completly dead falcon

[mehlab]

Dear dml

thank you very much for your quick respons time and of course for the answer

will keep you inform all of the result sadly not before september thanks to acp i still have a firebee to work on it

[dml]

quick update on 2 dead Falcons - one has sparks of life.

Board was in a sorry state - looked like it had lots of repairs (videl had been replaced - can see slight pin misalignment and some flux around the chip), one memory expansion pin replaced - but length not trimmed, too long to fit a memory board on it (thats really strange!) some tracks cut and remade, L102 hanging off at one end, some other minor issues. I doubt it had been used since the last changes were made....

So I reattached L102 and tested, nothing. Noticed a lifted track near videl, held it down with a matchstick and retested, Atari logo and memory test ok. Not done yet but that's a good sign

If I don't find anything else needing fixed, I'll test my own flashed GAL & ROM chips in this board, and start populating the other one...

[simbo2]

make sure at least the metal strip is across all the sockets is in place with its threaded port screws are there
if you dont ground all the sockets together the falcon wont boot
i would happily lend a test cart and rs232 cable if you pay postage

you can get any one of the atari st etc test cart boards
just replace the eproms for say a falcon or tt etc if you cant get the models cart
the rom images are in the forum and i can even send a set if needed
http://sales1632.myzen.co.uk/acatalog/


its far better to see whats going on with the cart and rs232 to a pc and hyper terminal running

[candle]

thanks for the tip

i've measured continuity and resistance with my DM3054 bench meter and resistance with or without metal stripe is exacly 0.000 ohms

i guess i'll need to resort to diagnostic cartridge

tell me, this is just plain st-style cartridge with two eproms on it? (just in case i had to make it)

[dml]

quick update on my 2nd Falcon in repair - I made a seriously bad mistake (which simbo has pointed out to me when we talked last) in removing the combel i had probably *successfully* replaced, just because i saw strange signal levels on the expansion port. I forgot that only some devices are mapped to some of the bus and data lines so they are not all equal. Now I have to solder it back again.

What I saw (apart from black screen - which is probably something else) was patterns of pulses, mostly 1ms period, (750us high or 75% duty) on most of the data lines, and some different patterns on others. No signals on a few, and some with what looked like superimposed signals (strong 4v swing with another 1v waveform overlaid, either overlaid on low or high signal depending on the line in question). This freaked me out and i removed the chip thinking I had introduced a fault between pads under the chip.

It seem the signals in this machine are not very clean and some of the lines have different voltage levels and impedence and this is normal. :-z

Anyway, simbo gave me plenty of advice (thanks!) and once I've restored the combel (again) and see some activity on the bus I'll stop desoldering stuff and debug the black screen probably with the diagnostic cart...

The other thing I did was fully map out the combel and it's pins (that is an evil device to solder!) and made corrections based on the schematic and service guide (all of which have mistakes). I'll post the corrected mapping diagram when I boot my other machine up.

[simbo2]

viewtopic.php?f=27&t=16237&hilit=FALCON&start=25
background material read the whole topic may help someone
you need to have the metal base under the pcb and the metal strip along the back in place
if you get no boot at all
then you need to check it with a cart and rs232 before doing any replacing or repair
you need to check the 'reset' line is high and the HALT is high {power on reset chipset init}
does the reset pulse when you press reset
try pressing in reset and power on let reset go if this fixes it change C1 C3 caps

what ic's are warm
does the mpu get warm after 30 seconds
are all the clocks running

visual inspection of the pcb {upload a hi res pic or two of the pcb}

[simbo2]

thanks for the tip

i've measured continuity and resistance with my DM3054 bench meter and resistance with or without metal stripe is exacly 0.000 ohms

i guess i'll need to resort to diagnostic cartridge

tell me, this is just plain st-style cartridge with two eproms on it? (just in case i had to make it)

do not forget to check the reset switch isnt jammed up {copper jam the contacts turn into } may need to replace the switch


yes its any cart with 256k eprom pair
can be set to rom3 rom4 etc with solder pads
any atari cart will work you just change the roms

attached is a design someone can make up if they like
its untested i just have not had the time to do more work on it
pm me if you want to borrow the cart and cable {the whole test kit from best is about 150bucks +post+tax etc}
the cart and rs232 cable is enough to see whats going on on hyper terminal {i use a rs232 to usb converter board ftdi chip worked fine

[candle]

i've bought one for ste from the site you've linked previously, as i don't think i can do it cheaper than for 20 queens anyways, so now i'm waiting
reset switch is allright as far as i remember, but i'll check it again

[simbo2]

i've bought one for ste from the site you've linked previously, as i don't think i can do it cheaper than for 20 queens anyways, so now i'm waiting
reset switch is allright as far as i remember, but i'll check it again

in the archive above is the ic, images for the falcon diag cart

in another topic i posted here is the a few pics
viewtopic.php?f=27&t=19389&p=168353&hilit=diag#p168353

and here is the TT version

viewtopic.php?f=27&t=19784&p=172437&hilit=diag#p172437

all the carts pcb are the same just the rom3/4 signals are changed by the size of the eproms you use and the images in use

follow the details

i can take some pics of my falcon one if you like it has no extra TTL ic like the others
becouse it uses the right eproms ive posted the images if you need a set
i am almost ready to burn eproms again my machine is fixed and im building the o/s just now tomorrow ill burn off some sets i need to

[simbo2]

ok so you checked reset
check it at the expanssion port xreset pin 17 xhalt pin 18 on the 50 pin connector
it will on power on after a few hunders ms go high it will remain high
as will halt
if halt goes low at all then there is a chipset issue
you need to check out the mpu and combel as if the halt goes low processing has stopped
read the service guide on devdocs and use the cart is best at first
then you can test just three signals and it will tell you what ic is wrong
if you have out of range on the lcd then the videl is most often working
and the dma is faulty
take some pics of the pcb too see

[dml]

ok so you checked reset
check it at the expanssion port xreset pin 17 xhalt pin 18 on the 50 pin connector
it will on power on after a few hunders ms go high it will remain high
as will halt if halt goes low at all then there is a chipset issue

Originally (before I pulled the combel off first time round!) it was passing no clocks to the other chips and everything seemed to be in halted/reset state. After replacing the combel, the clocks were all alive, the reset/halt states looked healthy and the bus woke up and started doing something. However the display was still black and out-of-range. But then I did the bad thing and took the chip off again after seeing the funny overlapping signals.

you need to check out the mpu and combel as if the halt goes low processing has stopped
read the service guide on devdocs and use the cart is best at first then you can test just three signals and it will tell you what ic is wronge

Ok makes sense. I think the mpu is probably ok. It got slightly warm but not hot. I mapped voltage drops across the board - all chips - and everything was in the 0-2mv range. Only thing getting really warm was the video dac. Nothing going in or out of that chip. No activity out of the videl at all.

if you have out of range on the lcd then the videl is most often working and the dma is faulty take some pics of the pcb too see

You could be right about that - BTW what happens if you try to test the machine with no SDMA chip? Or with the bus grant control lines tied away? I expect it won't boot because the disk and audio stuff will fail to respond - it'll probably lock up while initializing the hw in ROM. But will the CPU and display function enough to validate those parts?

[simbo2]

you need the sdma in place to boot it has to handshake with the mpu for all memory transfers too and from the combel {ram controller} including the tos load up

in the explination of functions {perhaps you dont have this ill send it over on email or something is a description of the boot up
the mpu grants the bus when a chip is requesting it via int stack priority system
the BGE bus grant enable and BG bus grant BGR bus grant request
on each of the main ic's is released to the ic only when the mpu thinks its a good idea by priority
yes you have the hardware guide in the schematics i sent

anyway the long and the short of it is you prob need to resolder the new combel and i would change the dma
its far easier to solder and take off {bent pin etc to lift the legs } i use smd hot air its easier and very fast
although i wish i had an IR unit also

i have a spare dma chip its been used but i know its ok just to redirect the pins on placement with a pin or something needle wise
start with the ground pins on these ic;s and wear an anti static strap they are VERY sensitive to static

one thing that will stop it cold is a bad MFP

YOU NEED TO USE A DIAG CART !!! otherwise your working totally blind and if you do this you may as well replace the sdma videl and combel
then cry when the diag says its the mfp

[simbo2]

remember i said above about the metal lower case and metal strip{the one like a razor !!} needs to be on and the port screws need to to there also

i forget to add that if you plug the cart in in this setup
you get the red white hash on the video out
but if you have no cart connected then the machine just comes up 'out of range'
if you put the case lower on it boots

so defo one to watch out for

cold testing the address and data bus

you have no power connected
you just change a chip or have not fully inspected the pins of the ic's for any shorts etc
then you need to map out the resistance to ground of each of the address and data pins before applying power at all!!!
now not all chips in the chipset use all the address and data lines
so if you place the black probe on an lcd DVM and change it to diode range!!!!or 20K range
you can go along the pins on the expansion port and write down the resistance of each of the address and data pins

now youll notice that address 1-16 and 8-23 will have a value depending on the group
{22 address pins}
{16 data pins} 1-4 ,1-8, 9-16
and not short etc do the same with the data pins make a note
{the expansion port is best here to measure from }
you can do the same with the ram but if you have a scope you can look at each of the address and data
you can see that pressing reset the board will 'load' something to ram if the mfp is working they stop if the sdma isnt
and hold state and pulse rate {refresh } als check the ras cas clock strobes are running

also if a data or address line is state stuck
ie short to ground or has a strange low reading or higher reading than members of the group your checking
youll know there is a soldering issue or a small shard of metal or solder on a pin or two forming a short accross the address lines

this is by far the best thing to do !!!!BEFORE !!! you apply power
so all cold tests possible pin to pin and pin to ground can be done safely
without frying a chip you just payed out bucks for

[dml]

@simbo - Ok understood. I did a lot of this the last time round but not so organised / planned out. I don't really trust what my multimeter is doing in various modes so I was tempted to make a sensitive continuity tester WELL BELOW the diode drop range for pure continuity/short integrity testing (like 0.25v / limited current, with a builtin load to reduce capacitance).

I'm less sure about what to do for checking resistance because I could maybe kill some gates using anything above 0.65v, maybe even with stray capacitance in the tester. Any advice on this? Just go get a better meter made for that job? I still can't fully understand what is/isn't safe to do for cold-testing of boards with ICs.

(BTW thanks for the repair goodies! got them today, will make good use of them

[simbo2]

this is why i thought of using a riser board with buffers and leds on the expansion port for hot tests

for cold i just use a two bit lcd meter set to diode range or 20k range for cold tests or hot i use a logic probe

there is nothing special
you see the gates on ic's will just show the ecd circuit not the actual gate as a diode as it is...
most ic's need to initialise there port state most actual gates on the chips are open collector

so test away map out the results for each address and each data
you can soon see where a pin is disconnected etc...

this test cold test only shows the actual physical connections
when you place a new ic

to be sure it doesnt have stuck pins or shorts / opens

[simbo2]

here is the driver i though of for a bicolour led
for hot testing makes a pretty display for the front panel kind of space 1999 {im not serious!!!}


so address 1-22 + data 1-16 = 38 X2 = 76 gates and 38 leds 76/6 per ic
is around 13 ics for the address and data each led will require something like a 470R or 1K resistor in series with it and the not gate{inverter}

better with a logic probe

[dml]

here is the driver i though of for a bicolour led
for hot testing makes a pretty display for the front panel kind of space 1999 {im not serious!!!}

Nice. If I can dig up my edge connectors I could probably wire it up. Thats a *lot* of buffers though It's almost worth using a Xilinx IC!

(I've been busy working on PhotoChrome5 and posting updates in another thread. That's working now so I can slow it down a bit and go back to my repairs)

[candle]

got my diag cart (for ste) from 16/32 systems, and it has socketed chips, but doesn't have any additional logic
do i need one, or it was just for some other reasons? (ie not using righ sized rom chips)

also, can i use amiga to pc null modem cable for console output, or should i build another one?

[simbo2]

hi darren
can you take the one screw out the falcon diag cart
and take a picture in macro mode of both sides
post two jpg sized etc

as if i remember that cart has no extra address decoding logic
and is just a standard one

ill repost the eprom images as i cant find them on the board
they are here somewhere

[simbo2]

here is the driver i though of for a bicolour led
for hot testing makes a pretty display for the front panel kind of space 1999 {im not serious!!!}

Nice. If I can dig up my edge connectors I could probably wire it up. Thats a *lot* of buffers though It's almost worth using a Xilinx IC!

(I've been busy working on PhotoChrome5 and posting updates in another thread. That's working now so I can slow it down a bit and go back to my repairs)

as you say so the other way is to use a uc etc and then output serial from it .. to an lcd or something

[dml]

so address 1-22 + data 1-16 = 38 X2 = 76 gates and 38 leds 76/6 per ic
is around 13 ics for the address and data each led will require something like a 470R or 1K resistor in series with it and the not gate{inverter}

So if I to use a 74240/octal buffer each chip divided into gate pairs could drive 4 LEDs, so that's 9 ICs plus a lot of LEDs and resistors. The pin wiring for an octal buffer is also a more straightforward pattern, with a simple U-turn per LED. Might make layout/assembly easier.

36 + LEDs + resistors though... that's a lot of effort for a stripboard kiddie like me. This one is best done with layout and etchant. It looks like fun to try but only if the wiring is all done on the pcb and not with wire loops!

(here is a controller board I built with TTL years ago, using stripboard and wire loops - painful, painful!

[mehlab]

Hello all

In july 2012 i post a text requesting help for a dead falcon

After having found an old body who kept the falcon for almost 10 month without having time to do anything (i told him stop programing DNA stay with 68k cpu (for the slow one of you he got 4 kids))

I went to a high end sound shop in my town explain them the situation and they told me to come back when the electronic guy is here

Came back with the falcon the guy was really happy to see a falcon he even ask me if he was allowed to take picture of it (he had about 10 years old when the falcon was out)

He fix it with the help of all your comment and now it is the second time that i boot my falcon today

PS For the tech among you he check all the cold soldering and replace the RTC and of course he put it on a socket

Thanks to all of you all around the world for your time and patience

[Dal]

Nice story - good to hear another dead Falcon spring back into life.

[Shredder11]

That's a heart warming story and great to hear you are now enjoying your Falcon!