Suggestions for VT103?

Jerome H. Fine jhfinedp3k at compsys.to
Sun Dec 28 20:36:29 CST 2008


 >Tony Duell wrote:

>I refuse to beleive it takes you over an hour to solder one connection! :-)
>  
>
Jerome Fine replies:

Well, maybe soldering a VT103 backplane was not so bad, but I seem to 
remember the
problems I once had with a DRV11 module.  It needed to be strapped, but 
the only way that
DEC provided to change the CSR value was with zero ohm resistors.  
Removing a strap just
meant cutting out a resistor.  But I found it impossible to add a 
resistor.  Finally after several
hours of unsuccessful attempts (and almost damaging the board), I solved 
the problem by
removing a few male connector pins from a damaged board and soldering 
them into the
appropriate holes where the zero ohm resistor leads would normally be 
placed.  Since the
very tiny pin was easy to manipulate and could be easily inserted into a 
melted solder hole,
I ended up with two pins which I could them wire wrap since the pins 
were very similar to
wire wrap posts in the first place.  Why DEC had not done that to start 
with I don't know,
but I finally did change the CSR value.  And it took about TWO hours 
each to finally insert
each pin.

Needless to say, I am not a bit fan of making solder connections at this 
point.

>>As far I my experience is worth, the upgrade to a 22 bit backplane with 
>>the version DEC
>>provides in the VT103 works VERY well.  I watched over at least 6 VT103s
>>
>That does not suprise me. I was under the impression that some very early 
>Q-bus modules used at least one of those pins for something else (I've 
>seen 3rd-party Q-bus cards with 22 bit DMA capability where the bus 
>driver for thsoe upper 4 address lines (normally a '38 or similar) is 
>socketed with instructions to remove it if used in certain backplanes) 
>but I suspect the VT103 is late enoguh for this not to be an issue.
>  
>
As far as I know, the LSI-11 CPU modules (both dual and quad) do use 
some of those address lines.

But the M8186, M8189, M8190 and M8192 can't since they all support 22 
bit addresses for
memory.

So the VT103 backplane from DEC with 18 bit address lines probably 
supports the use of
the LSI-11 CPUs, but not after modification to 22 bit addresses.  Since 
the PDP-11/73 CPUs
are now readily available, I can't see anyone using an LSI-11 CPU at 
this point except in VERY
unusual situations which require the LSI-11 CPU for a special reason - 
like the microcode which
can be modified.  I don't know of anyone who ever modified the microcode 
for an LSI-11 CPU.

That is not to say that I will not invent new PDP-11 instructions such 
as an UNSIGNED multiply,
32 bit multiply and divide which will be implemented under Ersatz-11.  
But as the fellow in Irma
La Duce said "That is another story."

>>The really cool reason to use a VT103 is that a hard drive can be placed 
>>right under the CRT.
>>
>I wonder about stray magnetic fields from the yoke and/or flybackj 
>transformer. Not that they'll corrupt the magnetic patterns on the disk, 
>but that they'll be picked up by the read amplifier anf cause random data 
>erros. But I guess it works OK.
>  
>
Not knowing about stray magnetic fields, I just put an ST412 (actually a 
DEC RD51) under the
tube and started to run.  This used a Sigma RQD11-B MFM controller (dual 
board with boot
ROMs) with an M8186.  Worked great.  Made them available to Ontario 
Hydro as a work
station.  Since they were already using the VT103 with a dual RX02 
floppy drive, the hard
drive was a huge improvement.  They ran RT-11 and having a 10 MByte hard 
drive rather
than a 0.5 Mbyte floppy made a huge improvement.  Expensive at the time, 
but worth while
for commercial use.

>>The one problem of using the VT103 is that the power supply is really 
>>too limited, although with
>>only 4 slots, not a lot of power needed.  Tony, perhaps you might be 
>>able to suggest how
>>the 5 amp supply could be enhanced?  On the other hand, with a BA23 
>>    
>>
>
>Do you mean '5 amp' or '5 volt' here? I was under the impression it was 
>around 15A or so at 5V.
>  
>
Yes!  I did mean the 5 Volt which is limited to 16 Amps on the VT103.  
And that includes
all of the boards, including the VT100 video card and anything else in 
the VT100 which uses
the 5 Volt level.

>Increasing the rating of a PSU is not easy in general. Many of the 
>components would need replacign with higher-rated parts, including the 
>transformer (whether linear or switch-mode), the rectifiers, smoothing 
>capacitors (increase in capacitance value), chopper transistors (if an 
>SMSPU), pass transistors (if a linear design), etc.and of course you'd have 
>modify any current limit circuitry. It'd probably be easier to design a 
>replacement PSU from scratch to fit in the same space.
>  
>
I thought as much.  I will continue to use the BA23 and BA123 for now.  
Since the core 2 duo
CPU runs Windows XP which runs Ersatz-11 which runs RT-11 at more than 
100 times the
speed of a PDP-11/93, it is not likely that I will be using a real DEC 
CPU much in any case.
By the way, with SATA II drives, the disk I/O is probably 200 times 
faster than any ESDI or
SCSI drive connected to a PDP-11.

>>As for modes of failure, how often should a power supply be used to be 
>>sure that keeping it out
>>of service does not cause a failure when the power supply is used after 
>>a few years?  Does anyone
>>have any recommendations?
>>
>About the only thing that'll fail from not being used are electrolytic 
>capacitors, and I am not convinced this is a major problem with 
>modern-ish ones. Certainly it's not a failure I've ever encountered (yes, 
>I've had electrolytics fail, but not by the oxide-film disolving due to 
>them not being used).
>
I probably turn on the PDP-11/83 about 2 times a year.  Since I had 2 
BA123 power supplies
fail in the past 10 years, I wondered about having to use them or loose 
them.  At one point,
I was told by a company that I did some software programming for that 
their major customer
required them to run the PDP-11 systems every 3 months until delivery 
which was not to be
for 2 years.  Thus the reason for my question.

Sincerely yours,

Jerome Fine



More information about the cctalk mailing list