What's new
By:

In this thread we build the projects that others have shared - TD-IO (JVS to JAMMA)

hatmoose

hatmoose

Enlightened
Joined
Mar 25, 2021
Messages
1,333
Reaction score
2,781
Location
New Zealand
I've been wanting to mess round with Naomi hardware for a while now, but the whole "massive rats nest" of cables to set it up put me off. But not any more! TD-IO to the rescue

Thomas Daede makes amazing things like the TD-CRT project which is an open-source reimplementation of a CRT chassis with every single cool CRT feature ever invented.

He also makes other cool projects like the TD-IO which is a JVS to JAMMA adapter. As well as being zero latency it has a number of very cool features such as
  • Chainable JVS I/O implementing a 2L12B panel and coin inputs
  • JVS to JAMMA video converter compatible with all monitors
  • Mono/stereo audio amplifier for driving JAMMA speakers
  • JVS power connector providing 3.3V via an on-board converter
  • High reliability, low heat generation and all solid state caps
  • Upgradeable firmware
The whole project, hardware and software is available here
https://github.com/tdaede/td-io
If you want to to buy one premade I found this site
https://birdlandarcade.com/products/td-io

A number of the features are very attractive to me; For example I hope that the built in 3.3v output means I could potentially run a Naomi with my current Minigun + RT125a combo rather than having to mess around with Axunworks or a Sun power supply.

Screen Shot 2023-05-19 at 8.58.11 PM.png Screen Shot 2023-05-19 at 9.08.13 PM.png

Not sure if anyone else has looked at these, or has an interest? If so lets work together to share our experiences!
 
Last edited:
I'm imagining that these will have some kind of case on them, probably in the style of the Minigun "Pacifier" case.

So I'll adjust the JAMMA edge and change the footprint slightly to accommodate, maybe something like this
Screen Shot 2023-05-20 at 7.44.17 PM.pngScreen Shot 2023-05-20 at 7.44.24 PM.png
 
The most recent version of the PCB (v1.4) as some new fuses added to the board but I'm struggling to figure out what they do or why they are there

F1 makes perfect sense - that's incoming +5v from the JVS to the TD-IO (USB1.1 should provide a maximum of 500ma)
F6 makes perfect sense - that's directly from +5 on the JAMMA edge into VSYS (+5v) on the Pi Pico (the official Pi Pico power supply produces a maximum of 3.1amp)

F2,F3,F4,F5 are pins 8 9 J and K on the JAMMA edge which are Meter 1, Meter 2, Lockout 1, Lockout 2.
Screen Shot 2023-05-20 at 9.53.48 PM.png

The schematic shows that they are absolutely doing "stuff"
Screen Shot 2023-05-20 at 9.54.26 PM.png

But I'm struggling to figure out what stuff they are here for, why it needs such lavish protection when nothing else does, and what this part of the circuit actually does when all the JAMMA looms ive seen (and the Minigun, and the pacifier) don't have these pins wired at all

Does anyone have any thoughts on what this is, what it does, and why its here?
 
Last edited:
What got me down the Meter 1 Meter 2 Lockout 1 Lockout 2 rabbit hole was selecting a suitable fuse for them.

But before I could do that I took a gentle detour into the vestigial organs of the JAMMA spec that don't get used much.

For the meter I read a whole load of stuff and eventually landed (where I always do) with the real Bob Roberts
http://www.therealbobroberts.net/meter.html
F2+F3 = Either 5v or 12v depending on the coin counter - so that's going to be 12v

For the lockout I read a bunch more stuff and eventually landed here
https://wiki.neogeodev.org/index.php?title=Coin_lockout
F4+F5 = 12v because all coils are 12v

The fuses are in front of a pair of AO4818B's that are good for 30V and 8A each - I thought this was absurdly over-spec and extravagantly over engineered for a coin counter until I did some research and realised that everything I knew about Mosfet arrays (not much to begin with!) was 30 years out of date and that 8A was now the lowest spec on the market. Progress!

The fuses are also protecting a pair of MMBZ18VALT1G that are good for 25v and 1.6a each - so that's going to be what we need to protect with the fuse. This makes even less sense to me because a coil can easily pull 7a or 8a.
https://www.digikey.co.nz/en/products/detail/onsemi/MMBZ18VALT1G/1749023

Anyway, the point is that 1.6a is what we have to protect, so...
F2, F3, F4, F5 = 12A / 1.6A
F1 = 6v/1.25A
F6 = 6v/3A

The little tiny traces on the board (.4mm) felt a bit skinny when coils were involved so I fattened them up to 1mm.
Screen Shot 2023-05-21 at 12.25.05 AM.png
On the absurdly tiny off-chance that someone actually does have lockout coils and the even tinier off-chance that one fails closed those nice fat traces should give the fuse time to do its thing. It's probably over-cooked but I kind of like how they look, it was fun practise, so I'll leave them at 1mm for now
https://www.4pcb.com/trace-width-calculator.html

Which after three hours of pleasant times-wasting brings me back to "what does this part of the thing do? I'm not really too sure."
 
Last edited:
Thinking about picking one of these up and wanted to see what your experience has been. Capcom I/O boards are expensive as add a frame of lag and the Sega 838 I/O boards, while affordable, don’t have a straightforward way to implement a kick harness in a New Astro City cabinet. The TD-IO looks to be a great replacement for the Capcom I/O without the lag.

Thanks for the great write up.
 
That's as far as I got to be honest - so I can't share my experience yet. My use case is going to be a little different (glorified supergun).
Going to be distracted for quite a while making CPS2 digiAV boards now that I have managed to find some (expensive!) FPGA's.
 
Sorry to necrobump but I was starting to put the TD-io together and thought I may post my progress here also. Maybe @hatmoose may even get back to finishing one too.

The fuse info actually saved me a bit of research so thanks.

One problem I have hit so far is with U5 which is an sic437a buck converter (0.6mm pitch). You must either use green solder mask and reduce a pad or 2 by 1mil to allow an appropriate mask dam width between pads or redesign with gang relief. I took option 1.

Still looking into what I have on hand a what I need to order but it's starting to look like it may be difficult to find everthing in a single place.

So far it's been a fun project. I'll update when I have more information.
 
A few things to update here:
There are no orientation dots on the pcb for anything. This is not unusual for a commercial product but can make putting one together by hand a pain. From what I have looked at all packages are aligned such that the writing is the right way up when the JAMMA edge is at the bottom regardless of where the actual orientation dot is on the IC. The exception is a few ICs that are aligned 90 degrees to the pcb and these are readable starting from the JAMMA edge direction.
The second point of note is be really careful planning your way around the amp if you are doing this majority with an iron. The large 50v caps attach a bit like the smd inductors in that you have to flow under the with only a tiny bit of pad exposed. This is not a problem unless you have installed the smaller caps above them first. I did and I had a hell of a time actually getting to the pad as there is now a speed hump in your way. I typically work flattest package to tallest, if I had of been on the ball I probably would have picked this up but I didn't.

I did the bottom pad of the amp by tinning the pad first and then hitting it with hot air because I didn't order a stencil. This was easy and once you have correct alignment the legs are easy. Because there are 4 vias under the amp you don't even need to be that careful with how much you tin the pad as most excess will flow down the vias (within reason). Same technique for the tiny no feet buck converter package (sic437a) I tinned everything and hit it with hot air. Again you can over tin some of the pads with vias but many of these don't so be careful as the spacing between pads here is so tight that the board could only be produced with a green solder mask without going to a gang mask arrangement.

That's my update I have tried buzzing out the sic and I think it's all connected properly but the final test will be what is coming out of the jvs power plug.
 
I have sent a few of these boards to people so maybe, if we are lucky they will also post their experiences.

One other thing to note is that I only saw @hatmoose post about enlarged traces after I ordered the boards, so unless the github has been updated it is probable that my pcbs don't include this. Also I have gone with @hatmoose recommendation of fuse values. Looking at the Birdcage images it looks like they have used the 3a across all polyfuse locations (its a bit hard to tell because the image is not that clear). Given that @hatmoose put in the work to check what was required I have just gone with that as it probably would have taken me days to get to a similar outcome, but I thought it was worth mentioning.
 
Big_P Big_p said:
I have sent a few of these boards to people so maybe, if we are lucky they will also post their experiences.

One other thing to note is that I only saw @hatmoose post about enlarged traces after I ordered the boards, so unless the github has been updated it is probable that my pcbs don't include this. Also I have gone with @hatmoose recommendation of fuse values. Looking at the Birdcage images it looks like they have used the 3a across all polyfuse locations (its a bit hard to tell because the image is not that clear). Given that @hatmoose put in the work to check what was required I have just gone with that as it probably would have taken me days to get to a similar outcome, but I thought it was worth mentioning.
Click to expand...
Fun!

There is a (now closed) GitHub issue on the original repo as well. that confirms the fuse values.
 
Great. I was doing a little more tonight and there are orientation lines on the pcb. My eyes are just not as good as they used to be.
 
Working my way through this build. I ordered the wrong size inductor for the power side and the vga header I had on hand while fitting the pins was not long enough to reach past the edge of the pcb. There were also a few unusual resistor values on the power side which I didn't have on hand and had to order.

I am not really happy with the rca solution either and will look into other options.

I had a bit of a problem soldering the J lead switches for the coin chutes. Typically I don't find these difficult but they sit so close to each other that the middle inside pad on the last one placed is really difficult to reach with even my smallest longest tip. Soldering that pad first I ended up with a leaning switch and second a little bit of plastic burn. Ideas?

I have 3 boards in production so hopefully I end up with the 2 I need from this and maybe a spare. Hopefully I also work out these practical problems and can put one together for anyone interested trying this out too.
 

Attachments

  • 20241107_182524.jpg
    20241107_182524.jpg
    231.9 KB · Views: 78
I grabbed a couple of these from @Big_P and am working on getting these built up

Have just ordered all of the ancillaries from Aliexpress (all of the non JST/Hirose stuff) and working on a Digikey cart for everything else

If anyone in Australia needs RS485 connectors (A and B type), switches, or a volume knob, I will likely have spares
 
Thank you, for some reason when I looked up the fuses section that package size wasn't even listed - didn't realise they are PTC fuses

Edit: And Digikey order done. I also realized I ordered the wrong volume pots so will have to order those again
 
Last edited:
My Digikey order arrived and I built up most of one of these today (waiting for the remaining parts from Aliexpress) - I ordered a few incorrect ICs and will need to reorder but most seems to be all good

Unfortunately I am not getting any 3.3v and this is most likely related to some damage underneath the SIC437 - this board may be a failure. Trying to get that little buck converter down was extremely difficult and when I used some braid to try and clean up the pads I lost some of the solder mask on the board. All of the pins on that side look to be going to the inductor though so I wouldn't have thought it would make too much of a difference

For those who are looking at building these - there is a very heavy ground plane. You will need a lot of heat to get the power connector soldered down, be patient

Still going to finish building the board for the experience if nothing else, I'm really happy with how it has turned out and I think everything other than the 3.3v will work fine so it will just mean running a separate 3.3v PSU like a Capcom or Sega IO

Edit: picture as it looked right before I added the Hirose and VL connectors
 

Attachments

  • IMG_8232.jpeg
    IMG_8232.jpeg
    134.6 KB · Views: 41
I am still waiting on the power inductor to test it out, but that little QFN package was a #~<#er. I had a similar issue on my first go at putting it down. I got the thing tacked down and was pretty happy that the pads below were connected but when I went to flow the side near the inductor I over heated with the iron and lifted mask, really similar result. I probably should have warned everyone but I was feeling stupid and out of practice. If you need another board just let me know. I still have a few spares and l am just happy to see them being used
plus feeling bad that my pride stopped me warning people, so no need for another donation.

You are 100% correct in that the ground plane is heavy but I suspect that the solder mask at that point may also be less than ideal. Getting the pcb to pass manufacturing with nextpcb required a little bit of adjustment of the pads of that package as they were too close together for individual masking even with green mask. The other option was going to a gang mask but I couldn't see that being any easier. The close up photo may help. The bits circled are independent of the plane you may be able to touch it up (but typically I find I just make it worse when I do that)

The fuse at the bottom of that plane can give similar problems as it takes a bit of heat to get it to flow. I caught it in time but it was close.

My advice would be to remove R42 to R45 before you attempt to put the switches in above them. I did a similar thing, then removed them and still had a few issues getting to the middle leg of the second switch put down. If you find a way to do this let me know. I tried to straighten it with tape and hot air but ended up with a melted switch end. I think to lift it i will need to tack a thick copper wire in a 'U' around to the legs and use the big iron on a bit of the wire pointed up (it took me a while to come up with this plan)

Thanks heaps for sharing your build. I have found it a fun project. It definitely pushed me to the edge of my soldering skills at a few points but that was offset by plenty of stuff I could do easily if I was feeling stressed about it. It was great to see that you had many of the same issues I had. I was feeling a little stupid when I lifted the mask on my first go at the SIC, hopefully it works both ways.
 

Attachments

  • 20241119_135652.jpg
    20241119_135652.jpg
    177.7 KB · Views: 34
Thank you for the offer of the replacement board. I’m going to remove the chip and re-tin the pads to see if I can get it to work. I still built the rest of the board just for the experience. Interested to see if everything else works
 
You must log in or register to reply here.
Back
Top