SCSI corruption & SCSI patch

[wongck]

I located a SCSI-clock patch on one of my Falcon and this particular falcon experiences data corruption on the external scsi devices.
On my other Falcon, no such SCSI-clock patch and it work flawlessly with the external devices.

Can someone let me know if this scsi-patch looks correct?
It seems to be too simple and missing some wires, when i compare to some docs found on the web.

(click to view larger version)

Should I remove it?

[Shredder11]

The photo link you posted gave me this...

Ooops...
This page (hosted by 110mb.com) is missing.

[wongck]

The photo link you posted gave me this...

Ooops...
This page (hosted by 110mb.com) is missing.

Yikes... was working last night !??!
May be bandwidth limited, so I moved it to imageShak.

[FatRakoon]

I had a lot of issues with my SCSI when I got my CT63

I originally had the 7408 IC but I changed it to the 7404 Patch and now, what I am using is no SCSI Patch at all, but I am using the resistor trick
( Dont know if its helping but since its working flawlessly, I dont intend to remove it to find out )

So, yes, I would say remove it.

[simbo]

yes remove it totaly when using a ct63 and restore the three 33R resistors {current limiters}

rodi recommends you change one of the bypass capacitors near to the dma chip instead
the mod is on the ct63 site...

for pal machines only...
i also change the three transistors in the dot clock section {beside the nvram to the right}
use bc547 type {watch the pinout its the reverse of the ones there}
the 2n type used is a general purpose but has not much gain
so you get a week clock carrier and intermix
however bc547 is a higher hfe factor so gives a cleaner and more amplitude
of clock carrier and intermix for the colour clock

for ntsc machines i would do nothing except change the bypass capacitor
as it doesnt have the same clock setup



you can also if TV output is something you dont want in a pal machine
remove inductor L102
on the left side of it run a coax from it to the dsp 32mhz clock crystal
i do this with the nemesis so i just hated the idea of modulationg a clock
that should be crisp and solid
doing this one step will cure the dma issues completely
as it fixes all clocks to the dsp clock
there is a small pad next to it as there is two types of size catered for on the falcon pcb


another thing is the patch above is wrong
the ic is too far away from the resistors and the chip used has no bypass cap of its own
{100nf accross + - supply end of cap close as possible to the +5v of the chip}
will cause 32mhz to leak into the supply pin of the pal chip below

[wongck]

I had a lot of issues with my SCSI when I got my CT63

I originally had the 7408 IC but I changed it to the 7404 Patch and now, what I am using is no SCSI Patch at all, but I am using the resistor trick
( Dont know if its helping but since its working flawlessly, I dont intend to remove it to find out )

So, yes, I would say remove it.

Thanks, saw your earlier post about scsi patch as well. Not sure how it went for you as there was no follow up on that thread (or may be I missed it last nite).
So you been there & done that.
Excellent, thanks.

[wongck]

yes remove it totaly when using a ct63 and restore the three 33R resistors {current limiters}

Ok, Just minutes ago a step on to my old 28.8K modem from US Robotics. Wonder if I can find any 33R resistor inside
Yeah. I don't have such resistor handy around the den.

for pal machines only...
i also change the three transistors in the dot clock section {beside the nvram to the right}

Yes, I saw that there is also a bunch of electronics on this particular Falc on RHS of the NVRAM.
They were not on my US Falc. So it's the PAL system stuff.... Thanks, solved the mystery for me
I was like "wtf are these... could it be this stuff killing the scsi?"

another thing is the patch above is wrong
the ic is too far away from the resistors and the chip used has no bypass cap of its own
{100nf accross + - supply end of cap close as possible to the +5v of the chip}
will cause 32mhz to leak into the supply pin of the pal chip below

Thanks... man, that mod will sure have to go then. Seems to be messing up several things.

Thanks !!!

[simbo]

here is http://www.powerphenix.com/CT60/english/fitting63.htm

the fitting and dma problems page
at the bottom of the page is a few things to do...
i would follow this first

[wongck]

Thanks Simbo. I always looked at the pages for CT63.
Both pages does look similar.

[FatRakoon]

Thanks, saw your earlier post about scsi patch as well. Not sure how it went for you as there was no follow up on that thread (or may be I missed it last nite).
So you been there & done that.
Excellent, thanks.

Ok, well at the time it was all confusing for me.

I knew the CT63 was not quite right.
I never had the issues with the SCSI Clock before.
Rodolphe told me to use the 7404 Patch and then told me to not use any patch

There was more to it than that, but I edcided to just ship the CT off to Rodolphe and while he had a look at that, I would restore the Falcon to basics and when I got it back, it all seemed to just work great.

Turns out to be a combination of issues rather than just the one.

[wongck]

Rodolphe told me to use the 7404 Patch and then told me to not use any patch

Ha Ha.... Typical.... can you try this.... no, try this instead..... no no... try this one....
Looks like stabbing in the dark...
I guess he resolved it to what he had on his website, like Simbo said.

[FatRakoon]

No, I dont think thats quite fair.

I think there wasa lot lost in the translation, after all, English is NOT his strong point and french is not mine.

No, I think what went on there, is that I had the 08 Patch and he was telling me that there are various other patches, and he then sent me the 04 IC with the wires soldered on, and said that this would be better.

When I said there does not seem to be any difference, he then told me not to use a patch at all... This just confused me otherwise, why would he send me the 7404 IC and tell me to use it?

[simbo]

its just my point of view

but replacing the three daft gp 2n crap with BC547 will fix the clock
adding a resistor as a termination to the dma is a good idea also
as is changing the 0R resistor to a 33R as mentioned{ive done this now to the nemesis setup}

im sure i also have seen a small bypass capacitor to change the value of

fixes things im sure i have read about it being the cap that bypasses the dma vcc
and also to change some of the values of electrolytics to get more current resoviour

and a last resort was to fix all the grounds of major chips to actual 0v as they are badly connected sometimes
esp rev A and B machines

[wongck]

When I said there does not seem to be any difference, he then told me not to use a patch at all... This just confused me otherwise, why would he send me the 7404 IC and tell me to use it?

I had some experience like that on my ct63. He was suggesting to try this and that.
Stuff like pressing CTRL/SHIFT and up to changing my sdram. There was just a blank screen.
I had to return the board to him in the end.
It turns out there was some production debris and shorted something I guess.

And yes, sometimes it does take some doing to suss out what he says.

[wongck]

but replacing the three daft gp 2n crap with BC547 will fix the clock
adding a resistor as a termination to the dma is a good idea also
as is changing the 0R resistor to a 33R as mentioned{ive done this now to the nemesis setup}
<snip>
and a last resort was to fix all the grounds of major chips to actual 0v as they are badly connected sometimes
esp rev A and B machines

Sorry. please bear with me as I a noobie in electronics.

I don't understand - replacing the 3 gp 2n - that only affects the pal signaling, no?
I am using a VGA and so that has no affect?

My pal system is a revision J. By fixing the grounds you mean reflow of the soldering on Vcc & Gnd pins?

[simbo]

if your machine is a pal type falcon
then there will be a colour clock ~4mhz and a 32mhz crystal
and some components to the right of the nvram

to the right of the nvram is the three transistors in a row Q7 Q8 Q9
they make the 32mhz crystal oscillate and also mix 4mhz with it
for dot sync
if your machine is an american ntsc version
it will have just one crystal oscillator block U65 below
if it has the block then ignore what i said
if its a pal machine it will it tell you on the bottom sticker

maybe beyond some people to change transistors
but im sure its far better to do that than fit a buffer ic etc

here is the two datasheets for you to compare

i would use the C type like bc547C .. its the highest hfe {gain factor}
and has a very reduced noise figure to the 2n3904

http://www.radiovilag.hu/images/BC546_50.pdf
http://www.nxp.com/acrobat_download2/da ... 3904_4.pdf

basicaly atari has used a very poor device here indeed
and its no more than a switching device
not ment for vhf amplification so has very poor output ...amplitude...

its mostly the pal machines suffer from dma issues
never heard of a ntsc guy complain,.

as your in the far east
if you cant get bc547C then use a 2sc1815 but its not as good

[simbo]

http://cgi.ebay.co.uk/100-pcs-BC547B-BC ... 0469939616

100 mixed b and c for about $6 from HK

[wongck]

its mostly the pal machines suffer from dma issues
never heard of a ntsc guy complain,.

Thanks Simbo...
I have both pal & ntsc machine.
the one sufffering from scsi issues is the pal machine.
the ntsc machine have been able to access sscsi but it's never twice the same colour
sorry strike that part about never twice the same colour (ntsc) I can't resist it... No the colour is fine on it

I still got to get around to removing that 7408 patch and getting this 33R resistors.

[wongck]

Need the help of the hardware Jedi master, hardware black belt, hardwire guru again.

The scsi-patch I have in OP. I do not see any cutting of traces. It could be so fine or done else where I cannot find it (like I said, I am a noobie here).
Can someone tell me if there are any cut traces that I need to somehow repair??
Thanks a bunch.!

[simbo]

the clock patch doesnt cut any traces just remove three resistors

so replace the three 33R smd and change the 0R body type resistor to 33R

then change the three transistors and fit the dma termination resistor as per the details
on the above link
you just scrape the paint off the track make the bended part and cut to suite and
then tin the pcb trace and resistor legs you scrapped and solder on the resistor

[wongck]

Thanks Simbo for the clarifications.

[wongck]

So I found some 33R SMT resistors.
Pretty small stuff... No problem.... I have moved some SMT resistors on the floppy driver before and it worked.

Man.... am I so very wrong.
After more than 2 hours, I finally have both of them in place.
Took the Falc from the well lighted dinning table back to the den, plug power, LCD, keyboard, HDD & turn the power on.

Not a single sound... mmm.... I did not plug in the speakers....
1. good I did not short anything.
2. no good as the HDD did not turn on.
3. LCD complain about wrong frequency.

So here we are... I just killed the US$900 machine (by today's pricing ebay 190444845314)

[wongck]

Ok, so what will happen if I just disregard those 33R current limiter?
Will the thing burn?
I need some expert advice before I try it... Thanks,.

[simbo]

fgs dont panic
you prob fried the resistors after so long to solder

you must use tweezers {beeding tweezers and a stong pair of mag glass headset or an eye loop}
and a 17 watt pencil iron wet all contacts with fresh solder first
then solder the resistors .....


now use a multimeter set to diode range
and test all parts make contact

then before all
take a photo of the resistor area using macro mode and post
and dont touch it further!!!


you should have replaced 3 resistors
dont take them off yet ...

[FatRakoon]

Yes, I think you have done what I keep doing... Jumped too quickly into it.

Take your time, you probably havent done any real dammage

The way to solder resistors, is, for me, to "Tin" or melt a tiny little dot of solder onto the place where you want to stick the resistor, and also to melt a bit of solder on the resistor too!
Its a balance of heating up the resistor to a temperature thats hot enough to melt the solder once you touch the solder onto it, and not hot enough to kill the resistor ( or any other component )

But anyway, you heat up the part with the iron and then touch it with the solder wire... If everythign is good, the wire should melt a tiny dot there.

Its somethign you get used to over time.

But anyway, I will put up some pictures of my Falcons SCSI stuff when I was doing mine.

The pictures are not very good because the guy who did the original patch made a serious mess of it. I then tried to apply a second patch before restoring the original 33r resisitors.
I have also done the one Resistor patch further up the board and thats there too!

Ok, well, as you can see here, the trace has been cut, but its also been utterly demolished too
http://www.fatrakoon.co.uk/atari/temppix/wong-2.jpg


I have used standard 33r resistors instead of SMD ones ( Surface mount or small ones )
On this pic however, you can still see the old 7408 and the 7404 ICs - The 7408 is superglued down, and the 7404 is not, but I have just moved the legs out of the way... I have since doing that pic, and after some tests to be sure, now removed the 7404 but the 7408 is still there.

http://www.fatrakoon.co.uk/atari/temppix/wong-3.jpg





I have simply curled up the back leg and looped the front leg over... Dotn knock it. Its done the job


And this is the single resistor under the floppy drive.
http://www.fatrakoon.co.uk/atari/temppix/wong-1.jpg



The only other fixes that I have done, are

The RESET DELAY CAPACITOR on the Falcons MOBO

The SHORT on the CT63... I forget why