can test it because of the holiday weekend.

Backwards V0.5

Didn't want to put the sound chip at the end of the pci or underneath for straighter lines?

I found some smd pci slots if it helps.
https://www.sullinscorp.com/products/?f ... &toggle=in

I honestly thought isa was so basic and slow that it'd handle almost anything 😒

What you links were PCIE slots... I'm in fact going to use a 1x surface mount slot for the mapper card, but I don't think anyone made an older surface mount PCI slot.

I'm okay with the sound chip being on top though. I like looking at it! 🫣😁

"I honestly thought isa was so basic and slow that it'd handle almost anything 😒"

Well... I'm sure it's robust enough, but the engineering couldn't foresee my poor routing. I'm happy with how this version turned out though. And the termination is probably warranted with trace lengths >250mm.

No... The link with filters didn't work properly.
Try this:
That's only one of them.
https://drawings-pdf.s3.amazonaws.com/C10858.pdf

Oh, how interesting!

It's bedtime for me, but I'll look closer at it tomorrow. I wonder if they're still being made and how expensive they are??

Backwards V0.5

Short update: Newest prototype boards worked out great. Only one minor bodge, but everything works. Yay! The CD-ROM channel of the ES1869F works as expected, although, getting the DOS side of the CD-ROM driver working seems to be an issue with some of my SBCs. The Allen Bradley cards seem to be the most restrictive and troublesome when it comes to IRQ and DMA management. My PCISA cards are much more forgiving. The ISA termination appears to be doing its job. Need to test some more, but things look better on the scope. The reversible mapper card also works as expected and is a slightly cheaper option than the pin heads.

While this could be a V1.0 board, I may adjust the mounting holes. Right now the ATX mounting looks good. The intention for mounting the backplane, other than something obviously custom, is to have it mountable in a full ATX case as well as an ITX motherboard in its normal location. This fits as expected, but I may add a second set of holes to move the board one more space lower to accommodate a double-spaced PCIE card in the ITX. This requires some wires to be moved, but the clearance on chips looks fine. This may be a niche usage case and it will grow the board by 5mm, but it might be worth it in the long run. You can see here how the backplane fits into the case.

Damn looks great!
I'd definitely move it down one slot

This looks fantastic. Please let me know when you decide to sell boards. I’ll certainly buy at least a couple of them.

- Alex

myne wrote on 2024-12-08, 16:18:

Damn looks great!
I'd definitely move it down one slot

It's something I should have added in the first place, but I was designing it originally to fit within a micro ATX case. Now that I've mounted this, I can see it really should be one slot down to take advantage of a 7 slot case and leave room for an ITX with a larger graphics card. Not that this would be every use case, but that seems to be a common arrangement for home users.

For me, this backplane is going into my Retro Rig with a custom mounting location. It uses parts from an open-frame ATX-style aluminum chassis like this. These style cases are fully adjustable because of their construction and would work great for a stand-alone retro machine.

ajacocks wrote on 2024-12-08, 22:32:

This looks fantastic. Please let me know when you decide to sell boards. I’ll certainly buy at least a couple of them.

- Alex

tabm0de wrote on 2024-10-14, 06:21:

Nice project, it it possible to purchase board? 😉

So... Now that I'm happy with how it's turning out, where do I go from here?

I would like to sell a few of these but I'm not sure how to go about it. What I really need is some honest feedback from others who have SBCs to test with. At its core, this backplane is a purely passive device and would function just fine with only the ATX power and card slots. However, I think I would benefit from someone else having some hands-on time with one of these 0.5V boards before I make any final changes to a V1.0.

This might be a bit presumptuous, but I need a few volunteers who have compatible SBCs to help me test...

Full disclosure, I'm making these boards fully by hand and they take about 5 hours to assemble. In the prototype quantities I'm ordering, they cost about $80 to make. Whether or not I can shave the cost a bit will determine the final price I might have to charge. The price can easily vary for instance between a basic passive version or a fully loaded one with all the active components. The final quality of these boards that I'm able to produce will be determined by the current availability of some of the parts which will ultimately be limited to small number. I've already stockpiled some stuff so I can make a few dozen right now.

Okay then...

For whoever is interested, I need two or three volunteers to test this V0.5 backplane. If you want to DM me here on Vogons, first send me a message of your interest in participating and a photo of one or more compatible PISA/PCISA, Allen Bradley, or ISA "half-sized" SBC along with the model number so I can verify what you'll be testing with. Within the next week, I'll have a few of the V0.5 boards ready to test with. I will provide these prototypes for free, but I'd ask that the volunteers provide the shipping costs. I can't promise this will move quickly as I do this as a hobby in my free time, but I will continue to make forward progress.

The goal is to have honest feedback about functionality and possible changes for a final version and what should be included in the documentation for the end user. Thx ahead of time for anyone that's interested in helping out.

Honestly...

It looks like you counted the slots, counted 5 and then started the pcb from the top. Forgetting of course that one isa/pci is shared.

That's the kind of mistake I'd make.

Dunno if this will help
Re: The thing no one asked for: KICAD 440bx reference schematic

You might appreciate this too.
Just finished!
Convert old ASUS ASC boardviews to KICAD PCB!

Any updates on this?

BTW what takes the most time with assembly?
If it's hand soldering every pin, a pre-heater plate and solder paste might do the job if it can get hot enough.

I made the attached too - which is probably too late to be useful to you, but you might wish it already existed before you started!

Yes still making progress. I'm not in a position to look at the attachment ATM but will later.

I only had one taker for a free prototype. While I'm not relying on feedback for progress, I will not make anymore V0.5 boards right now. This is mostly due to the limited number of slots I currently have. And, I've ordered a preliminary V1.0 with the refinements I wanted. I should be able to assemble one soon.

I have the assembly time down to about 4 hours. This could be much shorter, and it's mostly due to my non-optimized assembly process. I'm hand soldering the roughly 700 through holes on the bench. I've tried to "cheat" with some solder paste but it didn't turn out well. I also tried solder paste and hot air reflow on the back, but the old EISA slots don't tolerate the heat well and the alignment pegs melted under my hot air.

A pre-warmer might help with the hand soldering since some of the pins are sinking large amounts of heat to the ground planes.

The process right now is, solder paste hot plate reflow for the top SMDs, solder paste and hot air reflow the back side SMDs, and finally hand solder the slots and through hole parts. It's getting smoother but there's time that could be saved if I really wanted to optimize the process.