Date   

Differential probes, ungrounded ringing, dual P6009s for high voltages

Robin Whittle
 

To make these two messages from the thread "Re: Isolation transformers
are not a panacea [was: Variak talk]" easier to find in the archives,
here is a summary of them, with links to the messages, under a new
Subject heading.

Steve <ditter2@yahoo.com> wrote an interesting piece on differential
probes, including:

http://tech.groups.yahoo.com/group/TekScopes/message/89737

1 - Why ordinary probes are resonant at certain frequencies if the
ground connection to the device under test is not connected.

2 - The value and potential limits of differential probes.

3 - Difficulties measuring power transistor drive voltages in the
few hundred millivolt range when they are both at 900V or more.

4 - Achieving differential measurements with a pair of equal length
cable P6009 probes. Apparently the DC gain can be finely adjusted
on these 100x attenuation 120MHz 1500V probes:

http://www.barrytech.com/tektronix/probes/tekp6009.html

David <davidwhess@gmail.com> continued the interesting discussion:

http://tech.groups.yahoo.com/group/TekScopes/message/89746

with mention of success using this dual probe technique, keeping the
cables close together, adding a ferrite core around the two cables,
modifying other probes to achieve the required close match of
attenuations, and: "As Glenn mentioned in the concurrent Cheap Probes
thread, if you run your hand along the probe cable and the waveform
changes, then you have problems."

- Robin http://www.firstpr.com.au/tequip/


Re: Interesting 7K Extenders on eBay

John Griessen
 

On 03/13/2013 07:04 AM, jerry massengale wrote:
Hi,

Really good soldering work.
If you mean the photos on Chris hayes's email, that's one I assembled. Thanks!

The edges of the edge connector need to be trimmed down a bit. As it is you cannot daisy chain the
extender as the edge is wider than the black connector can take.
That is true. I just found it out after the last order of PCBs for Tek_7K_flex.
I plan to fix it so they can daisy chain if possible while making a profit, but
for the first runs I was more worried about having minimal connector tolerance
slop so no wrong electrical connections could be made.

John Griessen
(back from Louisiana)


Re: Interesting 7K Extenders on eBay

John Griessen
 

On 03/13/2013 07:04 AM, jerry massengale wrote:
Hi,

Really good soldering work.
If you mean the photos on Dave D's original email, that's one I assembled. Thanks!

The edges of the edge connector need to be trimmed down a bit. As it is you cannot daisy chain the
extender as the edge is wider than the black connector can take.
That is true. I just found it out after the last order of PCBs for Tek_7K_flex.
I plan to fix it so they can daisy chain if possible while making a profit, but
for the first runs I was more worried about having minimal connector tolerance
slop so no wrong electrical connections could be made.

John Griessen
(back from Louisiana)


Re: Cheap Probes

Rob <rgwood@...>
 

Can we please call them “economically challenged probes”?

 

I would not like to think that we are trying to differentiate based on their origin in a 3rd world country or some such. After all it isn’t there fault where they were manufactured and it is certain they are trying to be the very best probes they can. Please, let’s just put the probes past behind us and just talk in terms of their capability regardless of economic background going forward. Ok?  


Re: Cheap Probes

 

On Wed, 13 Mar 2013 21:19:00 -0000, "Hakan H" <hahi@telia.com> wrote:

On Wed, 13 Mar 2013 14:45:39 +1100, Don Black
<donald_black@...> wrote:

I've never used them but the cheap probes are available up to several
hundred MHz (250-300?) for a bit more money. Has anyone tried them and
know how good or bad they are above 100 MHz. It may be more demanding on
fast pulses rather than sine waves.

Don Black.
--- In TekScopes@yahoogroups.com, David <davidwhess@...> wrote:

I bought a pair of inexpensive x10 250 MHz probes (Texas TX5125R) for
my 2440 and they work fine. I verified their performance with my
PG506 using a coaxial connection to the probe tip.
Almost two years ago, on behalf of another list member, I made a
comparison between a cheap 500 MHz probe (Texas TX6150) and a Tek
P6139A on a couple of TDS scopes.
The result can be found here: http://www.hakanh.com/dl/ProbeComp.htm
Fun stuff.

When I tested the Texas TX5125R probes (250 MHz) on my 2440 (300 MHz)
which is my fastest 1 MOhm input impedance oscilloscope, I used the
PG506 that I repaired and verified the fast transition outputs on my
S-4 sampling head at about 500ps with no significant overshoot. Since
then I have also gotten identical results with an S-1 sampling head.

The probes with a 50 ohm feedthrough termination and coaxial probe tip
connection gave results which were basically identical to using just a
50 ohm connection directly to the oscilloscope.

On one hand I do not currently have a faster pulse generator or
oscilloscope to test them but on the other hand, at 250 MHz the x10
capacitive probe loading is so high that high frequency content is
largely smushed so I would not see it anyway.


Re: 7B70 external input / magnifier problems

 

I took a quick look at the 7B70 schematics.

What you are describing with the variable control causing the external
X input to shift seems like a problem with the DC balance. Page 5-6
of the service manual has a procedure that involves adjusting R186
(Schematic 2 external horizontal amplifier balance) and then R37
(Schematic 1 external input amplifier balance) while rotating the
variable control until the X position does not change.

I would start there as it will likely provide further information
about the problem.

On Wed, 13 Mar 2013 19:38:21 -0000, "woooey" <woooey@hotmail.com>
wrote:

I found the manual on BAMA.

Relay K780 has an o/c coil. Replacing this with one borrowed
from a 7B71 fixed the 'magnifier' problem but had no effect
on the problem with external input. Still, this is encouraging,
so I will peer at the schematic some more.

--ian

--- In TekScopes@yahoogroups.com, "woooey" <woooey@...> wrote:

In 'amplifier' (external input) mode, the variable control
shifts the trace off to the left when turned clockwise. There
is some low-level signal on the X-axis but this seems to have
no relation to a signal applied to the external input.

The Magnifier no longer operates. The numeric indicator on
the display changes correctly, but the timebase itself doesn't
change.

Is it worth tapping on relays, etc, or should I just look for
a "new" 7B70?

Thanks,

--ian


Re: Isolation transformers are not a panacea [was: Variak talk]

 

On Wed, 13 Mar 2013 20:49:26 -0000, "Steve" <ditter2@yahoo.com> wrote:

--- In TekScopes@yahoogroups.com, Bert Haskins <bhaskins@...> wrote:


Two points that I would like to see discussed:
What do YOU do with the probe ground leads in the above two paragraphs?
I'm certainly not a newbe here ( have, use a 7A13 ), but I'm not an
expert either

Then do we really, really need need differential probes ( hard to
find and expensive ) in the above?
Yep, I know the CMRR is better....

The tek app note is OK but it really should have covered the grounding
issue to a greater depth.

Bert
The question of what to do with the ground leads often comes up. Experienced users have reported conflicting results from connecting them to a local (non-floating) ground – sometime improving the signal, other times it has no effect. I tried to model this a few years after I wrote the article to see if I could put some science into the conflicting results they reported. There is an explanation, but because a user will not know the impedances in their circuit, there is no hard and fast rule.

. . .

Since we usually do not know the relative impedances to ground of the common mode and differential mode components, we can forget the science and use an empirical method – try both and see which is better! Note that in many cases, you don't have a good local earth ground point in the circuit under test, so you can't connect the probe ground. In others, connecting the probe lead to the local earth ground creates a ground loop which you are trying to eliminate with the differential measurement!
This has been my experience. Under ideal conditions, modeling will
show what is going on but in practice the ground point impedance will
vary too much with frequency because of the physical limitations of
construction.

After taking extra precautions like forming the two probe cables to
run closely parallel for minimum loop area and maybe adding a common
mode suppression ferrite core around both cables, it is easiest to
just test with the differential probe grounds connected to "ground"
and then test with the probe grounds only connected to each other.

As Glenn mentioned in the concurrent Cheap Probes thread, if you run
your hand along the probe cable and the waveform changes, then you
have problems. This points to modeling the probe to circuit
connection as an RF balanced (one probe) to unbalanced (the circuit)
connection with an RF current on both sides of the probe cable shield
for three actual connections. Using two probes in a differential
connection should solve that problem just like using a balun but you
still have the center conductors of both probes plus a third and forth
connection comprising the inside and outside conductors of the two
shields which are now in parallel so if the ground is a poor RF
ground, you still get a current on both sides of the coaxial shields
which can affect the measurement results in some odd ways.

What about tying the two probe grounds together? Usually this will improve the high frequency response, but it may introduce a ground loop if the probe leads are not tightly twisted together.
I forget where but at least one of the Tektronix differential
amplifier manuals recommends this configuration for some measurements.
I usually lightly wire tie the cables of the two probes together so
they move as one.

Next Question - Differential probes in front of a 7A13? You are right - if you need the highest CMRR, then this is the way to go. You can calculate the rough amount of CMRR that you need if you know the common mode voltage and differential voltage you want to measure. When calculating CMRR needs, you need to determine how much common mode signal leaking into your measurement is acceptable. A good starting place is 10% or lower. Designers of motor drives and high power SWPS that have 700-900 V bus voltages need well matched probes when trying to measure upper gate drive. In this application you need to resolve a few hundred mV riding on a 900 V dynamic signal. Remember, the first order term that sets the maximum CMRR your measurement system will achieve is the relative gain (or attenuation) match between the + and – inputs.

A practical note – if you can find a pair of P6009 probes, they can be roughly matched for differential use. There is a fine adjustment on the DC attenuation in the comp box. I don't think this was intended for differential matching, but rather just to get the attenuation with spec. It was difficult to do with the divide by one hundred attenuation these probes have. The tip capacitor and resistor have minimal voltage coefficient, which more modern differential probes tend to have, so they maintain their match well over a wide common mode range. If yo go this route, be sure to get a pair with the same length of cable. I think Tek offered 3 versions – 3.5, 6 and 12 feet.
I originally suspected that the DC trim in the P6009 was used to
adjust the probe attenuation factor closer to the ideal value so that
it could support an extended temperature range and still be within its
specifications but the similar P6008 (x10) lacks the DC trim.

At some point I anticipate either modifying a pair of P6008 probes to
add a DC trim or picking up a pair of P6055 or P6135A probes.


Re: OT - but help needed.

Kenneth G. Gordon <kgordon2006@...>
 

On 13 Mar 2013 at 16:05, Bob Albert wrote:

Is the 3020 close enough?

http://bama.edebris.com/manuals/b&k/3020/
Nope. I have already checked that one out. It isn't very close. Double the number of controls,
and less than 1/2 the frequency range, not the same inputs and outputs.

Thanks, though.

Ken


Re: OT - but help needed.

Bob Albert
 

Is the 3020 close enough?

http://bama.edebris.com/manuals/b&k/3020/

Bob


--- On Wed, 3/13/13, Kenneth G. Gordon wrote:

From: Kenneth G. Gordon
Subject: Re: [TekScopes] OT - but help needed.
To: TekScopes@...
Date: Wednesday, March 13, 2013, 3:35 PM

 

Might anyone have a schematic, or even a users or maintenance manual for the B&K Precision model 3025 SweepFunction Generator?

Mine crapped out a few years ago, and I would really like to get it working.

B&K has so far been no help whatever.

Ken W7EKB


Re: OT - but help needed.

Kenneth G. Gordon <kgordon2006@...>
 

Might anyone have a schematic, or even a users or maintenance manual for the B&K Precision model 3025 SweepFunction Generator?

Mine crapped out a few years ago, and I would really like to get it working.

B&K has so far been no help whatever.

Ken W7EKB


Re: Fw: Is this Tektronix 577 CRT kaput?

ehrico
 

Well I am going a bid for it now, fingers crossed.

--- In TekScopes@yahoogroups.com, HankC <hankc918@...> wrote:



From: HankC <hankc918@...>
To: "TekScopes@yahoogroups.com" <TekScopes@yahoogroups.com>
Sent: Tuesday, March 12, 2013 7:16 PM
Subject: Re: Is this Tektronix 577 CRT kaput?


I believe that display section is the same as the storage section of a 5103/D11.
I have one of those & mine adjusted to a uniform flood.
I'd be willing to bet the CRT is OK & can be adjusted per the manual


 
HankC, Boston
WA1HOS


Re: Cheap Probes

Peter Gottlieb <hpnpilot@...>
 

Of course, if your sweep source and scope are leveled!

Try connecting the sweep source to the (properly terminated!) scope input and run the sweep with nothing in between the two. If the envelope is level enough for you, you're good. But remember that the scope gives you a linear scale and you may want log.

On 3/13/2013 6:00 PM, Javier Alberola wrote:
Hi Hakan,

Are you measuring the frequency response of the probes with a sweep sine generator an an oscilloscope, viewing the envelope? Would it be a cheap method to obtain, for example, the frequency response of an amplifier? (without a network analyzer)


El 13/03/2013, a las 22:19, Hakan H escribi�:



--- In TekScopes@yahoogroups.com <mailto:TekScopes%40yahoogroups.com>, David <davidwhess@...> wrote:

I bought a pair of inexpensive x10 250 MHz probes (Texas TX5125R) for
my 2440 and they work fine. I verified their performance with my
PG506 using a coaxial connection to the probe tip.

On Wed, 13 Mar 2013 14:45:39 +1100, Don Black
<donald_black@...> wrote:

I've never used them but the cheap probes are available up to several
hundred MHz (250-300?) for a bit more money. Has anyone tried them and
know how good or bad they are above 100 MHz. It may be more demanding on
fast pulses rather than sine waves.

Don Black.
----------------------
Almost two years ago, on behalf of another list member, I made a
comparison between a cheap 500 MHz probe (Texas TX6150) and a Tek
P6139A on a couple of TDS scopes.
The result can be found here: http://www.hakanh.com/dl/ProbeComp.htm
/H�kan
--------------------------------------------------------------------------------

No virus found in this message.
Checked by AVG - www.avg.com <http://www.avg.com>
Version: 10.0.1430 / Virus Database: 2641/5669 - Release Date: 03/13/13


Re: Cheap Probes

Javier Alberola
 

Hi Hakan,

Are you measuring the frequency response of the probes with a sweep sine generator an an oscilloscope, viewing the envelope? Would it be a cheap method to obtain, for example, the frequency response of an amplifier? (without a network analyzer)


El 13/03/2013, a las 22:19, Hakan H escribió:

 



--- In TekScopes@..., David wrote:
>
> I bought a pair of inexpensive x10 250 MHz probes (Texas TX5125R) for
> my 2440 and they work fine. I verified their performance with my
> PG506 using a coaxial connection to the probe tip.
>
> On Wed, 13 Mar 2013 14:45:39 +1100, Don Black
> wrote:
>
> >I've never used them but the cheap probes are available up to several
> >hundred MHz (250-300?) for a bit more money. Has anyone tried them and
> >know how good or bad they are above 100 MHz. It may be more demanding on
> >fast pulses rather than sine waves.
> >
> >Don Black.
> >
> ----------------------

Almost two years ago, on behalf of another list member, I made a
comparison between a cheap 500 MHz probe (Texas TX6150) and a Tek
P6139A on a couple of TDS scopes.
The result can be found here: http://www.hakanh.com/dl/ProbeComp.htm
/Håkan



Re: Sony/Tektronix 318 manual with full schematic

Gala Dragos
 

So I've looked at the clock generator.

The main clock is divided bu U140 and some signals go through U144 U148 and U152 (which is a CMOS).

However there is still something that bugs me. Why don't I have the 20ns option in the menu, now that the instrument identifies itself properly?

I mean, even if the clock was not working, I should have all the timebase options available.

--- On Wed, 3/13/13, Gala Dragos wrote:

From: Gala Dragos
Subject: Re: [TekScopes] Re: Sony/Tektronix 318 manual with full schematic
To: TekScopes@...
Date: Wednesday, March 13, 2013, 9:57 PM

 

>> Do you suppose that your 318 is doing something like that?
Don't know, I don't own this for that long.

The story starts in January, when I've realized I need a logic analyzer, so I managed to find one close by, well in the same continent and fiscal area that is (Europe).

Got the instrument, but it came without the probes, which I have acquired from Jerry here on the forum.

When the instrument arrived I have checked that the menu displays what it should, including the 20ns, and that it actually trigger on to something, which it did.

During these tests I noticed that the fan blows a lot of dust, it looked like a diesel exhaust. So, naturally, I took it apart and cleaned the thing. Put it back together and, to my surprise, no 20ns clock was available.

I need that 20ns clock up and running as I have some apps that run at 40+ Mhz

>> That actually is a very good sign!
Good, but what to do next? It is still unexplained.

--- On Wed, 3/13/13, Chuck Harris wrote:

From: Chuck Harris
Subject: Re: [TekScopes] Re: Sony/Tektronix 318 manual with full schematic
To: TekScopes@...
Date: Wednesday, March 13, 2013, 9:26 PM

 

Hi Gala,

That actually is a very good sign!

I have used some logic analyzers where the glitch memory reduced
the storage speed by a bit... and others where the synchronous
storage mode was somewhat slower than the asynchronous storage
mode.

... And, I have seen other units where there were lower priced
siblings that were made specifically to use up the slow memory
boards (that otherwise would have to be discarded). It used to
be a very common practice with minicomputers.

Do you suppose that your 318 is doing something like that?

It's always better when it turns out to be operator error.

-Chuck

Gala Dragos wrote:
> So situation is now like this:- instrument type detection works, by adding or
> removing w118 from A04 board (acquisition)- the 20ns clock is still not available
> from the menu What the heck is going on? --- On Wed, 3/13/13, Gala Dragos
> <gala_dragos@...> wrote:
>
> From: Gala Dragos <gala_dragos@...> Subject: Re: [TekScopes] Re:
> Sony/Tektronix 318 manual with full schematic To: TekScopes@... Date:
> Wednesday, March 13, 2013, 11:28 AM
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Well that 20ns clock is missing in the diagnostics menu as well. TYPE ID is only
> tested at startup, I've just checked that by soldering a wire across W118 and
> removing it (with a switch) when I entered the trigger menu. When W118 is jumpered
> at startup the instrument is a 338 until next startup (or MPU reset). The 20ns
> clock is no where to be seen no matter the if W118 is soldered or not.
>
> --- On Wed, 3/13/13, sbirdasn <sbirdasn@...> wrote:
>
> From: sbirdasn <sbirdasn@...> Subject: [TekScopes] Re: Sony/Tektronix 318
> manual with full schematic To: TekScopes@... Date: Wednesday, March
> 13, 2013, 6:32 AM
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Comments inline...
>
>
>
> --- In TekScopes@..., Gala Dragos wrote:
>
>>
>
>> I have that manual, I cannot find half of A02 board.
>
>
>
> You're right! page <4> is missing! The doc looked pretty good to me when I first
> looked at it. ;)
>
>
>
> That being said, my dead-tree version that is truly complete has the following
> circuits:
>
>
>
> 1) External clock input buffer with its threshold comparator/delay circuit.
>
> 2) Internal/external clock select logic.
>
> 3) Some buffers for qualifiers.
>
> 4) Threshold buffer amplifiers for the pods.
>
> 5) The signal routing to get one pod connector's differential signals to the
> interconnect header for the A01's differential to ECL signal w/ glitch detection
> circuits. (one pod is handled on A01, one pod on A02)
>
>
>
> For the problems you're experiencing, it probably won't help much, if any.
>
>
>
>> The full symptoms are these:- 20ns clock disappeared;- there are 4 selectable
>> groups in the setup screen;- each group has 16 bits available for display;- only
>> 2 pods can be selected for input (pod A and pod B), the rest are unavailable;
>
>
>
> By that, I think you mean that groups 3 & 4 are disabled by default, and you can
> only enter signals A0-7 or B0-7 into groups.
>
>
>
> Correct?
>
>
>
>> - all tests pass, including acq and sram;- the pods are capturing external
>> signals properly (checked them with the available calibration output);
>
>
>
> This would imply that upon power-up, it *does* ID itself as a 318.
>
>
>
>> I have noticed that the instrument can be "used" without the acquisitionÂ
>> board, albeit you can only browse the menus.
>
>
>
> Not surprising, since much of the hardware is write only or limited in how the CPU
> can interact with it.
>
>
>
>> Excerpt from the manual:Â "The chip select latch (A04U114) is used to enable
>> each 8-bit pair of the acquisition memory and for identifying instrument type.
>> It is written by the MPU with the WRITE BS signal from the A03 ACQ Control
>> board."
>
>
>
>> I have checked that circuit for continuity of traces and they are all ok. On the
>> schematic there is a jumper wire called W118 which in the 338 is mounted and in
>> the 318 is not mounted, checked that as well and it is not mounted.
>
>
>
> Since you've checked the signal connections, it sounds like a hardware failure in
> one or more chips.
>
>
>
> Consider the following (I have no idea how they wrote the firmware, so I have to
> make some educated guesses):
>
>
>
> The Bank Select pin used for Type ID is "wire-OR'ed" with the ACQ/Glitch Memory
> output data bus, which has pull-up/termination resistors to bring the bus to a
> known inactive state.
>
>
>
> Since the 318 has the jumper removed, then when the ID bank bit is driven active,
> the signal to be read *should* be in the "inactive" state.
>
>
>
> If one of the ACQ/Glitch SRAM's were to drive this pin to a "active" state, then
> the bit will be incorrectly read (there is also some status bits that are selected
> by the 2-1 muxes, so something could be wrong there too).
>
>
>
> When would this happen?
>
>
>
> Apparently, not on power-up, as it knows to be a 318 for pod count and memory to
> test.
>
>
>
> But perhaps when you enter the Trigger menu and start moving the clock rate, then
> the firmware *might* check the hardware jumper state again, read the wrong
> information, and prevent selecting the highest clock rate.
>
>
>
> This is just a guess, but it would be easy to check-
>
>
>
> Solder a jumper wire onto the W118 pad on the latch side for probing with a scope,
> and check for activity by the CPU to drive it active (to read the ID TYPE). The
> signal will be fairly slow, as it is driven active across several instructions,
> and thus will be in the micro-second range, unlike the sampling circuitry. Not an
> ideal situation for signal integrity, but then I doubt you have a pair of extender
> cards handy (made of unobtanium).
>
>
>
> I think the signals in this area are ECL, so the logic transition delta is about
> 0.8V between 1's and 0's, and does not go to either "ground" or V- (ECL is
> technically a -5.2V logic family).
>
>
>
> Explore the operation of the analyzer, and note when the signal goes active. Try
> various menus, field changes, etc. to see when the CPU fiddles with this signal.
> You might also look at when the Bank Select Latch is clocked too.
>
>
>
> You could check some other signals from the bank latch and the read side as sanity
> checks if necessary.
>
>
>
> I can't think of any other explanation as to why you can't select 20 nS clock.
>
>
>
> Good luck.
>
>
>
> Sbirdasn.
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>


Re: Cheap Probes

 

--- In TekScopes@yahoogroups.com, David <davidwhess@...> wrote:

I bought a pair of inexpensive x10 250 MHz probes (Texas TX5125R) for
my 2440 and they work fine. I verified their performance with my
PG506 using a coaxial connection to the probe tip.

On Wed, 13 Mar 2013 14:45:39 +1100, Don Black
<donald_black@...> wrote:

I've never used them but the cheap probes are available up to several
hundred MHz (250-300?) for a bit more money. Has anyone tried them and
know how good or bad they are above 100 MHz. It may be more demanding on
fast pulses rather than sine waves.

Don Black.
----------------------
Almost two years ago, on behalf of another list member, I made a
comparison between a cheap 500 MHz probe (Texas TX6150) and a Tek
P6139A on a couple of TDS scopes.
The result can be found here: http://www.hakanh.com/dl/ProbeComp.htm
/Håkan


Re: Isolation transformers are not a panacea [was: Variak talk]

ditter2
 

--- In TekScopes@yahoogroups.com, Bert Haskins <bhaskins@...> wrote:


Two points that I would like to see discussed:
What do YOU do with the probe ground leads in the above two paragraphs?
I'm certainly not a newbe here ( have, use a 7A13 ), but I'm not an
expert either

Then do we really, really need need differential probes ( hard to
find and expensive ) in the above?
Yep, I know the CMRR is better....

The tek app note is OK but it really should have covered the grounding
issue to a greater depth.

Bert

Bert,

I wrote this article a couple of decades ago, and have learned quite a bit more since then.

The question of what to do with the ground leads often comes up. Experienced users have reported conflicting results from connecting them to a local (non-floating) ground – sometime improving the signal, other times it has no effect. I tried to model this a few years after I wrote the article to see if I could put some science into the conflicting results they reported. There is an explanation, but because a user will not know the impedances in their circuit, there is no hard and fast rule.

Single ended probes can add a ring to a waveform when the ground lead is not connected. The tip capacitance has a return path through the ground (either the lead or the cable) which is inductive. This forms a series LC resonate circuit. At the resonate frequency, the input impedance of the probe would drop to zero. This is mitigated by series resistance, which spoils the Q factor – spreading the range of frequency that resonates at the same time increasing the minimum impedance it drops to.

The ring occurs whenever the resonate frequency range is within the frequency content of signal being measured, and within the passband of the scope. You lower the resonate frequency in a series LC circuit be either increasing the C and/or the L. The C stays constant, but the L really increases when you switch from a short probe tip ground lead to a long lead, or none at all (then the cable becomes the ground lead.

So how does this map to differential? Well, waveform distortion resulting from artifacts of resonance that is induced by the common mode will be removed by the differential amplifier, the signals are matched in both input paths. Therefore, the simple answer is that local ground is important when the differential mode impedance to scope ground is much lower than the common mode impedance to the same ground. If this is not the case, (as it is not in a truly balanced differential line), then you will not see an improvement in measured signal quality by locally grounding the probes. If the differential mode impedance to ground is similar or lower than the common mode, then connecting the probe tip to a local ground (not floating – must be earth ground) usually will improve the signal quality. My modeling exercise showed other factors as well, but these are second and third order effects and probably have no practical effects in the real world.)

Since we usually do not know the relative impedances to ground of the common mode and differential mode components, we can forget the science and use an empirical method – try both and see which is better! Note that in many cases, you don't have a good local earth ground point in the circuit under test, so you can't connect the probe ground. In others, connecting the probe lead to the local earth ground creates a ground loop which you are trying to eliminate with the differential measurement!

What about tying the two probe grounds together? Usually this will improve the high frequency response, but it may introduce a ground loop if the probe leads are not tightly twisted together.

Next Question - Differential probes in front of a 7A13? You are right - if you need the highest CMRR, then this is the way to go. You can calculate the rough amount of CMRR that you need if you know the common mode voltage and differential voltage you want to measure. When calculating CMRR needs, you need to determine how much common mode signal leaking into your measurement is acceptable. A good starting place is 10% or lower. Designers of motor drives and high power SWPS that have 700-900 V bus voltages need well matched probes when trying to measure upper gate drive. In this application you need to resolve a few hundred mV riding on a 900 V dynamic signal. Remember, the first order term that sets the maximum CMRR your measurement system will achieve is the relative gain (or attenuation) match between the + and – inputs.

A practical note – if you can find a pair of P6009 probes, they can be roughly matched for differential use. There is a fine adjustment on the DC attenuation in the comp box. I don't think this was intended for differential matching, but rather just to get the attenuation with spec. It was difficult to do with the divide by one hundred attenuation these probes have. The tip capacitor and resistor have minimal voltage coefficient, which more modern differential probes tend to have, so they maintain their match well over a wide common mode range. If yo go this route, be sure to get a pair with the same length of cable. I think Tek offered 3 versions – 3.5, 6 and 12 feet.

- Steve


Re: Sony/Tektronix 318 manual with full schematic

Gala Dragos
 

>> Do you suppose that your 318 is doing something like that?
Don't know, I don't own this for that long.

The story starts in January, when I've realized I need a logic analyzer, so I managed to find one close by, well in the same continent and fiscal area that is (Europe).

Got the instrument, but it came without the probes, which I have acquired from Jerry here on the forum.

When the instrument arrived I have checked that the menu displays what it should, including the 20ns, and that it actually trigger on to something, which it did.

During these tests I noticed that the fan blows a lot of dust, it looked like a diesel exhaust. So, naturally, I took it apart and cleaned the thing. Put it back together and, to my surprise, no 20ns clock was available.

I need that 20ns clock up and running as I have some apps that run at 40+ Mhz

>> That actually is a very good sign!
Good, but what to do next? It is still unexplained.

--- On Wed, 3/13/13, Chuck Harris wrote:

From: Chuck Harris
Subject: Re: [TekScopes] Re: Sony/Tektronix 318 manual with full schematic
To: TekScopes@...
Date: Wednesday, March 13, 2013, 9:26 PM

 

Hi Gala,

That actually is a very good sign!

I have used some logic analyzers where the glitch memory reduced
the storage speed by a bit... and others where the synchronous
storage mode was somewhat slower than the asynchronous storage
mode.

... And, I have seen other units where there were lower priced
siblings that were made specifically to use up the slow memory
boards (that otherwise would have to be discarded). It used to
be a very common practice with minicomputers.

Do you suppose that your 318 is doing something like that?

It's always better when it turns out to be operator error.

-Chuck

Gala Dragos wrote:
> So situation is now like this:- instrument type detection works, by adding or
> removing w118 from A04 board (acquisition)- the 20ns clock is still not available
> from the menu What the heck is going on? --- On Wed, 3/13/13, Gala Dragos
> <gala_dragos@...> wrote:
>
> From: Gala Dragos <gala_dragos@...> Subject: Re: [TekScopes] Re:
> Sony/Tektronix 318 manual with full schematic To: TekScopes@... Date:
> Wednesday, March 13, 2013, 11:28 AM
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Well that 20ns clock is missing in the diagnostics menu as well. TYPE ID is only
> tested at startup, I've just checked that by soldering a wire across W118 and
> removing it (with a switch) when I entered the trigger menu. When W118 is jumpered
> at startup the instrument is a 338 until next startup (or MPU reset). The 20ns
> clock is no where to be seen no matter the if W118 is soldered or not.
>
> --- On Wed, 3/13/13, sbirdasn <sbirdasn@...> wrote:
>
> From: sbirdasn <sbirdasn@...> Subject: [TekScopes] Re: Sony/Tektronix 318
> manual with full schematic To: TekScopes@... Date: Wednesday, March
> 13, 2013, 6:32 AM
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Comments inline...
>
>
>
> --- In TekScopes@..., Gala Dragos <gala_dragos@...> wrote:
>
>>
>
>> I have that manual, I cannot find half of A02 board.
>
>
>
> You're right! page <4> is missing! The doc looked pretty good to me when I first
> looked at it. ;)
>
>
>
> That being said, my dead-tree version that is truly complete has the following
> circuits:
>
>
>
> 1) External clock input buffer with its threshold comparator/delay circuit.
>
> 2) Internal/external clock select logic.
>
> 3) Some buffers for qualifiers.
>
> 4) Threshold buffer amplifiers for the pods.
>
> 5) The signal routing to get one pod connector's differential signals to the
> interconnect header for the A01's differential to ECL signal w/ glitch detection
> circuits. (one pod is handled on A01, one pod on A02)
>
>
>
> For the problems you're experiencing, it probably won't help much, if any.
>
>
>
>> The full symptoms are these:- 20ns clock disappeared;- there are 4 selectable
>> groups in the setup screen;- each group has 16 bits available for display;- only
>> 2 pods can be selected for input (pod A and pod B), the rest are unavailable;
>
>
>
> By that, I think you mean that groups 3 & 4 are disabled by default, and you can
> only enter signals A0-7 or B0-7 into groups.
>
>
>
> Correct?
>
>
>
>> - all tests pass, including acq and sram;- the pods are capturing external
>> signals properly (checked them with the available calibration output);
>
>
>
> This would imply that upon power-up, it *does* ID itself as a 318.
>
>
>
>> I have noticed that the instrument can be "used" without the acquisitionÂ
>> board, albeit you can only browse the menus.
>
>
>
> Not surprising, since much of the hardware is write only or limited in how the CPU
> can interact with it.
>
>
>
>> Excerpt from the manual:Â "The chip select latch (A04U114) is used to enable
>> each 8-bit pair of the acquisition memory and for identifying instrument type.
>> It is written by the MPU with the WRITE BS signal from the A03 ACQ Control
>> board."
>
>
>
>> I have checked that circuit for continuity of traces and they are all ok. On the
>> schematic there is a jumper wire called W118 which in the 338 is mounted and in
>> the 318 is not mounted, checked that as well and it is not mounted.
>
>
>
> Since you've checked the signal connections, it sounds like a hardware failure in
> one or more chips.
>
>
>
> Consider the following (I have no idea how they wrote the firmware, so I have to
> make some educated guesses):
>
>
>
> The Bank Select pin used for Type ID is "wire-OR'ed" with the ACQ/Glitch Memory
> output data bus, which has pull-up/termination resistors to bring the bus to a
> known inactive state.
>
>
>
> Since the 318 has the jumper removed, then when the ID bank bit is driven active,
> the signal to be read *should* be in the "inactive" state.
>
>
>
> If one of the ACQ/Glitch SRAM's were to drive this pin to a "active" state, then
> the bit will be incorrectly read (there is also some status bits that are selected
> by the 2-1 muxes, so something could be wrong there too).
>
>
>
> When would this happen?
>
>
>
> Apparently, not on power-up, as it knows to be a 318 for pod count and memory to
> test.
>
>
>
> But perhaps when you enter the Trigger menu and start moving the clock rate, then
> the firmware *might* check the hardware jumper state again, read the wrong
> information, and prevent selecting the highest clock rate.
>
>
>
> This is just a guess, but it would be easy to check-
>
>
>
> Solder a jumper wire onto the W118 pad on the latch side for probing with a scope,
> and check for activity by the CPU to drive it active (to read the ID TYPE). The
> signal will be fairly slow, as it is driven active across several instructions,
> and thus will be in the micro-second range, unlike the sampling circuitry. Not an
> ideal situation for signal integrity, but then I doubt you have a pair of extender
> cards handy (made of unobtanium).
>
>
>
> I think the signals in this area are ECL, so the logic transition delta is about
> 0.8V between 1's and 0's, and does not go to either "ground" or V- (ECL is
> technically a -5.2V logic family).
>
>
>
> Explore the operation of the analyzer, and note when the signal goes active. Try
> various menus, field changes, etc. to see when the CPU fiddles with this signal.
> You might also look at when the Bank Select Latch is clocked too.
>
>
>
> You could check some other signals from the bank latch and the read side as sanity
> checks if necessary.
>
>
>
> I can't think of any other explanation as to why you can't select 20 nS clock.
>
>
>
> Good luck.
>
>
>
> Sbirdasn.
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>


Re: 7B70 external input / magnifier problems

woooey
 

I found the manual on BAMA.

Relay K780 has an o/c coil. Replacing this with one borrowed
from a 7B71 fixed the 'magnifier' problem but had no effect
on the problem with external input. Still, this is encouraging,
so I will peer at the schematic some more.

--ian

--- In TekScopes@yahoogroups.com, "woooey" <woooey@...> wrote:

In 'amplifier' (external input) mode, the variable control
shifts the trace off to the left when turned clockwise. There
is some low-level signal on the X-axis but this seems to have
no relation to a signal applied to the external input.

The Magnifier no longer operates. The numeric indicator on
the display changes correctly, but the timebase itself doesn't
change.

Is it worth tapping on relays, etc, or should I just look for
a "new" 7B70?

Thanks,

--ian


Re: Sony/Tektronix 318 manual with full schematic

Chuck Harris <cfharris@...>
 

Hi Gala,

That actually is a very good sign!

I have used some logic analyzers where the glitch memory reduced
the storage speed by a bit... and others where the synchronous
storage mode was somewhat slower than the asynchronous storage
mode.

... And, I have seen other units where there were lower priced
siblings that were made specifically to use up the slow memory
boards (that otherwise would have to be discarded). It used to
be a very common practice with minicomputers.

Do you suppose that your 318 is doing something like that?

It's always better when it turns out to be operator error.

-Chuck

Gala Dragos wrote:

So situation is now like this:- instrument type detection works, by adding or
removing w118 from A04 board (acquisition)- the 20ns clock is still not available
from the menu What the heck is going on? --- On Wed, 3/13/13, Gala Dragos
<gala_dragos@yahoo.com> wrote:

From: Gala Dragos <gala_dragos@yahoo.com> Subject: Re: [TekScopes] Re:
Sony/Tektronix 318 manual with full schematic To: TekScopes@yahoogroups.com Date:
Wednesday, March 13, 2013, 11:28 AM


























Well that 20ns clock is missing in the diagnostics menu as well. TYPE ID is only
tested at startup, I've just checked that by soldering a wire across W118 and
removing it (with a switch) when I entered the trigger menu. When W118 is jumpered
at startup the instrument is a 338 until next startup (or MPU reset). The 20ns
clock is no where to be seen no matter the if W118 is soldered or not.

--- On Wed, 3/13/13, sbirdasn <sbirdasn@yahoo.com> wrote:

From: sbirdasn <sbirdasn@yahoo.com> Subject: [TekScopes] Re: Sony/Tektronix 318
manual with full schematic To: TekScopes@yahoogroups.com Date: Wednesday, March
13, 2013, 6:32 AM



























Comments inline...



--- In TekScopes@yahoogroups.com, Gala Dragos <gala_dragos@...> wrote:

I have that manual, I cannot find half of A02 board.


You're right! page <4> is missing! The doc looked pretty good to me when I first
looked at it. ;)



That being said, my dead-tree version that is truly complete has the following
circuits:



1) External clock input buffer with its threshold comparator/delay circuit.

2) Internal/external clock select logic.

3) Some buffers for qualifiers.

4) Threshold buffer amplifiers for the pods.

5) The signal routing to get one pod connector's differential signals to the
interconnect header for the A01's differential to ECL signal w/ glitch detection
circuits. (one pod is handled on A01, one pod on A02)



For the problems you're experiencing, it probably won't help much, if any.



The full symptoms are these:- 20ns clock disappeared;- there are 4 selectable
groups in the setup screen;- each group has 16 bits available for display;- only
2 pods can be selected for input (pod A and pod B), the rest are unavailable;


By that, I think you mean that groups 3 & 4 are disabled by default, and you can
only enter signals A0-7 or B0-7 into groups.



Correct?



- all tests pass, including acq and sram;- the pods are capturing external
signals properly (checked them with the available calibration output);


This would imply that upon power-up, it *does* ID itself as a 318.



I have noticed that the instrument can be "used" without the acquisitionÂ
board, albeit you can only browse the menus.


Not surprising, since much of the hardware is write only or limited in how the CPU
can interact with it.



Excerpt from the manual:Â "The chip select latch (A04U114) is used to enable
each 8-bit pair of the acquisition memory and for identifying instrument type.
It is written by the MPU with the WRITE BS signal from the A03 ACQ Control
board."


I have checked that circuit for continuity of traces and they are all ok. On the
schematic there is a jumper wire called W118 which in the 338 is mounted and in
the 318 is not mounted, checked that as well and it is not mounted.


Since you've checked the signal connections, it sounds like a hardware failure in
one or more chips.



Consider the following (I have no idea how they wrote the firmware, so I have to
make some educated guesses):



The Bank Select pin used for Type ID is "wire-OR'ed" with the ACQ/Glitch Memory
output data bus, which has pull-up/termination resistors to bring the bus to a
known inactive state.



Since the 318 has the jumper removed, then when the ID bank bit is driven active,
the signal to be read *should* be in the "inactive" state.



If one of the ACQ/Glitch SRAM's were to drive this pin to a "active" state, then
the bit will be incorrectly read (there is also some status bits that are selected
by the 2-1 muxes, so something could be wrong there too).



When would this happen?



Apparently, not on power-up, as it knows to be a 318 for pod count and memory to
test.



But perhaps when you enter the Trigger menu and start moving the clock rate, then
the firmware *might* check the hardware jumper state again, read the wrong
information, and prevent selecting the highest clock rate.



This is just a guess, but it would be easy to check-



Solder a jumper wire onto the W118 pad on the latch side for probing with a scope,
and check for activity by the CPU to drive it active (to read the ID TYPE). The
signal will be fairly slow, as it is driven active across several instructions,
and thus will be in the micro-second range, unlike the sampling circuitry. Not an
ideal situation for signal integrity, but then I doubt you have a pair of extender
cards handy (made of unobtanium).



I think the signals in this area are ECL, so the logic transition delta is about
0.8V between 1's and 0's, and does not go to either "ground" or V- (ECL is
technically a -5.2V logic family).



Explore the operation of the analyzer, and note when the signal goes active. Try
various menus, field changes, etc. to see when the CPU fiddles with this signal.
You might also look at when the Bank Select Latch is clocked too.



You could check some other signals from the bank latch and the read side as sanity
checks if necessary.



I can't think of any other explanation as to why you can't select 20 nS clock.



Good luck.



Sbirdasn.







































Re: Sony/Tektronix 318 manual with full schematic

Gala Dragos
 

So situation is now like this:
- instrument type detection works, by adding or removing w118 from A04 board (acquisition)
- the 20ns clock is still not available from the menu

What the heck is going on?

--- On Wed, 3/13/13, Gala Dragos wrote:

From: Gala Dragos
Subject: Re: [TekScopes] Re: Sony/Tektronix 318 manual with full schematic
To: TekScopes@...
Date: Wednesday, March 13, 2013, 11:28 AM

 

Well that 20ns clock is missing in the diagnostics menu as well.

TYPE ID is only tested at startup, I've just checked that by soldering a wire across W118 and removing it (with a switch) when I entered the trigger menu. When W118 is jumpered at startup the instrument is a 338 until next startup (or MPU reset).

The 20ns clock is no where to be seen no matter the if W118 is soldered or not.

--- On Wed, 3/13/13, sbirdasn wrote:

From: sbirdasn
Subject: [TekScopes] Re: Sony/Tektronix 318 manual with full schematic
To: TekScopes@...
Date: Wednesday, March 13, 2013, 6:32 AM

 



Comments inline...

--- In TekScopes@..., Gala Dragos wrote:
>
> I have that manual, I cannot find half of A02 board.

You're right! page <4> is missing! The doc looked pretty good to me when I first looked at it. ;)

That being said, my dead-tree version that is truly complete has the following circuits:

1) External clock input buffer with its threshold comparator/delay circuit.
2) Internal/external clock select logic.
3) Some buffers for qualifiers.
4) Threshold buffer amplifiers for the pods.
5) The signal routing to get one pod connector's differential signals to the interconnect header for the A01's differential to ECL signal w/ glitch detection circuits. (one pod is handled on A01, one pod on A02)

For the problems you're experiencing, it probably won't help much, if any.

> The full symptoms are these:- 20ns clock disappeared;- there are 4 selectable groups in the setup screen;- each group has 16 bits available for display;- only 2 pods can be selected for input (pod A and pod B), the rest are unavailable;

By that, I think you mean that groups 3 & 4 are disabled by default, and you can only enter signals A0-7 or B0-7 into groups.

Correct?

> - all tests pass, including acq and sram;- the pods are capturing external signals properly (checked them with the available calibration output);

This would imply that upon power-up, it *does* ID itself as a 318.

> I have noticed that the instrument can be "used" without the acquisition board, albeit you can only browse the menus.

Not surprising, since much of the hardware is write only or limited in how the CPU can interact with it.

> Excerpt from the manual: "The chip select latch (A04U114) is used to enable each 8-bit pair of the acquisition memory and for identifying instrument type. It is written by the MPU with the WRITE BS signal from the A03 ACQ Control board."

> I have checked that circuit for continuity of traces and they are all ok. On the schematic there is a jumper wire called W118 which in the 338 is mounted and in the 318 is not mounted, checked that as well and it is not mounted.

Since you've checked the signal connections, it sounds like a hardware failure in one or more chips.

Consider the following (I have no idea how they wrote the firmware, so I have to make some educated guesses):

The Bank Select pin used for Type ID is "wire-OR'ed" with the ACQ/Glitch Memory output data bus, which has pull-up/termination resistors to bring the bus to a known inactive state.

Since the 318 has the jumper removed, then when the ID bank bit is driven active, the signal to be read *should* be in the "inactive" state.

If one of the ACQ/Glitch SRAM's were to drive this pin to a "active" state, then the bit will be incorrectly read (there is also some status bits that are selected by the 2-1 muxes, so something could be wrong there too).

When would this happen?

Apparently, not on power-up, as it knows to be a 318 for pod count and memory to test.

But perhaps when you enter the Trigger menu and start moving the clock rate, then the firmware *might* check the hardware jumper state again, read the wrong information, and prevent selecting the highest clock rate.

This is just a guess, but it would be easy to check-

Solder a jumper wire onto the W118 pad on the latch side for probing with a scope, and check for activity by the CPU to drive it active (to read the ID TYPE). The signal will be fairly slow, as it is driven active across several instructions, and thus will be in the micro-second range, unlike the sampling circuitry. Not an ideal situation for signal integrity, but then I doubt you have a pair of extender cards handy (made of unobtanium).

I think the signals in this area are ECL, so the logic transition delta is about 0.8V between 1's and 0's, and does not go to either "ground" or V- (ECL is technically a -5.2V logic family).

Explore the operation of the analyzer, and note when the signal goes active. Try various menus, field changes, etc. to see when the CPU fiddles with this signal. You might also look at when the Bank Select Latch is clocked too.

You could check some other signals from the bank latch and the read side as sanity checks if necessary.

I can't think of any other explanation as to why you can't select 20 nS clock.

Good luck.

Sbirdasn.

94601 - 94620 of 183392