Zikzak - crummiest 8bit console/computer ever .. but I'm making it!

[skeezix]

Or even .. If the cart is an expensive atmel eeprom there what was it, the 28c256 or whatever else with similar control pins .. Hmm, if we expose everything on the bus..


Pc -> Gpu -> z80 -> cart or


Pc -> Z80 -> cart


Ie: reflash ez80 via gpu, or reflash cart via gou or ez80 ..


Hmm

 

[skeezix]

This whole thread is full of my 10s mental barfs .. Hope these make sense between bursts of RL madness :0

 

[Asiyura]

Having the GPU chip able to flash the Z80 using the st-link is IMHO a really good idea.
The less tools you need to get the zikzak to run or to modify its code, the better it is.
Even more if the tool "eliminated" is expensive.

---

I'm looking at Mockingboard's datasheet/schematics since I would like to make a nice PCB able to accommodate at least two 3910. I have some troubles to pick the capacitors I should use to get a clean sound signal, so I'll probably just copy their circuitry for this part

I 'm also considering designing a PCB for one 40 pin ZIF socket, to easily connect it to breadboards with ribbon cables. I've seen how easily I can bend chip pin while trying to remove them from breadboards. -_-

LQFP adapters are still on their way, I hope to get those ASAP since the ST-Link/V2 should arrive soon. I'm eager to get started with ARM MCU even if they seem no as straightforward as PIC, AVR or MSP430, from what you've told.

---

I received 3 SOP20/SSOP20 adapters (with those useful bridge pads), the pcb are pretty thick and seem really sturdy. They came with good quality machined pin headers.

I may buy more of this kind of pin headers, they look far more sturdy than the cheap ones with square pins. And the round machined pins are quite slimmer, better for using with breadboards and sockets, I think.

Other good news : I received the japanese Pop'n Twinbee: Rainbow Bell Adventure cartridge I ordered.
Now, I'm waiting for the super famicom to try it. The seller told me that they were late for the shipping, so they'll upgrade my shipping to airmail instead of economical shipping for free.

 

[skeezix]

Damn the boar.d. wrot eup a enormous reply..

and now board is not accepting

baby crying.. will repost it tomorrow

 

[skeezix]

Having the GPU chip able to flash the Z80 using the st-link is IMHO a really good idea.

The less tools you need to get the zikzak to run or to modify its code, the better it is.

Even more if the tool "eliminated" is expensive.

yeah, theres a few options buzzing around my head. I'd like things to be flexible, but not complicate the design.

So..

1) Programmer header for the STM32F4 GPU

2) Programmer header for the eZ80 (ZDI, Zilog Debug Interface header)

3) Maybe also hook up STM32 to eZ80 ZGI, so that we could code up a proxy or whatever

4) Serial/USB CPU: Given 64KB flash in the eZ80 model I've chosen to test with (may or may not be final brain, but I sure like the specs,

and it ditches the I/O chip need for us, bringing us very close to final design!), it may be possible to flash a bootloader on there, of

sorts; ie: wake up, check for various things, and boot up cartridge if present; if no cart present, show a terminal of some sort? If cart

present, wait a few secs, then boot cart .. if something hits (say) serial in those secs, open terminal. -> Terminal could have an option

to reflash the 64KB flash, or even write to the cart (since some or all carts will have eeproms .. my carts so far can be written to by any

cpu.)

5) Serial/USB GPU: May be able to build the same into this, but I forget offhand how muvch of itsd flash it can write to while live

So in my brain, an STM ST-Link V2 cable would be a good thing to have, for debug and flash the GPU, and through that maybe also the cpu and

maybe cart. But even without that, seems reasonable that it coudl be coded up so the cpu and carts could be flashed with pruely a usb

cable, or say a usb-serial-ftdi cable if I don't put all those FTDI's on board.

Remember with the Zikburner test.. I use an FTDI on the board, and talks seiral over usb, and can program the cart right there live; wrote

a little serial server on the avr controller there, which does hexdumps and verifies and all that good stuff. Same sort of thing running on

the eZ80 would rock

---

I'm looking at Mockingboard's

datasheet/schemati

cs since I would like to make a nice PCB able to accommodate at least two 3910. I have some troubles to pick the capacitors I should

use to get a clean sound signal, so I'll probably just copy their circuitry for this part

Nice Pretty awesome board, but seems way more complicated. In a way its great to be an electronics noob and free of worries about safety

.. so far I've had very good lluck with my very simple boards working well, ignoring all the extra circuitry peopel used to throw in;

really, I think this is a side effect of modern specifications.. the old chips would fry if you looked at them funny, but nowadays, tough

as hell. I dunno.. Mockingboard had a lot of stuff on it last time I looked, but really just needed a couple chips and call it a day

I 'm also considering designing a PCB for one 40 pin ZIF socket, to easily connect it to breadboards with ribbon cables. I've seen how

easily I can bend chip pin while trying to remove them from breadboards. -_-

Geez, you using a spoon to pry them out?! (I use a ball point pen..)

Never had this problem.

Buy yourself a couple sockets then .. I mean, locally they're expensive for me, but I ordered a stupuid pile of them from Futurlec long

ago.. got like 20 various 40pin ones (a few widths), some 32, 28, 16, etc, assorted sizes; got a huge bag of about 100 sockets I've been

known to stick a few mcu's into them, and jst keep wedging those into the BB's, let the sockets bend up and save the mcu. But in practice,

I've not had a problem ..

My pricier 3 BB thing, is very tight; have to use pliers to stick wires in; that worries me a bit for cpu pins, but so far so good.

My big giant 10 BB monster uses those $6 cheapo BB's, and they're pretty loose; no problem, but you do run the risk of wires slipping out,

and on BB with 40 or 100 connections, you don't want to be debugging _that_ :/

When I had zikzak r3 (giant one with double buffering in hardware), a lot of my problems were just line inductance between the 'mcdonalds

arches' wires, and stray capacitance and... lose wires; a few wires were in place, but just not solid connection and you'd see all sorts of

display noise and problems, unless I held the wires just so or squished them, or breathed at them .. nightmare until I found out which were

the culprits.

So now I try to use the tight BB and good solid wire jumpers. (I bought assorted spools of wires, and hung them on a piece of wood.. think

toilet paper roll, with 5 or 6 spools across it. Black, Red, White, Blue, Green, etc..); use good wire. Expensive, but worth it, and re-use

them.

LQFP adapters are still on their way, I hope to get those ASAP since the ST-Link/V2 should arrive soon. I'm eager to get started with ARM

MCU even if they seem no as straightforward as PIC, AVR or MSP430, from what you've told.

You probably sleep more than me, and maybe have more experience? At least you have the benefit of my experience, which should put you into

work no problem.

ie: For me, it was the 'okay, I soldered it, now what? .. too many pieces to look at, a large surface area to attack; when you find 10

manuals, each 1000 pages, its sort of 'w-t-f'! .. but I can point you right at a diagram that shows the _actual_ required minimum pinouts

(I had an ST engineer write it out for me, since the specs are not entirely clear.). Once you get the ST Link pdf, it has the pinouts for

that, and coupled with th eminimum pinout, you can be up and going. Then.. how to write code? I didn't know where to start, from basic

reister dump in datasheet, to the ARM standard CMSIS register naming, to the "Standard Peripheral Library" (thats the main API, oddly named

as it is), to libopencm3 (open source library), ... Let me tell you, ignore the register dump, ignore CMSIS, and go right for Std

Peripheral Library. Its a very thin wrapper over the registers, but comes with a Windows style chm help file that documents how to use the

stuff, and includes examples in the zip file. Anyone in the ST forum asking questions will get answers in StdsPeriph lib format.

libopencm3 looks like its pretty good, but I found it .. a thin layer over the registers, withotu much documentation.. and since it doesn't

relate back to the StdPeriph, you end up going back to ARM CMSIS docs etc, which are a nightmare (too much info!).

So for 'day 1', get the Standard Periph library (I can point it out if you can't find it), and good to go.

Use gcc. Makefiles for my test experiments are in the 'zik80' directory in my github.

What I use is the libopencm3 Makefiles, of sorts, to get good linkerscripts and to get the function names to ISR mappings. ie: This way I

can refer to tim1_isr(), and the lilbopencm3 crap will slot that into the right address to work, but I'm not actually using libopencm3 much

(just to set the clock speed )

So you see, I rammed head first into a lot of complications; libopencm3 works, but confusing

oh, debugging and gdb stubs and such .. get the launchpad arm toolchain, .. that gets you gcc, gdb, etc; use debian/ubuntu, get the openocd

package for the gdb stubs, and to flash stuff; you'll see in my makefiles and sh-scripts, theres a ./build.sh script, and a ./deploy.sh

script (and the gdb-attach and gdb-start type scripts.)

-> save you a lot of time!

I also found otu the F2 was not well supported in libopencm3 and openocd etc .. the 'flash and restart' would crash ,for instance.. so

eventualyl I figured out I could do a full erase (slow), and then flash, and then start, 3 steps of commands; but when I switched to F4, it

all magicaly works very nicely.. and in general, F4 has much more support around the net. So you went F4, you're a foto up where I was.

So, yeah, a lot of things to wrestle.. toolchain, debugging, programmer, API, ARM, etc. But ... get launchpad toolchain, openocd, Standard

Periph lib, and use my makefiles and scripts, and you'll be up and going fast.

I received 3 SOP20/SSOP20 adapters (with those useful bridge pads), the pcb are pretty thick and seem really sturdy. They came with good

quality machined pin headers.

I may buy more of this kind of pin headers, they look far more sturdy than the cheap ones with square pins. And the round machined pins are

quite slimmer, better for using with breadboards and sockets, I think.

I got burned here a bit.. I wasn't sure of the diff, so when I ordered from futurelec, I ordere both machine pin sockets and regular

sockets (so damned cheap, at least at the time.) Around me though, machine pin headers are REALLY expensive, more than the already

expensive normal ones. I found them really hard to solder.. the ones I ahve don't stick to solder at all, which is annoying as shit A

normal chip can't fit into machine pin socket, etc.. just machine pin headers. So you end up with compatibility issues, but if you get all

the matching pins together, they snap together snug and tight, which is nice.

FWIW, a tight breadboard.. no problem. A loosey goosey BB, the machien pins may be too thin and slip out or make poor connection, so test

them out.

ebay again though.. I have a parts box just for header pins .. machine pin or not, right angle or straight, extra long, and extra extra

wire wrap long.. and female straight, female rightr angle; single and double row, etc. Hit up ebay, and stock up on all varieties

I need to keep ordering more.. drives the poor wife nuts, but I mean.. buy a baggy of 20 20-pin headers of each kind for a couple bucks..

spend $20 or so, get a big pile of headers.. you always need them, in every kind.

Other good news : I received the japanese Pop'n Twinbee: Rainbow Bell Adventure cartridge I ordered.

Now, I'm waiting for the super famicom to try it. The seller told me that they were late for the shipping, so they'll upgrade my shipping

to airmail instead of economical shipping for free.

So much fun in one post my man

My one kid was driving me insane on Friday; I ended up putting he rin her room, and jkeeping my back up against the door while she threw

her stuff around and raged, keeping the babies up and making the wife cry.. quite annoying. But during it, I retail therapied myself via my

tablet.. eep So I've got about 6 orders in, mostly some good prices, vbut took a bath on some stuff. Ordered 4 eZ80's ($5/ea), a bunch

of those 128KB RAMs I found ($2/ea, love them to bits from the datasheet, will use them in a lot of things), a couple 28C256Kb eeproms (I'm

running out, and making some carts for experimentation), some 2x16P .156" edge connectors (arcade style, so I can use old fingerboards on

them for testing), and some other junk; mostly ebay, but some digikey (ouch, for those ram and eZ80s).

Hard to find edge connectors, but I've been needing to order some; I've got a small stack of various non-36pin ones around from my arcade

repair stuff, mostly 44p and JAMMA 56p .. but I want 36pin specifically, since they fit into Atari 2600 carts still, and have the right

numebr of pins. I need to get explicit measurements (my dad gave me a nice digital caliper) so I can make Eagle parts for the board layout.

Since I'm going to socket a lot of stuff, and because all CADs suck, I'm basicly going to have to make all new parts. ITs quite..

despairing to think, great.I can sit down and redo the schematic and get it near final.. and then think 'oh, instead of that 1-2 night

task, I have 10 nights of finding parts, redoing them, tossing htem out, redoign them again..'; you spend 90% of your time screwing around,

and you have to do it up front before you get to real work :/

If we go with eZ80, then it may be best to just put the DIP drill holes on the board, and then get those $1.50 futurlec adaptors; I put an

order in for some of those ut who knows when they will come, or if they'll all be the same from one order to the next; but as long as I get

1, I can clone them if I have to since its just a straight up adaptor. Anyway, once I get _that_, I can measure it and clone it into

schematic .. all the pins need to be entere dinto CAD etc, exactly right.

So I can come up with a near-final schematic, but I'll have to redo it for board design anyway once I get those adaptors in etc..

.. and baby is up, another night down the tube

jeff

 

[Asiyura]

Geez, you using a spoon to pry them out?! (I use a ball point pen..)

I used to use my fingers and my nails, now I use a knife or a flat screwdriver. My current breadboard's connectors are tight, you really have to use force to plug or remove the chips.

FWIW, a tight breadboard.. no problem. A loosey goosey BB, the machien pins may be too thin and slip out or make poor connection, so test them out.

As said, my breadboard have really tight connectors. I can't plug adapters with square pin headers in it without using lot of force (I have to use the other breadboard). And removing them is a real pain, you have high risks to break or damage the adapter. That's why I'm interested in machined pins even if they're more expensive.I also have a "mini breadboard" glued to my arduino "screw shield", it has round holes instead of square ones, it can't accommodate square pin headers at all.

ebay again though.. I have a parts box just for header pins .. machine pin or not, right angle or straight, extra long, and extra extra wire wrap long.. and female straight, female rightr angle; single and double row, etc. Hit up ebay, and stock up on all varieties

That's where I got mine too, but I only needed straight ones for my small needs. Bought a bag of 50 pieces of male 40 pin headers for a few bucks. That way, I don't care if I screw up some 

Since I'm going to socket a lot of stuff, and because all CADs suck, I'm basicly going to have to make all new parts. ITs quite.. despairing to think, great.I can sit down and redo the schematic and get it near final.. and then think 'oh, instead of that 1-2 night task, I have 10 nights of finding parts, redoing them, tossing htem out, redoign them again..'; you spend 90% of your time screwing around, and you have to do it up front before you get to real work :/

That's a real pain. Lot of time lost searching for a library that contains you part. To find none or to find one but the part is not done right… then you have to spend some more time creating/modifying the part in your cad software.  -_- Even for the only PCB I designed, I had to modify some parts. IIRC it's not the most straightforward process (at least in eagle).

EDIT:

Just got a new delivery from maxim. Totally forgot that I ordered thermometer & real time clock chips (all-in-one) a few weeks ago with the other chips samples. Seems those were out of stock, so maxim delivered them as soon as they got them back.

I plan using it to display date/time mainly to be able to log some info with timestamp. But also to have date & time on a led display in the elevator and control a fan in the PCB enclosure to prevent overheating. (don't really know if my circuitry will generate lot of heat, so better safe than sorry)

 

[skeezix]

I've not ordered any samples from Maxim yet .. need to get on that I use way too much of their stuff. Gotta have 1-2 serial on every board I make

Over the weekend, picked up a QFP100 DIP adaptor (50*2) and alsoa aluminum stencil; should make a ez80 breadboard test pretty straighty forward, and will be good to see how the hot-air works on such tiny pins. Microscope + hot-air ...

Today is a Good Day (ignoring the credit card charges, that'll be a bad day) .. couple of packages sitting at FedEx, hopefully coming out during today while it snows Should be getting..

- ez80

- the $2 RAMs

- couple more eeproms for carts

- zilog usb smartcable for last resort (needs Windows; I can plug in my old laptop if I must, but prefer not.) Going to see if I can talk Zilog-Debug protocol from STM32 (or maybe avr first, since I've already got avr<->pc serial server/client going there for zikburner .. bite off one thing at a time.) So maybe avr -> zilog programmer, then move it over to stm32->zilog, and that'll prove the program-ez80-via technique.

The ez80 should (I think) be able to run without any additional stuff, like avr's etc (built in 4K RAM and some flash, built in crystal); so I can try to program it, and try to blink a gpio, to prove it all works. Get it powered up etc, prove the toolchain, all that. Man, I have way too many toolchains installed

Should be a fun week or two working on this angle; once can flash the ez80, next test will be stm32+ez80 using bus shared memory; should be fine, can't see why not.

Then pcb board design.

really, buddies of mine tell me I should've gone right to pcb day 1, and then you screw it up, and do another iteration or two, and you're done; but not the point... I've had a lot of fun over the year proving piece by piece and refining the bits, narrowing in on what I want to do, and making sure it all works. VGA being a right pita at the low end hardware of course, taking way too much time .. but implementing VGA in 6 different chipssets ftw

So who knows.. inching towards pcb design time; once those futurelec adapters show up, I think that'll be time. I've got 2 QFP->thruhole adaptors here now, but everybodies are different; I got I think 6 or 8 futurlec ones on the way (stupid me, it was late, I should've ordered 20 of them or something, since pretty cheap and save on shipping, etc.) If I use same ones for both chips, thats 2 per board... but makes it easier to design anyway, copy paste that part over.

Who knows how long futurlec will take (if it comes at all.. they didn't ask me about CC at all, and its been awhile since I ordered.. do they keep CC on file? and right CC? Their website is 3/4 broken so you can't check your account status or anything, so.. who knows?)

I could just order a stack of those little adaptors locally, but they're what, $4.50 or so each, sort of hurts. But could get them overnight So will wait 6 or 8 weeks I guess for futurlec, and if nothing shows up, order these local ones ..

jeff

 

[panik]

Who knows how long futurlec will take (if it comes at all.. they didn't ask me about CC at all, and its been awhile since I ordered.. do they keep CC on file? and right CC? Their website is 3/4 broken so you can't check your account status or anything, so.. who knows?)

Yes, your creditcard number is in the account settings (They tried to charge a 6 years old creditcard number on my last order that had been stolen in the mean time). It indeed felt weird: ordering without being asked to pay.
This link works for me: http://www.futurlec.com/YourAccount.html (redirects to somewhere, can't find the link on the website itself)

 

[skeezix]

I have to say ..

Doing these QFP100 guys, with the right gear is not too bad; its time consuming for sure.. like gtting it tapped into the right spot at all, takes more patience than I have But its doable, and not hard.. just tedious.

I used aluminum stencil and squeezed solder paste down, and squeegee'd it through stencil; getting stencil right was hard, and keeping it right; paste was more or less correct.. but I don't think I'd do it again :/ the paste ends up obscuring everything, so its hard to tell if your chip is in the right place or not; a little cap may self-adjust, but 100 pins are not going to slide into the right place magically. On the up side, the paste acts like resistance.. like glue, so it doesn't slide around so much. Some sort of non-conductive wax might be nice for a similar effect. It might be slicker to tack it this way.. paste into the corners, heat gun it, and if looks good, then paste around the top edge and heat gun it then.

Anyway,..

- andonstar usb microscope - invaluable! After doing the hot air rework, inspect and see what bridged; raise it up a touch, and can then solder and see the magnification live, to fix stuff

- hot air rework - definately pretty nifty; paste down, wave the wand around, and poof.. paste melts and like magic all the pins are soldered

- solder wick .. works nicely on SMT still, when a tiny bit too much solder

- flux pen - using it over and over to help guide the way; need to look up how to get the excess flux off.. its shiny, which is annoying under the microscope

Could likely do it just fine with the soldering iron .. I've done it before for the QFP64 and lower, certainly; drag solder or variations.

The important thing is.. the microscope, so you can clean stuff up.

Having the right tools adds confidence.. when you take a stab at something and its hard, you'll goof up; but here, with the hot air rework, I know I'll get it; I won't get it too far wrong at all (I can post pics, but those pins are pretty darn square on those pads, tiny as they are!), but even if I blow it.. or bridges under the chip where you can't see .. I can fix that; hot air, pop the chip off, restart, in worst case. But it doesn't look likely you'd get to worst case.

So, yeah, microscope is essential (and cheap if you're lucky!), and the hot-air rework is great to recover a bad spot .. or like I just did, just do the soldering with.

You can get them SparkFun, a fancy one, for $99USD (thats got some features too); thats driving thre cheapo ones down in price on ebay.. I forget now how uch I paid, but it was $70 or something? Anyway, totally worth it. Good for heat shrink and all sorts of little issues, and with these SMT parts, essential.

.. now, it does take an hour or two to get it down, inspect, test for brudged pins, repair, etc.

I'm using a DIP prototype board here, 2 rows of 50 pins on it, spaced for breadboard; its easy to check for bridgesd, before making it live.. just test the DIP pins; checking for connection from chip to pin is harder, but I rather assume..

i) if I'm touching one DMM lead to the right DIP pin, and drag in the right area on the chip (you can be damned close, even with QFP100), and it beeps.. you know its connected to something

ii) if you check continuitty beeper for adjacent pins, and no beeps, you know no bridging

iii) with microscope, you know the pins are in the right place

So.. you can check for connected to something, and prove no bridge, and know its in right place.. so should be able to 100% test the adapter, before putting into circuit.

With real ZikZak, I'm leaning to those futurlec adaptors if any good, or the lcoal shops (very good, very reliable quality products.. but $4.50 per, in this case..); this way I can sit dwn for a day and knock out all the chpis I need for the kitds for you guys, if you want me to solder them ofg royu, and I can ship off the pre-soldered pre-tested adapter; plug it into my zikzak, prove its good, take a photo, send it off to you; if its bad at your end, no my problem then right?

I suspect I'm way too slow at these operations; buddies of mine slam through, 10 mins theyu're done, but I'm careful. Too careful maybe. But its fun

jeff

 

[slaeshjag]

If you're going to need a lot of adapters, you might want to order your own custom PCB's for that. 10 5x5 cm boards is $10 + shipping at elecrow.com (which is what I use for prototypes.) And for a small fee, you can panelize out multiple of the same type of adapters onto the same board and have it V-grooved.

 

[skeezix]

they do the V-groove too eh? nice I was looking at seeed studio..

yeah, was thinking of knocking off my own, too; not going to fret about it yet.. but thanks for the website tip.. looks like a good site, I'll have to review in detail.

On the one hand, zikzak is probably mostly for me and a few of you guys, but there is the rare chance maybe it'll end up goign to 20-30 or more people. So on the first case, could just put the SMT pads on the main board, and I take care of them if you don't want to. No biggy. On the other, if expecting to sell a few, then make them adaptor modules, for all the reasons above.

Or maybe both .. first time, do the pads right on the board? *shrug* One advantage of doing adaptors, is your headers are well defined and right there, every pin; without them, I'll have to add headers for all the pins anyway, since the board won't use half of them and you may want to add SDIO or an LCD panel or whatever..

jeff

 

[skeezix]

Sweet. FWIW, got this sucker first time, QFP100 onto adapter. I'm really pleased .. when I got the LQFP64 and did them by hand with just the iron I think it took 3 shots, with the first 2 'ever so close' but a couple pins just grazing the edge of another pad, and no easy way to fix them (need to remove half or whole chip in that case.) But with microscope, even that could be avoided; but with rework, you know you can fix anything.

This is awesome

 

[Asiyura]

AMS packaging is impressive.

I sampled their magnet joystick and the 2D hall effect sensor chip.

Here is the packaging of the joysticks:



Yep, cardboard box with foam and plastic box with foam!

All the chips come in cardboard box with foam and the classic antistatic bag.

Now I need a PCB for the chip and its joystick. Of course I can't find an eagle library containing them that could have sped up the process -_-

And I can't be assed to create them right now, so it'll have to wait.

---

That hot air station does a good job, then.

That'll be my next big buy. I just wonder where I'm going to store it without girlfriend being mad.

I already can't get back my old oscilloscope because it's far too large…

If you're going to need a lot of adapters, you might want to order your own custom PCB's for that. 10 5x5 cm boards is $10 + shipping at elecrow.com (which is what I use for prototypes.) And for a small fee, you can panelize out multiple of the same type of adapters onto the same board and have it V-grooved.

There is also oshpark. Two layer board is 5$ per square inch and you get 3 copies of the board, free worldwide shipping. On the plus side, the boards have that nice purple color and ENIG finish

 

[skeezix]

AMS magnet joystick?! url?

..

Bit of a setback here ..

I thought I'd just hook up the eZ80 Vdd/Vss and see if the USB smart cable would talk to it; I didn't put any crystal in .. wasn't entirely sure if it could run without one or not, and thought no harm no foul, right? Anyway, hooked up the USB cable .. pretty straight forward pinout (nearly same as avr in fact), and powered up; I think the draw went to 10A quick and the USB Smart CAble (not so smart?) went hot, LED went out, and thats that. I think the smart cable is toast.. not sure if defective or if I cocked something up :/

Will have to keep at it, see if I can talk to this chip another way, to try and guess its alive or if the cable took it with it. (I hate working with unknowns like that.. is your code broken, or is the chip just buggered?)

The chip does have a JTAG interface, so maybe there is generic jtag software that can query it; the ZDI interface (zilog debug interface, the one the USB smart cable uses) is there, and isnm't too complicated a protocol (the datasheet documents it.)

I should be able to hook up an avr or stm to the ZDI and talk to it that way, as mentioned above.. wanted to investigate that anyway, so as to build flashing abilities right into the zikzak firmware.. no USB smart cable needed.

So I guess it just means I need to tackle it that route.. work on the ZDI interface day-zero (ugh) .. I'd hoped to at least use the cable to prove if the chip was alive or not

I forget.. will check the datasheet but maybe the external address lines are still toggling while its defaulting to internal flash? doubt it, but maybe .. be nice if I could just watch the addresses climb as it NOPs away or something.

Ah well, .. *rolls up sleeves*

jeff

 

[Asiyura]

AMS magnet joystick?! url?

http://www.ams.com/eng/Products/Position-Sensors/EasyPoint-Joystick-Position-SensorThere you can find the hall effect chip AS5013 and the joystick module N40P105 (with "pins" for better mechanical connection to the PCB, they call that combination "EasyPoint."

The joystick feels a bit like the psp or pandora nub, but with a pushbutton. I played a bit with one of them, they feel really good. They re-center fast, move around smoothly, the pushbutton does a nice sound when pushed, the amplitude of movement is more or less the same as pandora's nubs (2mm for the easypoint).

Since they only have the pushbutton as electric contact in the mechanical part, it's reeeeealy slim: only 2.5mm once mounted on a PCB.

The ship is also really slim since it's a QFN package : 0,55mm.

And they recommend, for best performances, a PCB that's between 0,3 and 1mm thick.

So all in all, the complete joystick is about 4mm thick, pretty nice IMHO.

EDIT: I'm currently looking around to manufacture prototype on thin PCB boards (between 0,5 and 1mm). The one slaeshjag mentioned, elecrow, does. My main problem with them, is that it would be a bit pricey for only 3 joystick boards:

You have to order at least five 5*10cm-sized boards to be able to have them v-grooved and it also adds a lot to the global price. I think a better option would be to fit two or more circuits on one board and cut them myself with a dremel or another tool once I receive them.

 

[skeezix]

More setback.

Dang, I was pretty stoked over these eZ80's over the last week.. datasheet shows all sorts of awesome stuff.

I've soldered up 3 so far to adapters; first one was being super anal, 2 hours; next one half an hour, third one about 10 minutes. (the second two were drag solder, and light on solder.) I've gotten really good at drag technique .. tack down the corners, and very little solder.. drag, and wick to pick up extra after.

All 3 seem borked .. theres no errata on the power supply pins (Vdd, Vss). I've checked the breadboard approximately 100 times .. 8 pairs of power pins, plus crystal.

Adaptor continuity test seems okay .. no bridges; then drop it onto breadboard, add power, turn power off.. adaptor now shoes all Vdd/Vss as shorted.

Its as if you turn it on, and poof, all things melted inside or something Never seen such a thing. The pins test clear, then test shorted after powering up; never seen it.. you either shorted up front, or not. How does it develop new shorts!?

Bad adaptor boards? bad eZ80 batch? I have to assume its my work somehow, but.. 3 times, all identical problem?

Maybe this is a hint to abandon eZ80 and go back to regular DIP Z80 and DIP avr8 for I/O; that makes it a little goofier, but at least they'd be thru-hole components so infinitely easier to work with. Hmm. Suddenly makes programming a pita for folks.

I was really loving the idea of having the stm32 program the eZ80, and eZ80 doing all the I/O stuff. Just becomes such a nice clean design.. eZ80 + STM32, and voila. (plus ram, cart, audio, etc, of course.)

Frak, I'm so annoyed right now :/

I don't even know whats going wrong, thats the bummer; I could order a few more eZ80's ..$4.50ea is not _too bad_ (killed 3 so far..); the DIP adaptor boards are $20/ea, and the square ones are cheaper ($4.50 for the ones I can get quick, while waiting for the cheapies from ebay and futurlec.)

But without knowing whats going wrong... why bother? If the Bb looks okay, just the 16 power pins populated and the chips just die .. wtf? what do you do :/

Bench power supply is not spiking on turn on or anything weird like that; avr8 on same board runs fine.

*mulls over*

 

[skeezix]

I think I'd like to do eZ80, ideally, so will keep it in back of my mind, and talk to a few people..

Not overly interested right this moment in soldering to a square adaptor and then in turn to a perfboard and then into breadboard, when I have no idea what went wrong with prior 3 (as above.)

..

So out of depression on it all, I think I may fiddle with Z180 (got 2 on tap still, DIP64) and avr8 talking over serial; if I can sort that out, then at least we have the backup plan of..

- Z180 as brain

- Z180 using up to 1MB of ROM/RAM via built in MMU

- Z180 sharing bus with STM32 GPU as was long planned

- Z180 talking to avr as I/O chip over serial as long planned

- atmega644 doing keyboard, joystick, etc, but not on the main bus

Sort of loses a few options, but is basicly what we were long thinking about, so not too big a deal. Complicates things for programming, .. but outside of AVR, could still probably work out a way to program it all over stlink. (ie: stm32 coudl write to memory what needs writing to card, and then have the Z180 program the cart.)

Hell, for that matter, we put the cart onto stm32, and could flash it direct; turn off the VGA, flash things, turn VGA on with status, or something.

So could still be slick.

Anyway, will mull it over more..

edit: Added some Apache Rewrite magic, so www.zikzak.ca goes right into the wiki without having to do a redirect; so no more /dokuwiki/ in the path, its all done behind the scenes; ie: any bookmarks you made (I know you didn't), will be broken. Take that.

Upside: Maybe google will not hate the site now

 

[Asiyura]

Too bad about the borked eZ80, I hope you'll get some of those working.

I can understand your disappointment, it's really annoying when you've done everything right and it still doesn't work.

-----

I got my ST-link/V2, 40 pins ZIF sockets and my super cheap Super Famicom today, girlfriend was pouting in a corner until we got our new oven

So, I'm still missing the crystal oscillators set and LQFP adapters to get started with STM ARM chips.

 

[skeezix]

Maybe I'll go back to zikzak r2 idea of pluggables .. Use ribbon cables and IDC connectors and make cpu module separate; use stm32 or ez80 or z180 or avr or whatever as brain, right? Like a cape/shield..


Like really, should use stm32 as brain and another as gpu for more power, but z80 is retro fun.


My hoal was retro and not awesome console, but easy mods .. But then dual lqfp64 stm32 f4 woukd rock too; shrug

 

[skeezix]

A quick review of the older zikzak modular design and thining a bit.. the reason I aborted was too many ribbon/connectors and complexity. Address bus (24pin), databus (8pin) and control pins (half dozen.) Connectors in towards cpu .. 2 joystick (say 10-16 pins), keyboard, maybe some misc (SDIO..), and then outward (audio); power bus. Some others. End up with like 5 sets of cables or connectors, and a hundred or two hundred pins. Crazy.

So maybe best to go with..

- simple default cpu (for now) .. known to work, easy to work with

-> leave drill holes to drop in header pins .. so can

i) skip putting in a cpu and optional I/O chip on board

ii) put cape or jumpers on header and use another board or perfboard to put another cpu and/or I/O on.

ie: So instead of being modular and making it complex, just leave the option to put a header on and go nuts that way; so it defaults to one cpu, and lets you fork off another.

Will have to schem it all up, need to redo it all anyway, and then see how much board will cost, and decide from there.

..

Still need to make up a cart prototype board, just to get into the whole pcb design and getting made feel.. get a board house picked.

Then see about making maybe a zikbutrner board (for fun, may not be needed anymore, depending), and then the first cut of easy zikzak board.

I'm sort of itchign to get to making a damn board, been too long; even if its all wrong and screwed up, it'd be progress

...

Then after getting a working base board, with an avr or stm32 as cpu, then figure out next step; ie: just go with it, does headers work etc, or redeisng/tweak a bit.

..

hmm

Oh, and, Thief 4 comes out Tuesday

jeff