23 Jul
2015
23 Jul
'15
6:56 p.m.
A few weeks ago, while I was testing a spare IPB-80 CPU card in an
Intel Series II MDS, the monitor stopped working, with the raster
collapsing to a very bright dot in the center of the screen. I hit the
power switch and pulled the line plug immediately, but the dot
persisted for several minutes, only gradually dimming.
I just started investigating it. With the monitor cable unplugged from
the IOC board, the +15V DC at the IOC connector reads 0.74V. Of the
two DC power supplies described in the service manual, this MDS uses
the Power One. The +15V rail uses a separate transformer winding, a
723C regulator, a house-marked (12500-4) NPN pass transistor, and a
zener and SCR crowbar. The 0.74V makes me think the crowbar has
tripped and the 723 regulator is current limiting. Nothing else in the
MDS uses the +15V supply.
I was a bit lucky that this particular MDS used "method A" of
installing the monitor, which makes it easier to remove. Once the
metal shield is removed, the label was visible. It is a Ball
Electronic Displays TV-120 monitor, which is a common enough model
that it was easy to find the service manual online. I'll test the
monitor on a lab power supply to see if it's drawing too much current.
I may kludge up Rich Ottosen's PIC-based TV test pattern generator on
a breadboard.
24 Jul
24 Jul
7:03 a.m.
The most common failures are the horizontal output transistor,
the flyback transformer and the dampening diode.
Dwight
Date: Thu, 23 Jul 2015 20:10:21 -0600
Subject: Re: diagnosing an Intel Series II MDS monitor failure
From: spacewar at gmail.com
To: cctalk at classiccmp.org
Al wrote:
does it have a horizontal oscillator?
some ball monitors require horizonal drive before the hv works
It doesn't appear to have a horizontal oscillator.
1 Aug
1 Aug
1:09 p.m.
I previously wrote about the monitor of my Intel Series II MDS going
out, which was because the +15V DC supply tripped the crowbar. The
voltage adjust was all the way to the minimum, and the voltage was
still about 17.4V with no load, and trips the crowbar with even a
small load. I think someone previously encountered this problem and
adjusted the pot rather than fixing it, and the problem has gotten
worse since.
The root cause of the failure appears to be that the internal voltage
reference of the uA723, which is specified as 7.15V ?0.35V, is now
9.4V. The uA723 needs to be replaced. The uA723 is still readily
available, but due to the construction of the power supply, replacing
it is relatively difficult. Just probing the pins of the chip was a
challenge. The pass transistors are mounted to 1/8 inch aluminum
plate, but the pins are soldered into rivets in the single-sided PCB,
entering from the etch side (which is closest to the aluminum), so
they will have to be desoldered to get at the etch side to replace the
723. I considered just cutting the pins of the 723 and soldering
another (or a socket) onto it from the component side, but pins 1-7 of
the 723 are right next to a big axial electrolytic so there's no way I
can solder that side. Input and output wires to the board are soldered
into single-pin cup thingies on two opposite edges of the board, so I
can't even get the board into an orientation suitable for working on
it without desoldering or cutting some of the wires.
I haven't bench-tested the monitor, because I haven't kludged up a
suitable signal source yet, but I don't think there's anything wrong
with it.
1:14 p.m.
The root cause of the failure appears to be that the
internal voltage
reference of the uA723, which is specified as 7.15V ?0.35V, is now
9.4V. The uA723 needs to be replaced. The uA723 is still readily
That's an unsual failure. I've had 723s go internally short-circuit and
then the output voltage is essentially the input voltage, but never
seen the reference drift.
available, but due to the construction of the power
supply, replacing
it is relatively difficult. Just probing the pins of the chip was a
challenge. The pass transistors are mounted to 1/8 inch aluminum
plate, but the pins are soldered into rivets in the single-sided PCB,
Ouch, that sounds horrible to work on.
One thing... I had a (SMPSU) that seemed impossible to work on, it was built
a bit like that. Then I found that those 'rivets' were actually sockets and the
transistors were plugged in, not soldered. All I had to do was unto the
mounting screws and it all unplugged. Of course I didn't realise this and
spent some time desoldering things.
I assume you have proved your pass transistors are soldered in place.
-tony
1:26 p.m.
On Sat, Aug 1, 2015 at 2:14 PM, tony duell <ard at p850ug1.demon.co.uk> wrote:
The circuit used in the PS only ties the reference output (pin 6 of
the DIP) to the non-inverting input of the amplifier (pin 5), so I
don't think there's any other likely explanation, though since I
haven't actually been able to pull the board out yet to inspect it
more closely, I suppose there might be a very small chance of
something else being shorted to those pins.
The root cause
of the failure appears to be that the internal voltage
reference of the uA723, which is specified as 7.15V ?0.35V, is now
9.4V. The uA723 needs to be replaced. The uA723 is still readily
That's an unsual failure. I've had 723s go internally short-circuit and
then the output voltage is essentially the input voltage, but never
seen the reference drift. I assume you have proved your pass transistors are
soldered in place.
It sure looks like solder joints, but I should double-check. I was
surprised by solder joints on the top (non-etch) side of the
single-sided board, but I've been told that Power-One sometimes
installed rivets into the PCB holes for that.
1:36 p.m.
On 2015-Aug-01, at 1:26 PM, Eric Smith wrote:
On Sat, Aug 1, 2015 at 2:14 PM, tony duell <ard at
p850ug1.demon.co.uk> wrote:
The circuit used in the PS only ties the reference output (pin 6 of
the DIP) to the non-inverting input of the amplifier (pin 5), so I
don't think there's any other likely explanation, though since I
haven't actually been able to pull the board out yet to inspect it
more closely, I suppose there might be a very small chance of
something else being shorted to those pins.
Just a suggestion: you're sure there no ground reference problem on pin 7 (V-) of the
723? (hence raising the Vref).
I've seen small 723-based supplies with sense leads where the 723 V- is on the neg.
sense lead rather than neg. output lead.
The root
cause of the failure appears to be that the internal voltage
reference of the uA723, which is specified as 7.15V ?0.35V, is now
9.4V. The uA723 needs to be replaced. The uA723 is still readily
That's an unsual failure. I've had 723s go internally short-circuit and
then the output voltage is essentially the input voltage, but never
seen the reference drift.
2:22 p.m.
On Sat, Aug 1, 2015 at 2:36 PM, Brent Hilpert <hilpert at cs.ubc.ca> wrote:
Just a suggestion: you're sure there no ground
reference problem on pin 7 (V-) of the 723? (hence raising the Vref).
I've seen small 723-based supplies with sense leads where the 723 V- is on the neg.
sense lead rather than neg. output lead.
I checked it, and don't see any offset from ground elsewhere in the circuit.
This particular supply rail doesn't use remote sense. They only use it
on the 5V 30A supply, which is on a different board.
4:52 p.m.
There we two different reference voltages on 723s. I don't recall which
was which but one had to change the feedback resistors if you used the
wrong one.
If seems there is only one ref voltage available today.
I think Jameco carries them. What package do you need. They
cam in cans and dips.
Dwight
5:43 p.m.
On Sat, Aug 1, 2015 at 5:52 PM, dwight <dkelvey at hotmail.com> wrote:
There we two different reference voltages on 723s. I
don't recall which
was which but one had to change the feedback resistors if you used the
wrong one.
There's a reference with a built-in current source and amplifier,
which outputs on pin 6 on the DIP. That's the one that's specified as
7.15V ?0.35V.
The 723 in DIP also has a 6.2V zener diode in series from the
regulator output. It's not bonded out in the metal can package.
I've never found a 723 datasheet that gives a different values for
either reference.
The 9.4V that I'm getting from Vref, which is tied to the amplifier
IN+, exceeds the absolute maximum rating for the amplifier input,
which is 8.0V or 8.5V depending on the vendor.
The modern 723 datasheets seem to be dumbed down from the original
ones, with some useful information removed. :-(
2 Aug
2 Aug
9:42 p.m.
From: spacewar at gmail.com
On Sat, Aug 1, 2015 at 5:52 PM, dwight <dkelvey at hotmail.com> wrote:
The one I had to replace on my Nicolet oscilloscope was made by
Lambda. I don't recall the reference voltage but I did have to change
the feedback resistors to get it to work.
I could dig up the schematic if you are interested more.
The Lambda one had failed. It was in a can and I could not get the
same reference voltage so made the resistor change.
Dwight
There we two different reference voltages on
723s. I don't recall which
was which but one had to change the feedback resistors if you used the
wrong one.
There's a reference with a built-in current source and amplifier,
which outputs on pin 6 on the DIP. That's the one that's specified as
7.15V ?0.35V.
The 723 in DIP also has a 6.2V zener diode in series from the
regulator output. It's not bonded out in the metal can package.
I've never found a 723 datasheet that gives a different values for
either reference.
9:56 p.m.
Well, I had to look it up. The LAS723 comes with a 2.5V reference
and not a 7V.
Dwight
10:33 p.m.
On Sun, Aug 2, 2015 at 10:56 PM, dwight <dkelvey at hotmail.com> wrote:
Well, I had to look it up. The LAS723 comes with a
2.5V reference
and not a 7V.
Cool! Thanks for the warning about it. Fortunately the Power-One
supply I'm dealing with only uses normal 723s, but I'll have to bear
that in mind when working on linear power supplies in the future.
9:56 p.m.
> I've never found a 723 datasheet that gives a
different values for
> either reference.
>
Nor have I
The one I had to replace on my Nicolet oscilloscope
was made by
Lambda. I don't recall the reference voltage but I did have to change
the feedback resistors to get it to work.
I seem to remember that Lambda used a semi-custom IC that was the
same as a 723 in pinout and function, but had a different reference
voltage. Whether this was simply to make you get spares from them or
for some actually useful reason I don't know.
Was the failed IC marked '723' or some house number?
-tony
3 Aug
3 Aug
6:18 a.m.
From: ard at p850ug1.demon.co.uk
Was the failed IC marked '723' or some house number?
-tony Hi Tony As I recall, it was marked LAS723-2.5Luckily, before I simply replaced it,
I checked the feedbackresistor values. They didn't make sense for 7V.My guess is that
it made using the 723 for 5V supplies morepractical, although it could be wired up.The
data sheet seems to indicate a better tolerance.The 2.5V references are usually bandgap
and are tightertolerance. As I recall, in the supply I was working on, it was a low
current 12V line.Dwight
6:33 a.m.
Was the failed IC marked '723' or some house number?
-tony
Hi Tony As I recall, it was marked LAS723-2.5Luckily, before I simply replaced it, I
checked the
feedbackresistor values. They didn't make sense for 7V.My guess is that it made using
the 723
for 5V supplies morepractical, although it could be wired up.The data sheet seems to
indicate a better tolerance.The 2.5V references are usually
bandgap and are tightertolerance. As I recall,
in the supply I was working on, it was a low current 12V line.Dwight
Possibly, although for most applications this would not be important.
-tony
3196
days inactive
3206
days old
16 comments
5 participants
participants (5)
-
aek＠bitsavers.org -
ard＠p850ug1.demon.co.uk -
dkelvey＠hotmail.com -
hilpert＠cs.ubc.ca -
spacewar＠gmail.com