Topics

7T11 horizontal memory


cmjones01
 

Is there anyone here intimately familiar with the internal workings of the 7T11 plugin? I ask because I'm trying to get one to work. The symptoms are that there's no trace visible on the screen.

I had it on an extender in the bench today and established that most of it is working: the trigger circuitry, strobe pulse generator and slow scan all seem to be doing what they should. However:
- the trace is blanked because the X/sweep output is out of range (it's at about +15V)
- it seems to be out of range because the horizontal memory isn't getting any strobe pulses
- the holdoff multivibrator, which generates said pulses, looks like it's working
- there's a peculiar arrangement of resistors, a diode and a 100pF capacitor which takes the outputs of the multivibrator (a 7400 chip) and is supposed to switch on a PNP transistor (Q546 I think) which has its emitter tied to ground (in equivalent time sampling mode). This is not happening. The transistor's base never goes below ground so it never switches on.

Looking at the circuit diagram, I can't see how it's supposed to work. I don't have the manual in front of me right now but does anyone know this part of the circuit well? My next move is likely to be replacing the 7400 to see if a different one somehow magics it in to operation.

Thanks
Chris


Albert Otten
 

Hi Chris,

The hold-off multivibrator "looks like it's working". Did you check the waveforms at TPs around U110? The waveform at TP109 should be the inverted version of that at TP108 as shown in graph 5. (Note that voltage levels at the probe tip are 10 times larger than shown in the graph.) But also verify that TP106 is already low when TP109 goes low. TP106 low helps to bias the base of Q546 in negative direction. What happens to Q546 base voltage when TP106 goes low?
The time constant associated with C114 and surrounding resistors is 0.2 us or so. Maybe you missed the short spikes at the base of Q546 for that reason?
Disclaimer: I concluded these things from the diagrams but didn't test it in my 7T11.

Albert

On Fri, Nov 30, 2018 at 08:53 PM, cmjones01 wrote:


Is there anyone here intimately familiar with the internal workings of the
7T11 plugin? I ask because I'm trying to get one to work. The symptoms are
that there's no trace visible on the screen.

I had it on an extender in the bench today and established that most of it is
working: the trigger circuitry, strobe pulse generator and slow scan all seem
to be doing what they should. However:
- the trace is blanked because the X/sweep output is out of range (it's at
about +15V)
- it seems to be out of range because the horizontal memory isn't getting any
strobe pulses
- the holdoff multivibrator, which generates said pulses, looks like it's
working
- there's a peculiar arrangement of resistors, a diode and a 100pF capacitor
which takes the outputs of the multivibrator (a 7400 chip) and is supposed to
switch on a PNP transistor (Q546 I think) which has its emitter tied to ground
(in equivalent time sampling mode). This is not happening. The transistor's
base never goes below ground so it never switches on.

Looking at the circuit diagram, I can't see how it's supposed to work. I don't
have the manual in front of me right now but does anyone know this part of the
circuit well? My next move is likely to be replacing the 7400 to see if a
different one somehow magics it in to operation.

Thanks
Chris


cmjones01
 

Albert,

Thank you for these ideas. I'll check the relationship between TP106 and TP109 when I'm back in the office on Monday. The waveform I see at the base of Q546 is odd: it's a sort of castellated shape with up bits and down bits, but never appears to go below ground. I can see what look like those 200ns spikes at the anode of CR116, but they're no longer in evidence at the base of Q546. I should look more closely for them, though, since they'll be clamped to -0.7V by Q546's B-E junction.

The big anomaly is that the service manual shows waveform 28 at the collector of Q546, with 3us-wide pulses heading from -15V to pretty close to 0V. I can't get anything like that. The best I have ever seen were similar pulses but only from -15V up to about -13V, clearly not enough to do anything useful!

Chris

On Fri, Nov 30, 2018 at 01:14 PM, Albert Otten wrote:
Hi Chris,

The hold-off multivibrator "looks like it's working". Did you check the
waveforms at TPs around U110? The waveform at TP109 should be the inverted
version of that at TP108 as shown in graph 5. (Note that voltage levels at the
probe tip are 10 times larger than shown in the graph.) But also verify that
TP106 is already low when TP109 goes low. TP106 low helps to bias the base of
Q546 in negative direction. What happens to Q546 base voltage when TP106 goes
low?
The time constant associated with C114 and surrounding resistors is 0.2 us or
so. Maybe you missed the short spikes at the base of Q546 for that reason?
Disclaimer: I concluded these things from the diagrams but didn't test it in
my 7T11.

Albert

On Fri, Nov 30, 2018 at 08:53 PM, cmjones01 wrote:
I had it on an extender in the bench today and established that most of it
is
working: the trigger circuitry, strobe pulse generator and slow scan all
seem
to be doing what they should. However:
- the trace is blanked because the X/sweep output is out of range (it's at
about +15V)
- it seems to be out of range because the horizontal memory isn't getting
any
strobe pulses
- the holdoff multivibrator, which generates said pulses, looks like it's
working
- there's a peculiar arrangement of resistors, a diode and a 100pF capacitor
which takes the outputs of the multivibrator (a 7400 chip) and is supposed
to
switch on a PNP transistor (Q546 I think) which has its emitter tied to
ground
(in equivalent time sampling mode). This is not happening. The transistor's
base never goes below ground so it never switches on.

Looking at the circuit diagram, I can't see how it's supposed to work. I
don't
have the manual in front of me right now but does anyone know this part of
the
circuit well? My next move is likely to be replacing the 7400 to see if a
different one somehow magics it in to operation.


Albert Otten
 

Hi Chris,

Yesterday I looked only at graphs 4 and 5, but I think I read these in the wrong order (or interchanged the TPs). Hence the purpose of C114 is the opposite of what I thought, this cap speeds up reverse biasing the gate of Q546. Today I studied the circuit description in detail (again, after many years), pages 3-35 to 3-40, and verified waveforms in my 7T11A. Nice exercise.
The memory gate is open when Q546 conducts, so when its base is pulled negative. For this to happen, *both* gates B and D outputs have to be low. In rest D is low but B is high. Then when B also goes low the base of Q goes negative and the memory gate conducts. Some 2-3 us later D goes high, base Q goes positive and the memory gate closes. Again some us later (end of hold-off) both B and D switch to the opposite states.
In my 7T11A, low is nearly 0 V and high is 3.7 V (B) and 4.0 V (D). But for some reason B shows the same overshoot to 4.5 V as shown in graph 4.
With B high and D low base Q was +0.8 V. With B low and D high base Q was +1.0 V. With both B and D low the base was -0.8 V.
The timings were somewhat different from Fig. 3-31 and more consistent with those in graphs 4 an 5.
I triggered the 7T11A from an SG503 and varied the frequency, in order to easier recognize the rest period between end of hold-off and next trigger event.
Hope this helps.

Albert

On Fri, Nov 30, 2018 at 11:13 PM, cmjones01 wrote:


Albert,

Thank you for these ideas. I'll check the relationship between TP106 and TP109
when I'm back in the office on Monday. The waveform I see at the base of Q546
is odd: it's a sort of castellated shape with up bits and down bits, but never
appears to go below ground. I can see what look like those 200ns spikes at the
anode of CR116, but they're no longer in evidence at the base of Q546. I
should look more closely for them, though, since they'll be clamped to -0.7V
by Q546's B-E junction.

The big anomaly is that the service manual shows waveform 28 at the collector
of Q546, with 3us-wide pulses heading from -15V to pretty close to 0V. I can't
get anything like that. The best I have ever seen were similar pulses but only
from -15V up to about -13V, clearly not enough to do anything useful!

Chris


On Fri, Nov 30, 2018 at 01:14 PM, Albert Otten wrote:
Hi Chris,

The hold-off multivibrator "looks like it's working". Did you check the
waveforms at TPs around U110? The waveform at TP109 should be the inverted
version of that at TP108 as shown in graph 5. (Note that voltage levels at the
probe tip are 10 times larger than shown in the graph.) But also verify that
TP106 is already low when TP109 goes low. TP106 low helps to bias the base of
Q546 in negative direction. What happens to Q546 base voltage when TP106 goes low?
The time constant associated with C114 and surrounding resistors is 0.2 us or
so. Maybe you missed the short spikes at the base of Q546 for that reason?
Disclaimer: I concluded these things from the diagrams but didn't test it in
my 7T11.

Albert

On Fri, Nov 30, 2018 at 08:53 PM, cmjones01 wrote:
I had it on an extender in the bench today and established that most of it is
working: the trigger circuitry, strobe pulse generator and slow scan all seem
to be doing what they should. However:
- the trace is blanked because the X/sweep output is out of range (it's at about +15V)
- it seems to be out of range because the horizontal memory isn't getting any strobe pulses
- the holdoff multivibrator, which generates said pulses, looks like it's working
- there's a peculiar arrangement of resistors, a diode and a 100pF capacitor
which takes the outputs of the multivibrator (a 7400 chip) and is supposed to
switch on a PNP transistor (Q546 I think) which has its emitter tied to ground
(in equivalent time sampling mode). This is not happening. The transistor's
base never goes below ground so it never switches on.

Looking at the circuit diagram, I can't see how it's supposed to work. I don't
have the manual in front of me right now but does anyone know this part of the
circuit well? My next move is likely to be replacing the 7400 to see if a
different one somehow magics it in to operation.


cmjones01
 

Albert,

This is fantastic. Thank you so much. It's exactly the sort of thing I needed to know and it really gives me something to compare mine with. I also see the funny overshoot on output B - I think the resistor pulling it up to +15V has that effect, if I remember the circuit correctly.

Now I have a good idea what I'm looking for and will report back. I too am using a pulse generator (though a cheap-and-cheerful Thandar TG105) to trigger the 7T11.

Chris

On Sat, Dec 1, 2018 at 10:49 AM, Albert Otten wrote:


Hi Chris,

Yesterday I looked only at graphs 4 and 5, but I think I read these in the
wrong order (or interchanged the TPs). Hence the purpose of C114 is the
opposite of what I thought, this cap speeds up reverse biasing the gate of
Q546. Today I studied the circuit description in detail (again, after many
years), pages 3-35 to 3-40, and verified waveforms in my 7T11A. Nice exercise.
The memory gate is open when Q546 conducts, so when its base is pulled
negative. For this to happen, *both* gates B and D outputs have to be low. In
rest D is low but B is high. Then when B also goes low the base of Q goes
negative and the memory gate conducts. Some 2-3 us later D goes high, base Q
goes positive and the memory gate closes. Again some us later (end of
hold-off) both B and D switch to the opposite states.
In my 7T11A, low is nearly 0 V and high is 3.7 V (B) and 4.0 V (D). But for
some reason B shows the same overshoot to 4.5 V as shown in graph 4.
With B high and D low base Q was +0.8 V. With B low and D high base Q was +1.0
V. With both B and D low the base was -0.8 V.
The timings were somewhat different from Fig. 3-31 and more consistent with
those in graphs 4 an 5.
I triggered the 7T11A from an SG503 and varied the frequency, in order to
easier recognize the rest period between end of hold-off and next trigger
event.
Hope this helps.

Albert


On Fri, Nov 30, 2018 at 11:13 PM, cmjones01 wrote:


Albert,

Thank you for these ideas. I'll check the relationship between TP106 and
TP109
when I'm back in the office on Monday. The waveform I see at the base of
Q546
is odd: it's a sort of castellated shape with up bits and down bits, but
never
appears to go below ground. I can see what look like those 200ns spikes at
the
anode of CR116, but they're no longer in evidence at the base of Q546. I
should look more closely for them, though, since they'll be clamped to -0.7V
by Q546's B-E junction.

The big anomaly is that the service manual shows waveform 28 at the
collector
of Q546, with 3us-wide pulses heading from -15V to pretty close to 0V. I
can't
get anything like that. The best I have ever seen were similar pulses but
only
from -15V up to about -13V, clearly not enough to do anything useful!

Chris


On Fri, Nov 30, 2018 at 01:14 PM, Albert Otten wrote:
Hi Chris,

The hold-off multivibrator "looks like it's working". Did you check the
waveforms at TPs around U110? The waveform at TP109 should be the inverted
version of that at TP108 as shown in graph 5. (Note that voltage levels at
the
probe tip are 10 times larger than shown in the graph.) But also verify
that
TP106 is already low when TP109 goes low. TP106 low helps to bias the base
of
Q546 in negative direction. What happens to Q546 base voltage when TP106
goes low?
The time constant associated with C114 and surrounding resistors is 0.2 us
or
so. Maybe you missed the short spikes at the base of Q546 for that reason?
Disclaimer: I concluded these things from the diagrams but didn't test it
in
my 7T11.

Albert

On Fri, Nov 30, 2018 at 08:53 PM, cmjones01 wrote:
I had it on an extender in the bench today and established that most of
it is
working: the trigger circuitry, strobe pulse generator and slow scan all
seem
to be doing what they should. However:
- the trace is blanked because the X/sweep output is out of range (it's
at about +15V)
- it seems to be out of range because the horizontal memory isn't
getting any strobe pulses
- the holdoff multivibrator, which generates said pulses, looks like
it's working
- there's a peculiar arrangement of resistors, a diode and a 100pF
capacitor
which takes the outputs of the multivibrator (a 7400 chip) and is
supposed to
switch on a PNP transistor (Q546 I think) which has its emitter tied to
ground
(in equivalent time sampling mode). This is not happening. The
transistor's
base never goes below ground so it never switches on.

Looking at the circuit diagram, I can't see how it's supposed to work. I
don't
have the manual in front of me right now but does anyone know this part
of the
circuit well? My next move is likely to be replacing the 7400 to see if
a
different one somehow magics it in to operation.


cmjones01
 

On Sat, Dec 1, 2018 at 10:49 AM, Albert Otten wrote:
The memory gate is open when Q546 conducts, so when its base is pulled
negative. For this to happen, *both* gates B and D outputs have to be low. In
rest D is low but B is high. Then when B also goes low the base of Q goes
negative and the memory gate conducts. Some 2-3 us later D goes high, base Q
goes positive and the memory gate closes. Again some us later (end of
hold-off) both B and D switch to the opposite states.
I've now had another look at this and found at least one fault with my plugin. I see the multivibrator behaving as you describe, and outputs B and D do indeed spend a couple of microseconds low together. The reason that Q546 was never turning on was that R542 and R541 have drifted in value. R542 should be 30k but measures 26.2, and R541 should be 27k but measures 28.5. That meant that Q542 was stuck on and thus clamping the base drive to Q546. I've desoldered R541 and R542 for now.

Now I have healthy gate pulses at the collector of Q546. However, I don't have any sweep at TP555 in equivalent time mode. Need to look in to this next.

Chris


Albert Otten
 

Hi Chris,

It's not so easy to interpret the graphs 26 and 27 the correct way. The shaded area in 26 actually consists of horizontal lines, the value of which are sampled when the memory gate conducts. In 27 something similar for the rest period between end of hold-off and next trigger event.
When probe TP555 at a slow sweep speed, 5 ms/div, trigger from this signal and play with the SCAN rate of the 7T11, you can get a more or less stable display. You will recognize the 7T11 sweep as a rising envelope with an amplitude of 5 V. The output at TP564 is a nice 2X inverted, down going sweep signal with amplitude 10 V. TP556 shows about the same as TP564 but with small amplitude.

Albert

On Mon, Dec 3, 2018 at 01:06 PM, cmjones01 wrote:


Now I have healthy gate pulses at the collector of Q546. However, I don't have
any sweep at TP555 in equivalent time mode. Need to look in to this next.

Chris


Albert Otten
 

Hi Chris,

One more suggestion. Set the 7T11 for MANUAL scan. Set the test scope at 5 or 10 us/div, trigger from the 7T11 pulse output. Then probe all TPs 512, 515, 525, 555 of the horizontal amplifier. In each period you will see one horizontal line at a level determined by the SCAN knob setting. The level increases from ground to some maximum when you rotate SCAN from left (ccw) to right (cw). The amounts and ratios depend on the setting of the 7T11 time/div within its range. At TP555 the maximum should always be 5 V.

Albert

On Mon, Dec 3, 2018 at 05:13 PM, Albert Otten wrote:


Hi Chris,

It's not so easy to interpret the graphs 26 and 27 the correct way. The shaded
area in 26 actually consists of horizontal lines, the value of which are
sampled when the memory gate conducts. In 27 something similar for the rest
period between end of hold-off and next trigger event.
When probe TP555 at a slow sweep speed, 5 ms/div, trigger from this signal and
play with the SCAN rate of the 7T11, you can get a more or less stable
display. You will recognize the 7T11 sweep as a rising envelope with an
amplitude of 5 V. The output at TP564 is a nice 2X inverted, down going sweep
signal with amplitude 10 V. TP556 shows about the same as TP564 but with small
amplitude.

Albert

On Mon, Dec 3, 2018 at 01:06 PM, cmjones01 wrote:


Now I have healthy gate pulses at the collector of Q546. However, I don't have
any sweep at TP555 in equivalent time mode. Need to look in to this next.

Chris


cmjones01
 

I have now replaced the faulty R541 and R542. I also found that U572B, a Tek special op-amp, was dead, I think due to a careless slip of my scope probe. For now I've kludged a 741 in its place. Joy of joys, I have a horizontal sweep in equivalent-time mode! Still blanked in real-time mode. I think there's still something not right about the horizontal memory, which seems to only be used in real-time mode (or am I wrong?). Anyhow, when I select a real-time-mode sweep range, the output of U572B wanders up to about +12V and stays there. I do have real-time sweep on the way in to the horizontal memory, though. Will investigate further.

Chris

On Wed, Dec 5, 2018 at 06:07 AM, Albert Otten wrote:
One more suggestion. Set the 7T11 for MANUAL scan. Set the test scope at 5 or
10 us/div, trigger from the 7T11 pulse output. Then probe all TPs 512, 515,
525, 555 of the horizontal amplifier. In each period you will see one
horizontal line at a level determined by the SCAN knob setting. The level
increases from ground to some maximum when you rotate SCAN from left (ccw) to
right (cw). The amounts and ratios depend on the setting of the 7T11 time/div
within its range. At TP555 the maximum should always be 5 V.


Albert Otten
 

Hi Chris,

I might have good news for you.
The real time multivibrator doesn't work when you don't have a 7S11 adjacent to (or otherwise connected to) the 7T11. So then the memory gate is always blocked and you have no sweep output at the front panel.
You mean U512 I suppose. In real time ranges the horizontal amplifier now has one continuous ramp as input during the whole sweep of the 7T11. It depends on the time/div setting which outputs of U512 are really used in the amplifier chain. Not used outputs can be overdriven within short time without any consequence, and stay at some very high maximum level during the rest of the sweep . It's what I see happening.
I didn't pay attention to this in my previous post about equivalent time sampling but you essentially have the same situation then.

Albert

On Wed, Dec 5, 2018 at 03:23 PM, cmjones01 wrote:


I have now replaced the faulty R541 and R542. I also found that U572B, a Tek
special op-amp, was dead, I think due to a careless slip of my scope probe.
For now I've kludged a 741 in its place. Joy of joys, I have a horizontal
sweep in equivalent-time mode! Still blanked in real-time mode. I think
there's still something not right about the horizontal memory, which seems to
only be used in real-time mode (or am I wrong?). Anyhow, when I select a
real-time-mode sweep range, the output of U572B wanders up to about +12V and
stays there. I do have real-time sweep on the way in to the horizontal memory,
though. Will investigate further.

Chris

On Wed, Dec 5, 2018 at 06:07 AM, Albert Otten wrote:
One more suggestion. Set the 7T11 for MANUAL scan. Set the test scope at 5
or
10 us/div, trigger from the 7T11 pulse output. Then probe all TPs 512, 515,
525, 555 of the horizontal amplifier. In each period you will see one
horizontal line at a level determined by the SCAN knob setting. The level
increases from ground to some maximum when you rotate SCAN from left (ccw)
to
right (cw). The amounts and ratios depend on the setting of the 7T11
time/div
within its range. At TP555 the maximum should always be 5 V.


cmjones01
 

On Wed, Dec 5, 2018 at 08:17 AM, Albert Otten wrote:

I might have good news for you.
The real time multivibrator doesn't work when you don't have a 7S11 adjacent
to (or otherwise connected to) the 7T11. So then the memory gate is always
blocked and you have no sweep output at the front panel.
Aha! Your news is indeed good :-) I put the 7T11 in to a slot next to my 7S11 and I now have a horizontal sweep in the rea ltime ranges.

You mean U512 I suppose.
It was U512 when I killed it, but I swapped it with U572 to check that it was actually faulty. My 741 kludge seems to work but I'm not convinced it's perfect. The Tek chip seems specified quite fast (80MHz gain-bandwidth product!) and I think the rather pedestrian 741 may be running out of speed (or slew rate) in the faster real-time ranges.

In real time ranges the horizontal amplifier now has
one continuous ramp as input during the whole sweep of the 7T11. It depends on
the time/div setting which outputs of U512 are really used in the amplifier
chain. Not used outputs can be overdriven within short time without any
consequence, and stay at some very high maximum level during the rest of the
sweep . It's what I see happening.
Yes, I've observed that in the U512A/B/C/D chain too. The ramp only really makes sense properly after the last stage.

The good news is that, coupled with my 7S11, I now have something resembling a working sampling system. I have three rather tired-looking sampling heads in unknown condition (an S-1 and two S-2s, one labelled 50ps risetime and the other labelled 75ps). I have only tried the S-1 but it does actually work. Observations so far:
- equivalent time mode works a lot better than real time mode. At least, the waveform I see on the screen in equivalent-time mode looks like what I'd expect. I haven't tried a really fast edge in to the sampling head yet to see how the performance is at the really short time'div settings.
- in real time mode, it's hard to get a trace that makes sense. presumably the 50kHz-ish clock from the real time multivibrator limits the display to about one sample every 20us, so some of the faster sweep speeds don't show a lot. Even on the slower sweeps, the edges of my test squarewave exhibit terrible risetime (about half a division) which is puzzling, and very different to the behaviour in equivalent time. Maybe it's something to do with the recovery time of the sampling head, or there's something amiss with the 7S11. I tried adjusting the 'dot response' control and making sure smoothing was switched off, but it didn't improve much.

Thank you for your help so far. You can probably tell that I'm new to the 7000-series sampling world. I regularly use my CSA803A/SD24 sampling setup, but that makes it all so easy...

Chris


Jim Ford
 

Chris, I'm envious that you have a CSA803 and SD24!  I had that setup about 25 years ago at my employer the time.  Nice!
Have to make do with 7904, 7S12, 7S11, and two S-4's in my garage lab for now.  I use an S-53 or S-51 for triggering.  Works for what I need it to do.  Someday I'll have a CSA803 again.  And an S-6 for TDR, etc., etc.
Jim F


Sent from my Verizon, Samsung Galaxy smartphone

-------- Original message --------From: cmjones01 <chris@stumpie.com> Date: 12/5/18 1:19 PM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] 7T11 horizontal memory
On Wed, Dec  5, 2018 at 08:17 AM, Albert Otten wrote:

I might have good news for you.
The real time multivibrator doesn't work when you don't have a 7S11 adjacent
to (or otherwise connected to) the 7T11. So then the memory gate is always
blocked and you have no sweep output at the front panel.
Aha! Your news is indeed good :-) I put the 7T11 in to a slot next to my 7S11 and I now have a horizontal sweep in the rea ltime ranges.

You mean U512 I suppose.
It was U512 when I killed it, but I swapped it with U572 to check that it was actually faulty. My 741 kludge seems to work but I'm not convinced it's perfect. The Tek chip seems specified quite fast (80MHz gain-bandwidth product!) and I think the rather pedestrian 741 may be running out of speed (or slew rate) in the faster real-time ranges.

In real time ranges the horizontal amplifier now has
one continuous ramp as input during the whole sweep of the 7T11. It depends on
the time/div setting which outputs of U512 are really used in the amplifier
chain. Not used outputs can be overdriven within short time without any
consequence, and stay at some very high maximum level during the rest of the
sweep .  It's what I see happening.
Yes, I've observed that in the U512A/B/C/D chain too. The ramp only really makes sense properly after the last stage.

The good news is that, coupled with my 7S11, I now have something resembling a working sampling system. I have three rather tired-looking sampling heads in unknown condition (an S-1 and two S-2s, one labelled 50ps risetime and the other labelled 75ps). I have only tried the S-1 but it does actually work. Observations so far:
- equivalent time mode works a lot better than real time mode. At least, the waveform I see on the screen in equivalent-time mode looks like what I'd expect. I haven't tried a really fast edge in to the sampling head yet to see how the performance is at the really short time'div settings.
- in real time mode, it's hard to get a trace that makes sense. presumably the 50kHz-ish clock from the real time multivibrator limits the display to about one sample every 20us, so some of the faster sweep speeds don't show a lot. Even on the slower sweeps, the edges of my test squarewave exhibit terrible risetime (about half a division) which is puzzling, and very different to the behaviour in equivalent time. Maybe it's something to do with the recovery time of the sampling head, or there's something amiss with the 7S11. I tried adjusting the 'dot response' control and making sure smoothing was switched off, but it didn't improve much.

Thank you for your help so far. You can probably tell that I'm new to the 7000-series sampling world. I regularly use my CSA803A/SD24 sampling setup, but that makes it all so easy...

Chris


Albert Otten
 

On Wed, Dec 5, 2018 at 10:19 PM, cmjones01 wrote:


- in real time mode, it's hard to get a trace that makes sense. presumably the
50kHz-ish clock from the real time multivibrator limits the display to about
one sample every 20us, so some of the faster sweep speeds don't show a lot.
Even on the slower sweeps, the edges of my test squarewave exhibit terrible
risetime (about half a division) which is puzzling, and very different to the
behaviour in equivalent time. Maybe it's something to do with the recovery
time of the sampling head, or there's something amiss with the 7S11. I tried
adjusting the 'dot response' control and making sure smoothing was switched
off, but it didn't improve much.
Just to be sure that the fault is in the 7S11, you might check that the terrible rise time is also present when you use the 7S11 in combination with an ordinary timebase unit in stead of the 7T11.
You probably did not verify that you could adjust the dot response for loop gain one, it only "improved not much". I would check this first of all. The sampling head has an internal GAIN adjustment that could be too far off to correct with the 7S11 dot response adjustment.

You can probably tell that I'm new to the
7000-series sampling world. I regularly use my CSA803A/SD24 sampling setup,
but that makes it all so easy...
I followed the normal(?) route. I started with sampling units for the 564, then for the 7000 series and now I also have a CSA803A with SD-20 and SD-26. Still waiting for a good opportunity to buy an SD-24.

Albert


Jim Ford
 

Say, since I have the 7S12 and 7S11, is there any reason to obtain a 7T11 or 7T11A?  Aside from the fact that it's another cool piece of Tek gear to own, I mean.  IOW, do I have a justification before the finance committee (my wife ;) )?
TIA.
Jim F


Sent from my Verizon, Samsung Galaxy smartphone

-------- Original message --------From: Albert Otten <aodiversen@concepts.nl> Date: 12/6/18 2:56 AM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] 7T11 horizontal memory
On Wed, Dec  5, 2018 at 10:19 PM, cmjones01 wrote:


- in real time mode, it's hard to get a trace that makes sense. presumably the
50kHz-ish clock from the real time multivibrator limits the display to about
one sample every 20us, so some of the faster sweep speeds don't show a lot.
Even on the slower sweeps, the edges of my test squarewave exhibit terrible
risetime (about half a division) which is puzzling, and very different to the
behaviour in equivalent time. Maybe it's something to do with the recovery
time of the sampling head, or there's something amiss with the 7S11. I tried
adjusting the 'dot response' control and making sure smoothing was switched
off, but it didn't improve much.
Just to be sure that the fault is in the 7S11, you might check that the terrible rise time is also present when you use the 7S11 in combination with an ordinary timebase unit in stead of the 7T11.
You probably did not verify that you could adjust the dot response for loop gain one, it only "improved not much". I would check this first of all. The sampling head has an internal GAIN adjustment that could be too far off to correct with the 7S11 dot response adjustment.

You can probably tell that I'm new to the
7000-series sampling world. I regularly use my CSA803A/SD24 sampling setup,
but that makes it all so easy...
I followed the normal(?) route. I started with sampling units for the 564, then for the 7000 series and now I also have a CSA803A  with SD-20 and SD-26. Still waiting for a good opportunity to buy an SD-24.

Albert


cmjones01
 

On Thu, Dec 6, 2018 at 02:56 AM, Albert Otten wrote:
On Wed, Dec 5, 2018 at 10:19 PM, cmjones01 wrote:
- in real time mode, it's hard to get a trace that makes sense. presumably
the
50kHz-ish clock from the real time multivibrator limits the display to about
one sample every 20us, so some of the faster sweep speeds don't show a lot.
Even on the slower sweeps, the edges of my test squarewave exhibit terrible
risetime (about half a division) which is puzzling, and very different to
the
behaviour in equivalent time. Maybe it's something to do with the recovery
time of the sampling head, or there's something amiss with the 7S11. I tried
adjusting the 'dot response' control and making sure smoothing was switched
off, but it didn't improve much.
Just to be sure that the fault is in the 7S11, you might check that the
terrible rise time is also present when you use the 7S11 in combination with
an ordinary timebase unit in stead of the 7T11.
You probably did not verify that you could adjust the dot response for loop
gain one, it only "improved not much". I would check this first of all. The
sampling head has an internal GAIN adjustment that could be too far off to
correct with the 7S11 dot response adjustment.
I've been back in the office today and have done some more experiments. I have no known-good sampling heads, only one S-1 and two S-2s bought very cheap and in unknown condition.

The risetime thing is interesting. If I use the 7T11 in real-time mode with my S-1 head, the risetime is about 100us. However, in equivalent-time mode, I can quite easily see the 350ps risetime of the head and the waveform looks accurate.

With my least-broken S-2 head, the risetime in real-time mode is definitely faster than with the S-1. But the waveform as seem through the S-2 is definitely not accurate, though in equivalent-time mode the its risetime is fast (definitely better than 100ps).

I'll start a separate thread about sampling head fault symptoms.

Chris


cmjones01
 

On Thu, Dec 13, 2018 at 05:28 AM, cmjones01 wrote:
I've been back in the office today and have done some more experiments. I have
no known-good sampling heads, only one S-1 and two S-2s bought very cheap and
in unknown condition.
This 7T11 is the most maddenly unreliable plugin I've ever had! It seemed to be working well so I thought I'd tidy up my repairs and put it back together. But while doing so, I found I'd left a screw undone which allowed the trigger board to touch the timing board briefly. The scope shut down and restarted, and after that the 7T11 still worked but with an offset on the trace in equivalent-time mode which worsened as the time/div knob was turned clockwise, as if the slow ramp was sitting on a DC level. Random sampling mode stopped working too. While trying to debug that, I found that my extender cable wasn't plugged in properly so +15V was missing which led me a merry dance. Sorting that out left me with no trace at all!

I've found that Q184, which is critically important for the horizontal memory, is short-circuit all ways, hence the lack of trace. Now I'll find a replacement for it and see if I can find the original fault, or if it will have magically disappeared...

Chris


cmjones01
 

On Tue, Dec 18, 2018 at 03:48 AM, cmjones01 wrote:
On Thu, Dec 13, 2018 at 05:28 AM, cmjones01 wrote:
I've been back in the office today and have done some more experiments. I
have
no known-good sampling heads, only one S-1 and two S-2s bought very cheap
and
in unknown condition.
This 7T11 is the most maddenly unreliable plugin I've ever had! It seemed to
be working well so I thought I'd tidy up my repairs and put it back together.
But while doing so, I found I'd left a screw undone which allowed the trigger
board to touch the timing board briefly. The scope shut down and restarted,
and after that the 7T11 still worked but with an offset on the trace in
equivalent-time mode which worsened as the time/div knob was turned clockwise,
as if the slow ramp was sitting on a DC level. Random sampling mode stopped
working too. While trying to debug that, I found that my extender cable wasn't
plugged in properly so +15V was missing which led me a merry dance. Sorting
that out left me with no trace at all!

I've found that Q184, which is critically important for the horizontal memory,
is short-circuit all ways, hence the lack of trace. Now I'll find a
replacement for it and see if I can find the original fault, or if it will
have magically disappeared...
Curiosity got the better of me and I replaced Q184 with a BC558 found on a random scrap PCB. Now everything seems to work again the way it did before, which is a relief. I've put the covers back on before I do any more damage!

Chris


Albert Otten
 

Hi Chris,
Without quoting here all the previous stuff I like to visit these findings once more:


The risetime thing is interesting. If I use the 7T11 in real-time mode with my S-1 head, the risetime is about 100us.
However, in equivalent-time mode, I can quite easily see the 350ps risetime of the head and the waveform looks accurate.

With my least-broken S-2 head, the risetime in real-time mode is definitely faster than with the S-1. But the waveform
as seem through the S-2 is definitely not accurate, though in equivalent-time mode the its risetime is fast (definitely better than 100ps).
In my opinion it's impossible to study these fast rise times in real time mode. The 7S11 sampling moments are quite random compared to the 7T11 sweep start moments. Suppose you have the "true" rising edge located somewhere on the sweep. The fraction of sweeps which show a sample dot on the rising edge is very small. So you see dots mostly at the bottom and the top line, and seldom an intermediate dot. Also the last dot in a sweep before the start of the edge is sampled most often quite some time before the start of the edge. Similarly the first dot after the edge is mostly much farther to the right on the screen. So was my observation at 500 ps/div at the 3rd ET range and 200 ps rise time. If you look at "some distance" then you might have the impression that the rise time is rather long. If you look at how far the bottom line dots appear to the right and the top line dots to the left, then you notice just a small time gap between those two, more realistic for the true rise time.

Albert


Albert Otten
 

(continued) Ah.. I meant 3rd RT of course, not 3rd ET range. Albert


Albert Otten
 

Hi Chris,

Now I confirmed my remarks quoted below, using the 7T11A/7S11 combo in my 7854. Real time range .5 ms/div, speed 200 ps/div. In 7854 analog mode of course again I only see sparsely scattered dots. But after ACQ (acquire) I get a nice stored waveform, with correct edge and rise time, exactly as in the 7T11 equivalent time ranges. This works because ACQ continues acquiring until (almost) all horizontal positions in the stored waveform are filled. With 1024 points per waveform this took many seconds!
BTW I used the square wave output of an FG504. One feature of this function generator is that rise and fall times can be selected. I chose 200 ps and 150 kHz repetition rate (quite arbitrarily, higher might be better to reduce jitter). The 200 ps transitions are still nearly linear with sharp corners at top and bottom, in stead of S-shaped. Very nice.

Albert

On Tue, Dec 18, 2018 at 04:36 PM, Albert Otten wrote:


Hi Chris,
Without quoting here all the previous stuff I like to visit these findings
once more:


The risetime thing is interesting. If I use the 7T11 in real-time mode with
my S-1 head, the risetime is about 100us.
However, in equivalent-time mode, I can quite easily see the 350ps risetime
of the head and the waveform looks accurate.

With my least-broken S-2 head, the risetime in real-time mode is definitely
faster than with the S-1. But the waveform
as seem through the S-2 is definitely not accurate, though in
equivalent-time mode the its risetime is fast (definitely better than 100ps).
In my opinion it's impossible to study these fast rise times in real time
mode. The 7S11 sampling moments are quite random compared to the 7T11 sweep
start moments. Suppose you have the "true" rising edge located somewhere on
the sweep. The fraction of sweeps which show a sample dot on the rising edge
is very small. So you see dots mostly at the bottom and the top line, and
seldom an intermediate dot. Also the last dot in a sweep before the start of
the edge is sampled most often quite some time before the start of the edge.
Similarly the first dot after the edge is mostly much farther to the right on
the screen. So was my observation at 500 ps/div at the 3rd RT range and 200 ps
rise time. If you look at "some distance" then you might have the impression
that the rise time is rather long. If you look at how far the bottom line dots
appear to the right and the top line dots to the left, then you notice just a
small time gap between those two, more realistic for the true rise time.

Albert