PC Engine 6 button controller

[pbj]

I have a Fluke multimeter and THAT IS IT.   

Been out of town this week, I'll dink around with this some more in the next few days.  I am borderline tempted to buy a PC Engine and have it sent to you.  All reason has left the building on this quest.   

They're not all that expensive, though, and the flashcart will set you back about $30 USD. 

 

[mourix]

I'll leave it up to you, haha. This is a bit of fun for me too so I don't mind ordering some PCB's as well.

My problem is PC engines being incredibly rare in The Netherlands so I'd have to import one from Japan which means big dollars.

[Malenko]

https://www.ebay.com/itm/385894281500

$125 free shipping and it comes with a game.

[lilshawn]

I'll leave it up to you, haha. This is a bit of fun for me too so I don't mind ordering some PCB's as well.

My problem is PC engines being incredibly rare in The Netherlands so I'd have to import one from Japan which means big dollars.

videogames are quite the opposite in japan. the market is so saturated with used games and systems that they can be had so so SOOOO cheap. there are shops where you can literally buy used snes games loose for under 1000 yen (about 6 bucks USD) each.

i can find tons of copies of the Japanese version of banjo kazooie for just over 10 bucks WITH free shipping. same game, north America version is topping 60 with shipping on top of that. call it 80.

hit up ebay. you can get pc engine... japan console... with controller and power supply with everything tested guaranteed working... 100 bucks.

NA "turbographics 16" system seller... untested, only console. nothing but the system so... probably broken and needs repair... for 250 with the shipping on top of that. (depending where you are could be about 50 bucks more.)

even with netherland bucks conversion, japanese consoles are still dirt cheap.

[pbj]

Well, TG16 was dead on arrival in the US, so it's going to be QMG LQQK R@RE no matter what you do.  That being said, my last several systems (SNES, Saturn, Dreamcast, Neo Geo, Neo Geo CD, PC Engine) have all been Japanese because nowadays it doesn't make any difference and they were a lot cheaper.  I wasn't really impressed with my experience using a Yahoo Auctions Japan broker.  Buyee?  Some BS like that.  I bought a bunch of PC Engine stuff and they pulled something like $350 out of my bank account for shipping without approval.  After my bank dealt with that fraudulent behavior it was all "forgiveness please, we can ship it FedEx for $80."  I hope they're enjoying that pallet of PC Engines in their warehouse.

I just stick to ebay these days.  Same sellers and auctions anyway.


Back on topic - I put a 100pf (type 101) capacitor on pin 11 of U1 and connected it over to Vcc on pin 16 of the same chip.  This has definitely improved the behavior.  In 3 button mode, I can use it to start up and play a CD game correctly (previously it was slamming select over and over and trying to delete my internal memory).

So, I figured what the heck and I tried a 0.1uF (type 104) capacitor across those same pins and now it's back to strange behavior.  I can get it to spit out select and run inputs and that's about it.

Then I tried a 220 Ohm resistor between pins 11 and 16.  It's back to the same behavior as if there's nothing there.

So I think I'm going along the right lines here..?

 












 

[lilshawn]

just a thought... i haven't perused your schematic much but your circuits behavior seems to indicate a floating or otherwise not properly pulled input causing random behavior.

typically in circuits, you tie an input line high through a 10k pull-up resistor to drive it definite high... then yank it low through your switch. i see some 47k resistors, perhaps the value is too high and not driving up the input of the chip high enough above the 0/1 switch point threshold.

check the Vil(max) and Vih(min) of your chips datasheets to make sure your high and low states with a voltmeter and see if it's within the ranges specified. (typical 74 series logic has a "0" in the 0.8v and under range and a "1" in the 2.0v and up range... and anything between 0.8 and 2.0 produces 0/1 random gitch switchingness)

something worth looking into anyways.

[pbj]

I had another electronics engineer look over the circuit.  I’ve only been feeding and inebriating his ass for 25 years.  His comment was similar - the 47k pull ups seem too high.  And he thinks my capacitor addition is filtering noise.  His suggestion was a 10k array.  Perhaps those circuits were all 4.7k and nobody noticed the typo for 35 years?

I tried another capacitor value while I had the box of them out - 220pF and it was the same behavior as 100pF.

I did some play testing tonight.  Some games are 100% playable in 2 button mode - Dragons Curse, Alien Crush, Devils Crush. 

Some will immediately throw a select input when the level starts, but otherwise play normally - Shinobi, Rondo of Blood.

SF2 plays fine, but it brings up a “define buttons” screen before each match and also skips the between round trash talk scenes.  Just like smashing the select button would do.  In both 2 and six button mode.

I did notice on most games that pressing select made the character have a down input.

I am so tantalizing close on this one…. But even as-is I’m having fun with it.

 

[pbj]

Well, I tried a 10k resistor on select.  No change.  Then I added the 100pF capacitor.  No change. 

 

[lilshawn]

honestly, I would go down as low as 1k on that circuit. you aren't worrying about fighting another circuit pulling it high/low there, might as well pull hard.

I've gone as low as 470 ohms on some projects just to get the Vih(min) high enough that it doesn't randomly trip the input. (especially on a microcontroller's unused reset line which benefits from being driven high to keep it from randomly resetting due to shenanigans.)

[mourix]

I've been chasing the resistor arrays for the most of these posts. 47k makes sense for short pcb traces but for long wires it may not.

1k too low though. Every button you press would pull 5ma from the PC engine which adds up if you push many at the same time. If 4.7K does not work and 1K does, there are different issues than your pull value.

[Mike A]

I had another electronics engineer look over the circuit.  I’ve only been feeding and inebriating his ass for 25 years.  His comment was similar - the 47k pull ups seem too high.  And he thinks my capacitor addition is filtering noise.  His suggestion was a 10k array.  Perhaps those circuits were all 4.7k and nobody noticed the typo for 35 years?

I tried another capacitor value while I had the box of them out - 220pF and it was the same behavior as 100pF.

I did some play testing tonight.  Some games are 100% playable in 2 button mode - Dragons Curse, Alien Crush, Devils Crush. 

Some will immediately throw a select input when the level starts, but otherwise play normally - Shinobi, Rondo of Blood.

SF2 plays fine, but it brings up a “define buttons” screen before each match and also skips the between round trash talk scenes.  Just like smashing the select button would do.  In both 2 and six button mode.

I did notice on most games that pressing select made the character have a down input.

I am so tantalizing close on this one…. But even as-is I’m having fun with it.

 

Keep it up. You will get it.

This is good stuff.

[pbj]

Just wanted to mention, same behavior on both assembled boards.  They both throw a Select when plugged in, even if I don’t have anything connected to the input pins.  If I turn the system on with the board already plugged in, I’m not seemingly getting that input but it may just be the system booting up.

I can leave it at the CD rom menu as long as I want, and it’ll never read a select input until I actually press it (it jumps to memory management instantly when you hit select). 

Don’t know what other testing I can do at this point….

[lilshawn]

a logic probe would really help you out here.

[pbj]

I do actually have a logic probe.  Can you type very slowly and tell me what I should do with it? 

[baritonomarchetto]

Could you reassume the situation so far? What is working and what is not?

[mourix]

Maybe it's a good thing to go put select directly to VCC with a wire to see if it still triggers accidentally.

[pbj]

Well, if I put select directly to Vcc, and then I ground out select, the console scrambles the graphics and freezes up, so that is a no go.  If I just leave it connected to Vcc and wait, after a few seconds the system registers a phantom select press.  I still get a select input when I plug in the controller cable with the system on.

[lilshawn]

I do actually have a logic probe.  Can you type very slowly and tell me what I should do with it? 

sorry, i didnt see this post. i can help you out here a bit.

you probe should have some power lead on it. (one or two, either a power and ground, or just ground) hook this up to your circuit.

your probe should have a setting for TTL/CMOS. (For TTL it is 4.75 V to 5.25 V while for CMOS it ranges between 0 to 1/3 VDD for a "low" level signal and 2/3VDD to VDD for a "high" level signal. set your probe for CMOS mode.

your probe may or may not have a switch for "pulse" or "mem" or the like. this will just illuminate an led for if it senses a logic level change that is extremely short. it will lock on a light or have it flash or something like that. you usually have to apply the probe to the circuit, then turn on pulse mode to keep from it being triggered by attaching it.

then just attach your probe tip to your circuit inputs and outputs to see where your weird pulses are coming from. make sure your button presses are properly registering as high and low and not "both"

couldn't hurt to check to see if your +5 vcc coming in to your board doesn't have a pile of ripple on it. measure it with your voltmeter set to AC. it should be less than 20mv. if it's higher, you have some bad power filtering somewhere in the console sending dirty power.

Well, if I put select directly to Vcc, and then I ground out select, the console scrambles the graphics and freezes up, so that is a no go.

kek, you essentially shorted your VCC to ground if you don't do it through a resistor. not sure why mourix suggested this.. maybe thinking your board had a seperate or different supply maybe?

[pbj]

Reading 1.5 - 2.6 mV AC.  Readings didn't do much even when loading large roms or booting a CD.  Oddly, even with the system off I'm still reading 1.4mV.   

Still digging around for the logic probe.  I bought it back when I was trying to fix some Williams pinball boards with a test rom.  It passed all the tests beautifully but wouldn't boot.  I bought repros.   

[lilshawn]

voltage seems good, not much ripple there so I don't think it's that.

[mourix]

Well, if I put select directly to Vcc, and then I ground out select, the console scrambles the graphics and freezes up, so that is a no go.  If I just leave it connected to Vcc and wait, after a few seconds the system registers a phantom select press.  I still get a select input when I plug in the controller cable with the system on.

The idea was to pull select hard high to rule out any chance for ghost presses by bad wiring or pullups not working, which it did. You just added the short circuit 

[pbj]

Was I supposed to leave it tied to Vcc and then not press it?  Type slowly, I'm begging.

[pdco_arcade]

Hi pbj,
Is the interface working now or is it still producing a select when initially connected ?

[pbj]

Nothings changed.  I throw an occasional bid on a PC Engine to mail it to mourix.  They’re selling crazy high lately.  I’ve played with it a little more.  Tied in Vcc to Ground with a 100pf cap on that first IC.  I notice that hitting select also sends a down input. 

[baritonomarchetto]

Being a simple design, I would suggest to emulate the circuit with a circuit simulator tool like circuitJS1.
This way one could at least see if those schematics makes any sense...

[lilshawn]

i still think your issue is errant signals triggering your IO chip... (see mspaint cad)

if you design the circuit as in the first instance, the inputs and outputs are left "floating" if a button is not pressed... this will cause the input to behave erratically such as ghost pressing and not registering presses. this is caused by all manner of weird things such as static buildup in the air and possibly fast moving particles like radiation and (depending on a circuit's content) possibly photons. i know you got some pullups going on so i dont think it's that... but...

we might need to yank these outputs high or low (depending on the circuit's needs) using a resistor too. (see mspaint cad 2nd half) this puts the inputs/outputs in a known state that we can change so there is a definitive change from a 5v state (high) to a 0v state (low)

im thinking it may be beneficial to drive the output high as well (unless it's an inverter or some kind of nor gate, (you'd have to see how the console's gamepad originally did it) in which case we would also want to drive the output high or low as well (sorry i havent really looked into the circuit too much TBH)

designing a digital circuit is kind of a pain in the ass especially if the digital circuit and the devices it's attached to is picky about how it's being run (as it seems to be in this case) and could really need/want/require the states of the inputs and outputs in a definite high or low state.

if i was having errant signals on the select button, i would hook a logic probe to the IC's output for the select and watch it for blips of signals. my logic probe has a mem/pulse switch... if i switch it to mem, if the state changes at all, it will illuminate the "pulse" LED on it until i reset it, so even if a short or single pulse happened (basically a state change from whatever the base is), the probe would show it by illuminating the led. a couple of alligator clips or a soldered on fly lead attached to the probe tip will make it hands free and able to check the circuit for long periods of time.

use of some circuit simulator would also be beneficial to sanity test the original circuit diagram, but they do have a pretty sharp learning curve to them. i have used "livewire" it's old (meant for pentium computers with 16mb of ram but runs just fine today on win 10), still a fairly capable program and it can be found  for free on the various ---No torrents please - thanks!--- places often bundled with "PCBwizard" which i also use occasionally... since i can export my circuits right out of livewire into it and dont have to redraw them. there are a couple formats they can be saved in that can be opened up in other software and imported to pcbway and the like. its a little clunky...but i like it better than cloud things where you don't own anything anymore if you make it with them.

[baritonomarchetto]

Schematics in first post actually show pulled up inputs: am I missing something obvious? The other inputs need no pullup or pulldown because signals from previous ICs will not float.

If (and only IF) the circuit is ok, next suspect should be ICs. Recently I bought a 10 pcs lot of quad XOR gates. Only three out of four gates (not ICs but GATES on the same IC) were working... You can bet this caused a circuit malfunction. All 10 had the same issue!
This is to say that you cannot trust those AliExpress chips 100%

[pdco_arcade]

Did you find the logic probe?

[Zebidee]

You could try putting one or two larger electrolytic caps (suggest 100uF/22uF to start with, but you might want to try higher-lower variations to see what works best) in as a "bypass" or "filter", across the 5v input and GND. This will help to smooth out variations ("bumpiness") in the 5v supply, and help with a poorer quality PSU. You'd know what I mean by "bumpiness" if you saw the waveform on an oscilloscope. Anyways, nothing terrible will happen if you experiment with some different capacitor values using the PCBs you already have. Just make sure to put them on the right way around (+ to 5v, - to GND). Best to do with an oscilloscope attached so you can see the results.

You already have some 0.1uF caps, but they are only there for local support of the chips they are close to. The electro caps would help with bulk power support.

If want to really go all out, a LDO (low drop out) voltage regulator could be added to further improve power supply stability.

[pdco_arcade]

I think the problem is related to the PCB layout.

Review of the PCB files show nodes J4.5, U1.11, RN1.6, J1.6, U1.1, U3.1, RN3.2 are all connected together.
Clearly this was not the intent.  There should be 2 nets - J4.5, U1.11, RN1.6 and J1.6, U1.1, U3.1, RN3.2.
I think the cause of this is both J1.6 and J4.5 use the label SEL on the schematic.
If the net lister assumes these labels are actual net names then the two net fragments are considered the same net and will be connected together in the net list.
Even though the actual schematic drawing does not have a line drawn between these nets, by having them using the same label/name they are logically connected.
The pcb tool rat's nest will show this connectivity and then routing the pcb from that will make them connected.
A method to correct this is to rename one of the nets.  In the attached schematic the J4.5 signal is renamed to SEL-BTN.  With that the net lister will treat them as different nets.

The attached pcb pdf shows where I think the problem is and make a suggestion about how to resolve the problem.
Would be great if some folks can review the information and verify if I am correct or not.

Cross check me!

[PL1]

Review of the PCB files show nodes J4.5, U1.11, RN1.6, J1.6, U1.1, U3.1, RN3.2 are all connected together.
Clearly this was not the intent.  There should be 2 nets - J4.5, U1.11, RN1.6 and J1.6, U1.1, U3.1, RN3.2.
I think the cause of this is both J1.6 and J4.5 use the label SEL on the schematic.

Nice catch.   

Haven't checked if your recommended PCB mod with trace cuts and jumper wires would make the PCB match the schematic, but your explanation makes sense and all seven nodes being connected to one net (SEL) is shown in the schematic PDF outline under Nets - SEL - Pins.


Scott

[pbj]

Since I've already got two assembled boards and this seemed like a simple enough mod, I went ahead and cut the traces and added the jumper wires.

Only difference in behavior is now it hits the system with a Button I and Select input at the same time when the PCB is plugged in.   SF2 - same deal.  Plays fine but skips the cut scenes and pulls up the input screen before each match.  Strip Fighter everything is fine in the menus, no movement during the game (3 or six button mode).  Everdrive Clone - pulls up the BIOS/Firmware information (same as a Button I input) when initially plugged in but no movement in menus.  CD menu - six button mode - seems to constantly pulse the main menu with Select, which brings up memory management screen.  In 3 button mode it behaves normally.  Rondo, he jumps in the air and does a super attack when plugged in (again, that's select and I at the same time).

So, it's like.... *slightly* different but mostly the same.

Should I throw in a filter cap somewhere?  This is my PCB where I haven't soldered in all those extra filter caps.

[Zebidee]

Should I throw in a filter cap somewhere?  This is my PCB where I haven't soldered in all those extra filter caps.

You need those 100nF/0.1uF/104 ceramic caps to support the four ICs Vcc. They must be physically close to the IC's Vcc pins. Otherwise they might not work right.

Suggest a 100uF cap (or similar) across the 5v and GND landing pads, +'ve to 5v and -'ve to GND. There is a 47uF listed on the schematic, seems to be within appropriate range, give it a go. This is a simple mod so you can just see how it goes and remove/modify the value if needed. Make sure you have the polarity correct.

The caps store and release power. We might call them "filter" caps, and in some ways they are, but in this instance they are more about providing a consistent power supply to the circuit, as needed.

This may or may not solve some/all of your issues, but will help you to eliminate power supply.

[pdco_arcade]

Now type slowly for me.  I do not know the jargon.
Have never owned or played these video game consoles.

via google I do now know the difference between Street Fighter and Strip Fighter!

Yes the power bypass capacitors for each IC is important.  The IC data sheets call for them to be used.

Is the assumption correct that a standard, off the shelf controller does not display the same effects when plugged in with the game console power ON?

And commercial controllers still operate the game console OK?

For now lets forget the events when the controller is plugged hot.  I mean plug in the controller and power up the game console.  And only test 2-3 button mode.  No 6 button mode yet.  Initially had all the configuration jumpers between pins 1 & 2 - so basic operating mode.

How does the PCB work under these conditions?

Would it be possible to assemble another PCB without the extra components for the 6 button operation?  Mainly do not populate U3 & U4. 
Just to keep their loads off the J1.6 SEL input.

Also be possible to verify static operation? To start:
   Not connected to the game console.
   Just connect +5V power & ground to J1.1 & J1.8 respectively. 
   All configuration jumpers ( J2, J3, J5, J6 ) pin1 to pin 2. 
   No connection to J4.
   Measure the outputs with a volt meter ( J1.2, J1.3, J1.4, J1.5)
   
Like this would expect all outputs to be near zero volts.
Now connect J1.7 ( CLR ) to ground.  Would expect all outputs to be near 5V.

Actuate button inputs ( tie to ground ) UP, RIGHT, DOWN, LEFT one at a time and measure all 4 outputs each time.
Here would expect measurements to be:
UP  ( J4.7) -  D0 low, all others high
RIGHT ( J4. - D1 low, all others high
DOWN ( J4.9) - D2 low, all others high
LEFT ( J4.10) - D3 low, all others high
Connect J1.6 ( SEL ) to ground.
Repeat measurements using button inputs I ( J4.3), II ( J4.4), SEL ( J4.5), RUN ( J4.6).

Have you found the logic probe?

[pbj]

Acknowledging your post, I will make another attempt to find my logic probe.  I have the materials to assemble 10 pcbs, so I will make a stripped down version as you’ve requested.  I may finally break down and buy an official six button controller just to compare behavior…

[pbj]

Okay, I assembled a new PCB.  I left off U3 and U4.  I cut the traces as indicated on your revised schematic, and added the jumper wire from pin 1 of U1 to SEL on the output.  J2, J3, J5, and J6 all set to pins 1 and 2 closed.

Everything working perfectly in 2 button mode.  I can control the Everdrive menu and make the morally upstanding women of Strip Fighter move around and fight.

 

So now what?


EDIT:

WOO HOO!

Installed sockets for U3 and U4, figured what the heck.  First I installed U4 - the SN74HC163N.  Games would recognize that I was toggling between 2 and 6 button mode but obviously weren't getting any six button inputs.  Then I stuck in U3. 

With the cuts to the traces and the jumper wires added, I am happy to report I am 100% working in both 2 and 6 button mode in SF2 and Strip Fighter.  Everdrive navigation working in either mode.  I can boot up to the CD menu and not have it jam select inputs over and over.  Rondo doesn't control correctly in six button mode, but he's not getting slammed with random inputs.  Plays fine in 2 button.  (I doubt very highly that game was programmed to work correctly with a six button controller, many of them aren't)

I will go back and double check my work on the previous 2 PCBs... or just bury them in a lonely field for closure and move on with life.


Many thanks to you both, pdco_arcade and mourix.  If I can make one more ask... would either of you be able to revise the PCB layout with the alteration to the Select / SEL lines so I can get it printed without needing jumper wires and cut traces? (It would also be nice if J2/J3 and J5/J6 were slightly spread further apart, I’m having to shave my JST connectors..)

 

[pbj]

Revisited the previous 2 PCBs.  Issue with one was that the trace running from Select to SEL wasn't fully cut.  It wasn't beeping for continuity but I guess there was juuuuuust enough trace remaining to screw things up.  Dragging a knife across that cut a few more times cleared that up.

The other PCB I had directly soldered the cable to the board because my pins and connectors hadn't arrived yet.  Cleaned all that up, which had been causing intermittent issues, and did the same thing with really gouging out the trace cuts. 

All three are operating correctly now.  Really need to invest in some better lighting and magnification at my work station, obviously.

 

[Mike A]

Nice work.

Way to see it through to the end.

 

[pbj]

So now it's time to finish up the joystick.  The Rocksoul had 5 buttons up top mounted to a PCB that was screwed into the underside of the control panel.  I didn't really want to dick around with cutting traces and adding jumper wires, so I just pulled all that out.

The holes leftover are just slightly too small to fit a 16mm button in.  Mike A suggested using a step bit.  So I very, very slowly spun it around by hand to widen each hole to 5/8".  At that point, the buttons can be jammed in.  The end result was clean enough that you could deconvert this joystick back to factory and I don't think you'd really notice anything was different.

I'm waiting on DPDT switches to arrive, the DPST throw won't work for the toggling we need for this PCB.  But I'll have I turbo off/on, II turbo off/on, Slow Motion off/on, and Mode A / B toggles.  The LEDs always stay on with these buttons, so I figure I'll use the 5 button to toggle the 5V line to all the LEDs.  Or just plug it.  Whatever.

That being said, if you tie together the jumpers for mode B (six button), and then add the jumper for mode A, the system will go back to recognizing it as a 2 button controller.  It's inelegant but will work for while I wait for the proper buttons to arrive.

So here it is with a 16mm DPST switch installed to toggle it between 2 and 6 button mode.

[pbj]

And while I’m talking to myself and making feature requests, it would be really really awesome if that 15 pin connector for inputs was split out into two 8 pins.  I cannot find a 15 pin JST plug for the life of me.  Even if the 16th pin did nothing, 8 pin connectors are plentiful and cheap.  As it is, I’m going to have to take a bench grinder to the ones I have to get the edges narrow enough to line them up.