Topics

Another scope 7854

Bruce Griffiths
 

No its just the usual formation of gold-tin intermetallics:

http://www.semlab.com/papers/goldembrittlementofsolderjoints.pdf

http://www.semlab.com/papers/intermetallicsinsolderjoints.pdf

However bonding of aluminium wires to gold pads can result in Aluminium gold intermetallics

One of these intermetallics (AuAl2) is purple (hence the name purple plague) and another (Au5Al2) is white in color.

Bruce

On 05 May 2017 at 16:41 "David @DWH [TekScopes]" <TekScopes@...> wrote:



You reminded me that I know of another failure in these vertical
plug-ins. Some of the early ones have a heavy gold plating on the
printed circuit board traces and the tin-lead solder can eventually
suffer from gold embrittlement breaking the solder connection.

If this happens, resoldering the connection will reveal what I suspect
is Purple of Cassius under the solder.

<https://en.wikipedia.org/wiki/Purple_of_Cassius>

The solution I found was to use a soldering iron on low heat to
repeatedly add and remove solder until the Purple of Cassius and gold
is removed. The tin in the solder dissolves the gold until the base
metal is reached. This is actually the recommended soldering
technique for heavy gold plating if tin solder is used.

Modern flash gold plating is deliberately thin enough to completely
dissolve into the solder preventing this problem.

On Thu, 4 May 2017 19:23:10 +0000, you wrote:

>I had a 7A26 with intermittent vertical instability in channel 2 only. I found that alongside the attenuator plug-in modules, there are discrete 1/8 watt resistors. They are also plugged in and one had developed high contact resistance. I replugged it about 4 times after using a q-tip to swab its leads with contact cleaner, being careful not to get it anywhere except on the ends of the leads. The problem never came back.
>
>Cliff Carrie





 

On Thu, 4 May 2017 17:38:45 -0500 (CDT), you wrote:

yOn Thu, 4 May 2017, David @DWH [TekScopes] wrote:

A shorted C375 shown on schematic 4 would cause about that much high
gain. A shorted C445 shown on schematic 3 could cause it if R445 is
adjusted to a low value.
Wow! I want to be you when I grow up! C375 was indeed shorted. I
understand why this affects the gain, and would be able to follow it
better if it were a modern OpAmp rather than discreet transistors (I can't
visualize those as well yet).
It helps that I am somewhat familiar with the details of the inside of
U270. Pins 8 and 9 expose the emitters of the channel 2 input
differential amplifier which has about 50 ohms of resistance to set
the gain. With C375 shorted, the 10 ohms of R375 is placed in
parallel with the internal 50 ohm resistance raising the gain by about
5 times.

Your observation that the gain was about 5 times high was a major
clue. There just was not that many things which could cause that
without other symptoms.

I got a suitable replacement from my local store and will swap it out
later today.
Recalibrating the transient response is not trivial on these; it
requires a reference level pulse generator like a PG506. What you
could do instead is apply the same fast pulse to both channels and
adjust C375 until the pulse response of channel 2 is as close as
possible to channel 1. If you do not have anything better, then the
sync or trigger output from a function generator is probably fast
enough.

The occasional triggering glitch I think must by the cam finger switches
on the AC/DC/GND/DC Offset switch. I got them to go away on channel 1
through exercising the switch, but channel 2 still has this glitch. I need
to figure out the best way to physically access these to clean them with
as little disassembling as possible. Will have a look at that too.
This problem might also be caused by the gold soldering issue that I
mention in my other post. Remove the aluminum cover over channel 2
and do a close inspection of the connections to the input circuits.
Mine had a problem with the AC coupling capacitor.

Also, in addition to the averaging the 7854 does have the cursors and
measurements built-in over the 7904. They are actually relatively easy to
use, about the same as my 2232, though having a knob for the cursors is
nicer than the buttons.
That is true but the 2230 and 2232 operate in real time and have a
much better interface. The complexity of using the 7854 limits most
users to its basic functions.

Did you know that the 2230/2232 can make ground referenced absolute DC
measurements? When the coupling switch is set to ground in storage
mode, the level is recorded and marked on the display as the first
couple of points of the waveform record. When one of the cursors is
placed on that spot, the voltage readout changes to show a ground
symbol indicating that the other cursor is now measuring relative to
ground.

The 7854 has a similar function which sets the ground reference level
with a special aquisition cycle.

 

On Thu, 4 May 2017 21:14:41 -0500, you wrote:

Trigger off "Mode" (in between Ch1 and Ch2) also has the glitches
"Mode" means that the trigger source depends on the vertical mode
selected:

Channel 1 - Channel 1
Channel 2 - Channel 2
Add - Channel 1 added to Channel 2
Chop - Channel 1 added to Channel 2
ALT - Alternating between Channel 1 and Channel 2

I will keep digging around... any insights appreciated.
The 7000 series uses a separate internal trigger path which originates
inside the vertical plug-ins and the dual trace plug-ins have their
own internal trigger channel switch which is shown on schematic 4.
This is what allows the horizontal slots to use any of the 4 channels
as a trigger source no matter what is displayed.

My guess is that there is a problem at the trigger channel switch
having to do with channel 2. Maybe the levels going into pins 4 and
14 are wrong or one of the parts connected to pins 8 and 9 is open or
shorted.

I doubt the channel switch is bad but U270 and U470 can be swapped to
find out.

 

I originally thought that but when I ran across the problem, removing
the solder revealed that the entire solder to gold interface was a
purple-brown color. I suspect this was a long term effect of whatever
flux Tektronix used during assembly.

On Fri, 5 May 2017 17:23:04 +1200 (NZST), you wrote:

No its just the usual formation of gold-tin intermetallics:

http://www.semlab.com/papers/goldembrittlementofsolderjoints.pdf

http://www.semlab.com/papers/intermetallicsinsolderjoints.pdf

However bonding of aluminium wires to gold pads can result in Aluminium gold intermetallics

One of these intermetallics (AuAl2) is purple (hence the name purple plague) and another (Au5Al2) is white in color.

Bruce

Craig Sawyers <c.sawyers@...>
 

I originally thought that but when I ran across the problem, removing the solder revealed that the
entire solder to gold interface was a purple-brown color. I suspect this was a long term effect
of
whatever flux Tektronix used during assembly.
Using tin-lead solder on gold plating is not a good thing. In space application, component leads are
gold plated to prevent oxidation of the underlying wire and maintain solderability. But the leads
are "degolded" prior to assembly using a solder bath (the gold plating dissolves in the solder). If
you don't degold, you get contact embrittlement because of the gold/tin/lead intermetallics.

Do a google search on degolding - there is a bunch of stuff on there including you tube videos, and
automatic machines to bulk degold.

Craig

Joe Laffey
 

yOn Fri, 5 May 2017, David @DWH [TekScopes] wrote:

On Thu, 4 May 2017 21:14:41 -0500, you wrote:

Trigger off "Mode" (in between Ch1 and Ch2) also has the glitches
"Mode" means that the trigger source depends on the vertical mode
selected:
Yes. Thanks for clarifying.


I will keep digging around... any insights appreciated.
The 7000 series uses a separate internal trigger path which originates
inside the vertical plug-ins and the dual trace plug-ins have their
own internal trigger channel switch which is shown on schematic 4.
This is what allows the horizontal slots to use any of the 4 channels
as a trigger source no matter what is displayed.

My guess is that there is a problem at the trigger channel switch
having to do with channel 2. Maybe the levels going into pins 4 and
14 are wrong or one of the parts connected to pins 8 and 9 is open or
shorted.

I doubt the channel switch is bad but U270 and U470 can be swapped to
find out.
I will try this. I did re-seat the two ICs.

The fact that the glitch waveform appears like it is being triggered by the opposite slope I wondered if it didn't have to do with the inverting part of the circuit.... The bias network around R428 perhaps (between Q426 and Q526)? Note I have Option 6 (the DC offset).

I am curious why enabling DC Offset coupling seems to fix the issue (I am not 100% sure it fixes it completely; I may have seen the glitch once with it on). All that DC Offset switch does is add DC bias to the base of Q520.

Also, with the invert switch turned ON (ch 2 inverted) the glitches occur MUCH less frequently, like every 2-5 seconds vs. 1-3 times per second.

I did see that the base of Q426 was about 4.6V DC, rather than the 3.2V on the shcematic. (in non-invert, as described in the manual). If Q426 and Q526 were near the edge of cutoff perhaps it would cause this issue?

FWIW the voltage rails in the plugin are spot on.

I did finally realize that most transistors are indeed in sockets as you or another list member said. At first I thought those sockets where just the old plastic spacers that hold transistors off the board. I was wrong. Anyway, re-seating made no difference.



--
73
Joe Laffey
The Stable
Visual Effects
http://TheStable.tv/?e40769M/

Joe Laffey
 

On Fri, 5 May 2017, David @DWH [TekScopes] wrote:

On Thu, 4 May 2017 17:38:45 -0500 (CDT), you wrote:

I got a suitable replacement from my local store and will swap it out
later today.
Recalibrating the transient response is not trivial on these; it
requires a reference level pulse generator like a PG506. What you
could do instead is apply the same fast pulse to both channels and
adjust C375 until the pulse response of channel 2 is as close as
possible to channel 1. If you do not have anything better, then the
sync or trigger output from a function generator is probably fast
enough.

For square waves I have a PG502, an FG504, an IEC F34, a Wavetek 145, and an old DataPulse 101. I think the PG502 is the fastest of those. I could also build a Jim Williams fast edge circuit.

When you say refernece level you mean that the peaks have to be very accurate calibrated? (The amplitude of the square wave) ??

Thanks again.

--
73
Joe Laffey

Bruce Griffiths
 

The older heavily gold plated HP boards used to develop distinctive dark gold tin intermetallics between the solder and the board after a few years. The forrmation of such intermetallics is why a solder pot is used to remove gold from the solder pockets on connectors for high reliability wiring harnesses. Indium solder is another approach but its more susceptible to corrosion

Bruce

On 05 May 2017 at 17:55 "David @DWH [TekScopes]" <TekScopes@...> wrote:



I originally thought that but when I ran across the problem, removing
the solder revealed that the entire solder to gold interface was a
purple-brown color. I suspect this was a long term effect of whatever
flux Tektronix used during assembly.

On Fri, 5 May 2017 17:23:04 +1200 (NZST), you wrote:

>No its just the usual formation of gold-tin intermetallics:
>
>http://www.semlab.com/papers/goldembrittlementofsolderjoints.pdf
>
>http://www.semlab.com/papers/intermetallicsinsolderjoints.pdf
>
>However bonding of aluminium wires to gold pads can result in Aluminium gold intermetallics
>
>One of these intermetallics (AuAl2) is purple (hence the name purple plague) and another (Au5Al2) is white in color.
>
>Bruce






[Non-text portions of this message have been removed]

 

On Fri, 5 May 2017 01:33:10 -0500 (CDT), you wrote:

On Fri, 5 May 2017, David @DWH [TekScopes] wrote:

Recalibrating the transient response is not trivial on these; it
requires a reference level pulse generator like a PG506. What you
could do instead is apply the same fast pulse to both channels and
adjust C375 until the pulse response of channel 2 is as close as
possible to channel 1. If you do not have anything better, then the
sync or trigger output from a function generator is probably fast
enough.
For square waves I have a PG502, an FG504, an IEC F34, a Wavetek 145, and
an old DataPulse 101. I think the PG502 is the fastest of those. I could
also build a Jim Williams fast edge circuit.

When you say refernece level you mean that the peaks have to be very
accurate calibrated? (The amplitude of the square wave) ??

Thanks again.
A reference level pulse generator has a fast edge with a very clean
pulse response and a minimum of aberrations so it can be used as an
absolute reference for transient response. The 7A18 is only 75 MHz so
your PG502 is fast enough and I think you can get by the clean edge
requirement by applying the pulse to both channels and adjusting C375
so that channel 2 matches channel 1.

 

On Fri, 5 May 2017 07:11:08 +0100, you wrote:

I originally thought that but when I ran across the problem, removing the solder revealed that the
entire solder to gold interface was a purple-brown color. I suspect this was a long term effect
of
whatever flux Tektronix used during assembly.
Using tin-lead solder on gold plating is not a good thing. In space application, component leads are
gold plated to prevent oxidation of the underlying wire and maintain solderability. But the leads
are "degolded" prior to assembly using a solder bath (the gold plating dissolves in the solder). If
you don't degold, you get contact embrittlement because of the gold/tin/lead intermetallics.

Do a google search on degolding - there is a bunch of stuff on there including you tube videos, and
automatic machines to bulk degold.

Craig
One of the NASA papers on the subject recommended removing the gold by
dissolving it in tin which is what I ended up doing. It took a few
passes of applying solder and removing it. The gold dissolves in the
tin better at higher temperatures but I had better results at a low
temperature applied for a longer time to avoid damaging the
polysulfone board substrate.

 

That sounds like the same thing but the color I observed was
definitely purple. It looked very much like the result of mixing
purple and yellow crayons.

I considered indium solder but removing the gold by dissolving it in
tin worked well enough and was less expensive.

On Fri, 5 May 2017 22:22:14 +1200 (NZST), you wrote:

The older heavily gold plated HP boards used to develop distinctive dark gold tin intermetallics between the solder and the board after a few years. The forrmation of such intermetallics is why a solder pot is used to remove gold from the solder pockets on connectors for high reliability wiring harnesses. Indium solder is another approach but its more susceptible to corrosion

Bruce

Bruce Griffiths
 

The colour of thin films can be misleading in that interference effects in thin films can impart colour to transparent films like soap bubbles, beetle wings, oxide films on steel produced by heat treatment etc.

Bruce

On 06 May 2017 at 00:53 "David @DWH [TekScopes]" <TekScopes@...> wrote:



That sounds like the same thing but the color I observed was
definitely purple. It looked very much like the result of mixing
purple and yellow crayons.

I considered indium solder but removing the gold by dissolving it in
tin worked well enough and was less expensive.

On Fri, 5 May 2017 22:22:14 +1200 (NZST), you wrote:

>The older heavily gold plated HP boards used to develop distinctive dark gold tin intermetallics between the solder and the board after a few years. The forrmation of such intermetallics is why a solder pot is used to remove gold from the solder pockets on connectors for high reliability wiring harnesses. Indium solder is another approach but its more susceptible to corrosion
>
>Bruce





 

On Fri, 5 May 2017 01:24:00 -0500 (CDT), you wrote:

yOn Fri, 5 May 2017, David @DWH [TekScopes] wrote:

The 7000 series uses a separate internal trigger path which originates
inside the vertical plug-ins and the dual trace plug-ins have their
own internal trigger channel switch which is shown on schematic 4.
This is what allows the horizontal slots to use any of the 4 channels
as a trigger source no matter what is displayed.

My guess is that there is a problem at the trigger channel switch
having to do with channel 2. Maybe the levels going into pins 4 and
14 are wrong or one of the parts connected to pins 8 and 9 is open or
shorted.

I doubt the channel switch is bad but U270 and U470 can be swapped to
find out.
I will try this. I did re-seat the two ICs.

The fact that the glitch waveform appears like it is being triggered by
the opposite slope I wondered if it didn't have to do with the inverting
part of the circuit.... The bias network around R428 perhaps (between
Q426 and Q526)? Note I have Option 6 (the DC offset).
I have seen this before where the waveform edges were distorted by a
lot of ringing. This might be produced if one of the active stages is
saturating or clipping.

I am curious why enabling DC Offset coupling seems to fix the issue (I am
not 100% sure it fixes it completely; I may have seen the glitch once with
it on). All that DC Offset switch does is add DC bias to the base of Q520.

Also, with the invert switch turned ON (ch 2 inverted) the glitches occur
MUCH less frequently, like every 2-5 seconds vs. 1-3 times per second.
In both cases, the operating point of the active stages changes. The
position control does the same thing; does it have any effect on the
glitches?

I did see that the base of Q426 was about 4.6V DC, rather than the 3.2V on
the shcematic. (in non-invert, as described in the manual). If Q426 and
Q526 were near the edge of cutoff perhaps it would cause this issue?
There are two version of the 7A18. The schematics show that the older
design has +3.2 volts at the base of Q426 and the newer design has
+4.7 volts at the base of Q426.

The same signal is applied to the signal channel switch and the
trigger channel switch so if it was distorted at that point, then I
would expect it to show up on the screen.

With the traces centered, I would compare the DC voltages between the
two channel switches with different channels selected to see if one
obviously has the wrong operating point.

I would also apply a square wave to channel 2 and measure the waveform
at the output of the trigger channel switch and at the collectors of
Q480 and Q580 to see if it is distorted. The signal levels at those
points are high enough to allow this.

FWIW the voltage rails in the plugin are spot on.

I did finally realize that most transistors are indeed in sockets as you
or another list member said. At first I thought those sockets where just
the old plastic spacers that hold transistors off the board. I was wrong.
Anyway, re-seating made no difference.
Those are Berg Miniserts. If you are interested and cannot find the
datasheet online, then I can make it available.

 

The color was awfully uniform compared to the transparency for that.
It looked a lot like a chemical stain but was difficult to remove; I
just used tin solder, low heat, time, and agitation.

The gold plating was really thick which is no surprise given what
happened. The unit was pretty old and I can probably find it to get
the rough date of manufacturing. I suspect either the problems with
tin solder and gold were obscure at the time or the designer at
Tektronix was just unaware of them.

It still struck me as an odd design mistake and I wonder if Tektronix
originally intended to use a gold compatible solder. Later boards of
the same design used a much thinner gold plating although it still
looks thicker than the flash gold over nickel that is normally used
now.

On Sat, 6 May 2017 01:21:34 +1200 (NZST), you wrote:

The colour of thin films can be misleading in that interference effects in thin films can impart colour to transparent films like soap bubbles, beetle wings, oxide films on steel produced by heat treatment etc.

Bruce

Craig Sawyers <c.sawyers@...>
 

One of the NASA papers on the subject recommended removing the gold by dissolving it in tin which
is
what I ended up doing. It took a few passes of applying solder and removing it. The gold
dissolves in
the tin better at higher temperatures but I had better results at a low temperature applied for a
longer
time to avoid damaging the polysulfone board substrate.


------------------------------------
Posted by: David <@DWH>
------------------------------------
We looked at a number of esoteric board substrates whose CTE was matched to typical gullwing IC's
and surface mount passives for strain-free with temperature cycling. I forget what we chose
eventually - but it was out of production (the certified board house had hoarded a load of the
stuff).

Plus there was a whole load of other requirements - no solder resist, no silk screen. And no traces
on the top and bottom - everything had to be in an inner layer other than ground planes. We pretty
quickly ended up with ten layers.

Craig

Craig Sawyers <c.sawyers@...>
 

The gold plating was really thick.
This was inherent in the way that Tek made this boards. They essentially used gold plating as an
etch resist - there is a period Tek video out there that shows the whole process.

Craig

 

On Fri, 5 May 2017 16:17:00 +0100, you wrote:

One of the NASA papers on the subject recommended removing the gold by dissolving it in tin which
is
what I ended up doing. It took a few passes of applying solder and removing it. The gold
dissolves in
the tin better at higher temperatures but I had better results at a low temperature applied for a
longer
time to avoid damaging the polysulfone board substrate.
We looked at a number of esoteric board substrates whose CTE was matched to typical gullwing IC's
and surface mount passives for strain-free with temperature cycling. I forget what we chose
eventually - but it was out of production (the certified board house had hoarded a load of the
stuff).

Plus there was a whole load of other requirements - no solder resist, no silk screen. And no traces
on the top and bottom - everything had to be in an inner layer other than ground planes. We pretty
quickly ended up with ten layers.

Craig
This was only the high impedance attenuator board so I suspect
Tektronix used a polysulfone substrate in desperation to solve the
problem with hook in FR4 type substrates until they could find a
better substrate. Later substrates were made of some white plastic
material which I have not identified.

Polysulfone has the advantage of being one of the higher temperature
thermoplastics so it works as a printed circuit substrate where
soldering will be used.

At least it *looks* like polysulfone. Maybe it is something else but
polysulfone fits the requirements and can be the right color and
transparency.

Chuck Harris
 

A really cool thing happens with gold plating, the gold atoms
are huge and the space between them fits all sorts of things,
such as etchant, oxygen, mercury, uranium hexafloride, and
other fun stuff. The gold, if plated too quickly, gets a
little mossy, which is why it appears matte finish rather
than shiny, and makes it even more susceptible to infiltration of
undesirable things. For instance, the electrolyte in batteries
like alkalines and nicads goes right through gold plating...

As a result, the mil specifications for gold plated fingers
requires a thin layer of nickel plating between the copper and
the gold.

I worked at one company, Defense Electronics, Inc (DEI) as a
metal plater, and they routinely flouted the rules on gold
plating. It wasn't so much that they didn't want to follow
the rules, but rather, that they couldn't afford my time, and
their money, to repair/replace their nickel plating tank. So,
they doubled up on the gold plating... at $32/oz, why not?

I later went on to a company called Electro Products, Inc, which
made circuit boards, and they too cheated on the gold plating,
no nickel inter layer.

To be fair, there was a lot of inbreeding amongst the defense
related electronics companies back then...

-Chuck Harris

David @DWH [TekScopes] wrote:

The color was awfully uniform compared to the transparency for that.
It looked a lot like a chemical stain but was difficult to remove; I
just used tin solder, low heat, time, and agitation.

The gold plating was really thick which is no surprise given what
happened. The unit was pretty old and I can probably find it to get
the rough date of manufacturing. I suspect either the problems with
tin solder and gold were obscure at the time or the designer at
Tektronix was just unaware of them.

It still struck me as an odd design mistake and I wonder if Tektronix
originally intended to use a gold compatible solder. Later boards of
the same design used a much thinner gold plating although it still
looks thicker than the flash gold over nickel that is normally used
now.

On Sat, 6 May 2017 01:21:34 +1200 (NZST), you wrote:

The colour of thin films can be misleading in that interference effects in thin films can impart colour to transparent films like soap bubbles, beetle wings, oxide films on steel produced by heat treatment etc.

Bruce

------------------------------------
Posted by: David <@DWH>
------------------------------------


------------------------------------

Yahoo Groups Links



 

The datasheet for the Berg Miniserts is in the Files area.

Dave Partridge

Joe Laffey
 

On Fri, 5 May 2017, David @DWH [TekScopes] wrote:

I did a lot of poking around this evening.

I put a scope on the output of both the trigger and signal followers Q280-Q380 Q480-Q580. I could not see any glitches in phase on these signals. I was triggering off the sync out from the siggen. I did notice a fair amount of noise on the signals, but that could just be the probes and the ground leads, etc. ** Shouldn't I see the glitching on these waveforms if the problem is in the vertical plugin?

I was able to get the glitches to occur with both Ch 1 and Ch 2 triggering. They are however much more often (3-5x more often) on the channel 2 trigger.

The glitches seem to only occur when the trigger level is negative and the slope is positive, or the trigger level is positive and the slope is downward on a normal centered signal (think AC coupled, or 0 V baseline).

It is very difficult to get the glitches to occur with a square wave. It really wants to have a slower rise time to cause the problem. Triangle and sine waves do it fine.



** Or perhaps the noise is what is causing the triggering. However the waveforms displayed on the 7854 are not particularly noisy. They look pretty thin and normal for a ~100MHz scope (with this plugin).

I am beginning to wonder if the problem might lie in the horizontal amplifier. I am pretty confused at this point.

Changing the trigger source form Ch 1 to Ch 2 DOES make a difference, even when displaying just Ch 1 (or Ch2). I have a T at the siggen and two wires come down to the scope. Both wires have terminator on them (same problem happens without termination). I tried another siggen and had the same problem. I tried putting this siggen into another scope and it triggered and looked fine. I have tried the internal calibrator, and I have seen the issue, but it is much less common on a square wave, especially a higher frequency square wave.

I swapped the two switching ICs to no avail. The DC voltages on the switching ICs read like the reference.

I should receive another Vertical amplifier plugin next week, and I guess I can use that to rule out or in the timebase.

The timebase is a 7B53A.

--
73
Joe Laffey