FlipChip availability?
dpi
pulled out some of the cards. Nobody admitted to doing this but the
machine was not operational as the R210, R211 cards were all gone.
These are the AC, Link, PC, MB, MA register cards. I was able to
obtain replacements for these from Federated Consultants at the time
for a reasonable price. I really wish I had purchased a few spares
but being short of money at the time I just got the minimum to allow
me to get the CPU working. I also didn't order the cards that were
missing from the DM01 thinking I could do that later when I got the
rest of it running again. I am fighting a broken R211 which seems to
have the MB bit stuck to zero. Since I don't have a card extender
that will support the double wide cards it is a lot more difficult to
diagnose this problem. I saw an R211 mixed in with a bunch of other 8
cards a couple of months ago on Ebay and I bid on it but it went for
too much considering all I wanted was the R211.
I will need to get the DM01 operating if I ever want to fire up my
DF-32's. Instead of cabling in the DF-32 controller they hard wired
it to the DM01 on the backplane. I can run the TC01 and the three
TU-55's dectape drives by cabling the TC01 directly to the processor
but will need the DM01 to operate the DF-32's and the dectape drives
at the same time. I haven't bothered to figure out what is missing on
the DM01. It is too depressing.
Does anyone have a stash of spares they would be willing to sell (or
trade for something) reasonably for a good cause? I would also like
to find another card extender or a dual one if there ever was such a
thing. I have a plan to make one but that will take at least a day of
effort to do a decent job and I would be sacrificing one of my spare
device selector cards to get the card edge connector.
If I can't find some cards I may have to pull all the parts off a
board and copy the foil patterns and etch some new cards. I know
there is no market for this so it would be a labor of love.
Thanks in advance!
Doug Ingraham
David Gesswein
dpi <doug.i...@gmail.com> wrote:
>
>I would also like
>to find another card extender or a dual one if there ever was such a
>thing.
>
http://www.douglas.com/hardware/pcbs/breadboards/digital.html
> I have a plan to make one but that will take at least a day of
>effort to do a decent job and I would be sacrificing one of my spare
>device selector cards to get the card edge connector.
>
If its no longer available they have blank cards.
>If I can't find some cards I may have to pull all the parts off a
>board and copy the foil patterns and etch some new cards. I know
>there is no market for this so it would be a labor of love.
>
I think I remember a posting of someone else looking for those cards.
I don't have any of them.
BobB
What IC's are on the R211, how many ? Since you think it's
narrowed it down to a MB bit, have you metered the
board, sometimes you can catch a bad chip that way or find
a bad discrete component.
If you get the dual card xtender you'll still need a
chip or component.
We use to repair them in the field if we had no spare.
Phil O. Sopher
"dpi" <doug.i...@gmail.com> wrote in message
news:6dfe61a7-0c0a-414f...@l34g2000vba.googlegroups.com...
> I am fighting a broken R211 which seems to
> have the MB bit stuck to zero.
99% of faults on old electronics are due to dry joints. Try touching up
easch joint with a soldering iron.
Of the remaining 1%, 99% of those are due to a dried or leaking
electrolytic.
Christian Corti
> What IC's are on the R211, how many ? Since you think it's
Nice joke ;-))
Christian
dpi
> What IC's are on the R211, how many ?
Zero IC's this is DTL logic.
> Since you think it's
> narrowed it down to a MB bit, have you metered the
> board, sometimes you can catch a bad chip that way or find
> a bad discrete component.
On the Straight 8 the R211 board has more parts than any other. Each
one has a single bit of the PC, MA, MB registers. In this case the MB
is stuck at zero. The PC and MA portions are fine. The test program
I wrote to stimulate the problem is:
0200 7040 CMA
0201 3177 DCA Z 0177
0202 3177 DCA Z 0177
0203 5200 JMP 0200
Since the MB is required to regenerate memory, running the program
will destroy itself unless the bit in all instructions in the program
is zero. In this case bit 3 is always zero so I put the card in that
bit position so I can run the program (CPU Chassis position CD10).
One of the transistors in the flipflop had nonexistent gain, and the
other was really low. However replacing both of these did not correct
the issue. It looks like there might be some diodes that are leaky
but they dont seem to be bad enough to show up on the diode test on my
Fluke. I am going to replace these tonight and start making my dual
extender card if I can't buy one.
> If you get the dual card xtender you'll still need a
> chip or component.
On the Straight PDP-8 99% of all the transistors are PNP types that
are easily substituted. There are only a couple of diode types as
well and they are not critical either. I suspect the parts were
selected based on cost more than anything else.
> We use to repair them in the field if we had no spare.>
This is the first time I have had any difficulty diagnosing one of
these problems. There is nothing that looks really broken now. I
will figure it out. Probably with a forehead smack and an exclamation
of why didn't I see that before.
Doug
BobB
I have a dual height extender somewhere, I think, just
have to dig it out. Plus I have some 8 components, diodes/
xsistors if you narrow the problem down and can't find a
replacement.
Where are you located ?
Good luck, nice to see you do component level repair.
Mike Ross
wrote:
>
>"dpi" <doug.i...@gmail.com> wrote in message
>news:6dfe61a7-0c0a-414f...@l34g2000vba.googlegroups.com...
>> I am fighting a broken R211 which seems to
>> have the MB bit stuck to zero.
>
>99% of faults on old electronics are due to dry joints. Try touching up
>easch joint with a soldering iron.
I seem to recall reading of one restorer who made a point of giving every single
board in a machine, faulty or not, a quick pass over a flow solder machine.
Overkill? Thoughts?
Mike
--
http://www.corestore.org
'As I walk along these shores
I am the history within'
dpi
> I seem to recall reading of one restorer who made a point of giving every single
> board in a machine, faulty or not, a quick pass over a flow solder machine.
> Overkill? Thoughts?
On the Flip Chips you could do this but you would need to protect the
edge connector somehow and remove the card handles. I get the feeling
from working on these that the solder used may not have been 60/40 as
it doesn't act the same when touching it up. Also exposed leads on
components are oxidized and might need some sort of cleaning. The
holes were not plated through on the double sided boards. Instead
they used something like a pop rivet which the solder flowed through.
If I had to guess I think the boards were first stuffed with the thru
hole pop rivets and then run through the soldering process component
side down to flow solder the pop rivets to the traces on the top side
of the board. Then they were stuffed with the components and run
through the solder bath normally. The solder on both sides of the
board is quite a bit thicker than normal over the traces. These
boards do not have solder masks or silk screening.
If you had a flow solder machine this might not be a bad idea. Not
sure what you would need to do to clean up the board before the
process. Some way to remove the surface oxidation at the very
minimum.
Doug Ingraham
dpi
Fixed the R211 card. I replaced a couple of low gain transistors that
make up the flipflop and ended up making it worse. The pulse type
logic DEC created for this requires pretty fast transistors although
the gain doesn't need to be that high. I put the transistors back in
and removed the transistors that cause the increment to occur (ISZ/
Data break/page zero auto increment) and now the MB flipflop worked
again. These other transistors were NPN's and one had a gain (hfe) of
11 and the other a gain of 18. It looked like just about any NPN
should work so I put in a couple of 2N2222's (original was a 2N3009)
and still no joy. It appears that these also need to be very fast
transistors. By very fast I mean that some of the data sheets call
these ultrafast and they can operate at speeds between 300 and 500
mhz. I found some subs that seem to work in a pack of generic NPN's
radio shack sells (sold?). So I am going on a hunt to buy a couple of
hundred of each type if I can find them. They really should sell for
20 cents each if I can find them.
So my 8 is back to working again (sort of). The memory checkerboard
only runs about 3 minutes so I need to figure out what is causing
that. It isn't detecting a real error when it stops. What is
happening is that the program is corrupting itself and then falsely
detecting an error. A few bits are being set when the contents of the
program are written back to memory during execution. It looks like
one of the sense amps is picking bits.
Doug Ingraham
Rapid City, SD
BobB
A similar component discussion and search for the same 2N3009 at:
http://archive.netbsd.se/?ml=cctech&a=2009-06&t=10855922
van...@vsta.org
> ... so I put in a couple of 2N2222's (original was a 2N3009)
> and still no joy. It appears that these also need to be very fast
> transistors. By very fast I mean that some of the data sheets call
> these ultrafast and they can operate at speeds between 300 and 500
> mhz.
I would never, ever have guessed that any PDP-8 of any vintage would
require a 300Mhz part. Fascinating.
Andy Valencia
dpi
It is not clear exactly why this particular application needed the
high speed transistor. I think it might be more the way it is being
used making the circuit sensitive to parasitic capacitance on the base
lead and the higher speed part has less capacitance. The 2N2222 I
used was the A version in the metal can. I was surprised that this
didn't do the job. It might be possible to use the 2N2222 if some
other parts in the circuit got changed as well. The basic clock in
the PDP-8 runs at 1,333.333 khz but this is used to generate TP1 and
TP2 so the actual machine operates at half that rate or 666.666khz.
If it weren't for the core memory speeds you could probably run at
several megahertz. I am not sure what the next limiting thing would
be. Probably the widths of the pulses in the stuff that uses pulses
which would end up being around 2.5mhz instruction rate. About 4
times faster and that would have been HUGE at the time. Not sure how
this would have affect the IOT's, they might need to be slowed down
for compatibility.
The guys that designed this stuff were very clever engineers and often
came up with very simple ways to save some components.
Doug Ingraham