Topics

S-52 only works with lowered +15 V power supply

unclebanjoman
 

Recently I got an s-52 pulser along with two s-6s, one 7t11, one a 7s11 and one 7s12.
Previous owner said they showed no signs of life.

First, I started looking at the s-52 pulser.
Inserted in the 7S12 it showed no signs of life.

Not having a Tek 285, I made myself an adapter using a suitable connector and powering the s-52 through a dual bench power supply.
I adjusted the dual power supply to +/- 15 V, lowering the negative branch using a 7912 voltage regulator.
As expected, no output signal from the pulser.
As a test setup I used a 7A26 and a 7A80 in conjunction with a 7854 mainframe.
I checked all the waveforms indicated in the schematic diagram: all regular except that the clock oscillator has a period of 1us instead of 0.5us, but I remember reading about that, in the latest versions Tek modified in this manner.

The only missing waveform was the TD output.
At this point I removed the original TD assembly welding, at the points indicated in the manual, one 1N3130 instead.
Power on...Still no result.

Remembering my previous experience with one 284 pulser some months ago, without too much belief I began to vary the + 15V branch, increasing and decreasing it to see if something happened.
To my surprise, around + 13V I notice that an output waveform appears, although not synchronized.
I connect the TRIG OUT of the S-52 to the 7B80 and the output of the S-56 to the 7A26 both with 50 ohm termination.
I set the trigger, 2us / div and ... Wow! Perfect! The waveform is absolutely identical to the 14A waveform reported in the manual.

Next I tried to vary the positive voltage (the negative one is fixed by the LM7912) and I notice that the TD pulser out started working at around 12-12.5 V, and stopping at around 14V.
In this range, I noticed that the duration of the automatic bias ramp and the subsequent pedestal prebias (when the TD fires and generates the "real" optimal pulse) increase progressively.
Above 14V, the impulse following the bias ramp disappears first and then the main impulse disappears also.

Powering the pulser with -12.6 V and +13.5 V the pulse is perfectly stable, and connecting it to my Tek 564 equipped with 3S1 and 3T77A, I can display a vbery sharp pulse a risetime of 350 ps, as expected ​since it's the bandwidth limit of the 3S1.

I have not yet checked with the 7S11 and 7T11 pair because I want to solve the malfunction of the S-52 first.
I can assume that this behavior is due to some derated components, certainly the carbon composite resistors. I have not yet measured the resistors one by one.
I tried fiddling with the trigger level R90, powering the S-52 with its nominal voltages ( -12.2 and +15 V) with no result, at the output I can only see the pedestal pre-bias and the reset intervals nut no positive-going TD impulse at all.

The snap-off diode works perfectly and the delay time is 85 ns as stated in the manual.

I would like to know your opinions and advice on how to restore the S52 to its original operating conditions.
Manufacturing date is surely 1985/1986 as deduced from the dates marked on SN7493 and SN74145 ICs. This sampling head looks as new.

As a very rude solution I could insert a couple of diodes in series to the + 15V coming from the rear PCB edge connector, lowering the + 15 voltage, but it seems a rough solution to me, not to mention that it could also last after a while if the derating of some component continues .. .

Thank you all,
Max

unclebanjoman
 

In the meanwhile I've carefully checked the resistor values.
All seems O.K. with the exception of the two 2.7 ohm in parallel, R65 & R68.
I had the precaution to disconnect (unscrew) the tunnel diode (CR69) to avoid false measurements (and also not to damage it!).
The parallel of these two resistors give a measure of 2.8 ohm. Seems that one of the resistor is open...
I think this can be a cause of the abnormal behavior.

Moreover, I've discovered that the capacitor C65 (1.8 pF) was replaced with a series combination of 430 ohm resistor + a small SMD capacitor.
Seems that this modification is Tek made, since I notice no trace of re-soldering or manipulation.
Undoubtedly in 1985 SMD components already existed, although not of widespread usage, I think.
My service manual don't report anything about that. I purchased my copy from Dave of ArtekMedia.

What do you think about all that?

The two resistor are carbon composite. I can find them at Mouser. Or can I use good quality SMD resistors?

Thanks to you all,
Max

Dan G
 

On Tue, Jun 23, 2020 at 05:02 PM, unclebanjoman wrote:

Moreover, I've discovered that the capacitor C65 (1.8 pF) was replaced with a
series combination of 430 ohm resistor + a small SMD capacitor.
Seems that this modification is Tek made, since I notice no trace of
re-soldering or manipulation.
Undoubtedly in 1985 SMD components already existed, although not of
widespread usage, I think.
My service manual don't report anything about that. I purchased my copy from
Dave of ArtekMedia.

What do you think about all that?
Hi Max,

The C65/R64 combination was changed starting with serial number B080000.
The new C65 part number is 283-0320-00 (1pF, 50V, leadless).
The new R64 part number is 317-0431-00 (430 Ohm, 5%, 1/8W, BB4315)

This is from the May 1981 revision of the service manual.

I hope this helps,
dan

unclebanjoman
 

Thanks Dan,
I manually edited my schematic with your values.

In the meanwhile I'm searching for some 2.7 ohm 1/8 W carbon composite resistors. Very hard to find in my country!
I found them in GB, price: approx 2.5 GBP for five pieces. Shipping fee: 12.96 GBP. Ouch!
If I don't find them in Italy, I buy those. Sgrunt.

Max

Albert Otten
 

In the meanwhile I'm searching for some 2.7 ohm 1/8 W carbon composite
resistors. Very hard to find in my country!
Hi Max,

A few thoughts.
I don't see how one of these resistors being open can be the cause for no pulse at higher (+15V) supply voltage while there is no problem at lower supply voltage. In my opinion its much safer to leave the 2.7R resistors alone and search for the true cause. I would start at the U30 outputs at D, E and B and follow there effect on the comparator pairs Q72/Q74 and so on. A fault there can also result in a missing bias calibration pulse.

An increasing supply voltage also has effect on trigger level as if the wiper of R90 is turned towards the top terminal (base Q90). Then the trigger pulse from the snap-off diode can be insufficient to fire the TD. This can be compensated by turning the wiper towards C90 Has the wiper good contact all over its whole range?

Albert

Albert Otten
 

Hi Max,

Any progress? Today I used an S-52 on an extender, case removed, and then thought of your problem. I read your first post once more:
The only missing waveform was the TD output.
At this point I removed the original TD assembly welding, at the points indicated in the manual, one 1N3130 instead.
Power on...Still no result.
The 1N3130 is a 50mA/25pF TD, while the original one is 50mA/1.5pF. Hence I think the small cap C65 will not pass enough current from the snap-off diode to fire the TD in a stable way. The bias current would have to be unrealistic close to the firing current.
To my surprise, around + 13V I notice that an output waveform appears, although not synchronized.
I connect the TRIG OUT of the S-52 to the 7B80 and the output of the S-56 to the 7A26 both with 50 ohm termination.
I set the trigger, 2us / div and ... Wow! Perfect! The waveform is absolutely identical to the 14A waveform reported in the manual.
I doubt so. The proper output pulse probably doesn't start nearly immediately after the sweep start (period 8) but earlier, in period 7, so up to 1 us earlier. Simply because the arming bias current in period 7 is already high enough to fire the TD.
In my previous post I wrote:
An increasing supply voltage also has effect on trigger level as if the wiper of R90 is turned towards the
top terminal (base Q90). Then the trigger pulse from the snap-off diode can be insufficient to fire the TD.
This can be compensated by turning the wiper towards C90 Has the wiper good contact all over its whole range?
Here I overlooked the current delivered via R97+R98. That current will increase when you increase the "+15V" level. Are the values of these resistors still in spec (not replaced by smaller values)?
When you say that the pulse disappears, does that mean that the output stays low, or that the output is high over a so long period that you don't consider this as the proper pulse?

Albert

unclebanjoman
 

Hi Albert,

sorry for the late reply but during the week I work hardly and in the evening I am too tired to take care of my Teks and I want to avoid mistakes.

Anyway I took just now some photos of the pulser while operating, varying the positive supply between 13.5 V and 15.5 V, in 0.5 V intervals.
I've created a new photo album, you can see it at the following link:

https://groups.io/g/TekScopes/album?id=249667

NOTE: The TD is the original one, carefully reassembled .
The optimal voltage seems to be around 14.5V. The pulse are stable and clean.NOTE: The TD is the original one, carefully reassembled .
You can see, at 15V the impulse disappears and only the impulse of pre-bias with the ramp remains.

The pot R90 (trig level) is almost fully counterclockwise (it's in the original position when I bought the thing).
At 15V ,turning it completely ccw I do not get appreciable effects anyway

I think the resistors, one or more, changed their values,
In the weekend I will check carefully their values.
I can confirm that the couple of 2.7 ohm resistors in parallel give me a reading of 2.8 ohm and not 1.4 as it should be ...

Max

Albert Otten
 

Hi Max,

Those pictures helped me a lot to understand what you previously said in words!
Just for comparison I added an annotated picture to your album (and somewhat destroyed the order).

You said R90 was almost CCW. But that produces almost the lowest arming current. When there is no output pulse in period 8 I would turn R90 CW, not CCW.

It struck me that at higher "+15V" there is something wrong in the output level. Only during the reset periods 3-6 and 9 should the output be true zero. In the "unused" remaining periods there should be some positive level as shown in your pictures up to 14.5 V. That's due to current via R97+R98. I suspect a state fault in comparator Q86/Q88. The zero output voltage also appears in period 7 at higher "+15V", probably due to the same fault.

Albert

On Wed, Jul 1, 2020 at 11:21 PM, unclebanjoman wrote:


Hi Albert,

sorry for the late reply but during the week I work hardly and in the evening
I am too tired to take care of my Teks and I want to avoid mistakes.

Anyway I took just now some photos of the pulser while operating, varying the
positive supply between 13.5 V and 15.5 V, in 0.5 V intervals.
I've created a new photo album, you can see it at the following link:

https://groups.io/g/TekScopes/album?id=249667

NOTE: The TD is the original one, carefully reassembled .
The optimal voltage seems to be around 14.5V. The pulse are stable and
clean.NOTE: The TD is the original one, carefully reassembled .
You can see, at 15V the impulse disappears and only the impulse of pre-bias
with the ramp remains.

The pot R90 (trig level) is almost fully counterclockwise (it's in the
original position when I bought the thing).
At 15V ,turning it completely ccw I do not get appreciable effects anyway

I think the resistors, one or more, changed their values,
In the weekend I will check carefully their values.
I can confirm that the couple of 2.7 ohm resistors in parallel give me a
reading of 2.8 ohm and not 1.4 as it should be ...

Max

unclebanjoman
 

Hi Albert,
these are the effects of work fatigue when I get home....

Here is the correct phrase:

"The pot R90 (trig level) is almost fully CLOCKWISE (it's in the original position when I bought the thing).
At 15V ,turning it completely CLOCKWISE I do not get appreciable effects anyway"

So, the arming current is insufficient using +15V

Max

unclebanjoman
 

Some progress.
Just now (in Italy it's 9.40 p.m.) I carefully measured the parallel R65/R68 with my Fluke 8040A, with a reading of 3.3 ohms. No good.
I carefully soldered a 3.9 ohm resistor, the only value I have, in parallel with the other two. Not a carbon composition but a metal film one, for a quick test it's enough.
Result: with +15V and the trim R90 fully clockwise the pulse appears and seems good. At +15.1 V disappears again. Sgrunt.
I think there are some other resistors out of tolerance. I'll check the whole "thing" quietly tomorrow night with great patience.
If other resistor's values are needed I will make a single order, the only place near Italy where I can buy them (carbon composition, 5% 1/4 W) at decent price is in G.B....
Otherwise I can get SMD resistors but I don't know if they are suitable. The seller (in Italy) guarantees them up to 1GHz only.

Max

Albert Otten
 

Hi Max,

I think the extra 3.9 ohm helps a tiny bit (makes the snap-off current to the TD somewhat stronger) but in my opinion its not the problem. Since at 15.5 V I suspected a wrong state of Q86/Q88 where Q88 should conduct I measured the waveforms at the bases of this comparator pair. The bases switch role with about 1 V voltage difference in both states. Roughly speaking 4.5 V vs 5.5 V and vice versa. The nominal Off value for base Q86 is 5.56 V and the nominal On value for base Q88 is 4.23 V (ignoring base currents).
R86 is easy to access from the bottom side. Less easy was to clip the probe tip at R89.
Hope this helps.

Albert

On Thu, Jul 2, 2020 at 09:54 PM, unclebanjoman wrote:


Some progress.
Just now (in Italy it's 9.40 p.m.) I carefully measured the parallel R65/R68
with my Fluke 8040A, with a reading of 3.3 ohms. No good.
I carefully soldered a 3.9 ohm resistor, the only value I have, in parallel
with the other two. Not a carbon composition but a metal film one, for a quick
test it's enough.
Result: with +15V and the trim R90 fully clockwise the pulse appears and seems
good. At +15.1 V disappears again. Sgrunt.
I think there are some other resistors out of tolerance. I'll check the whole
"thing" quietly tomorrow night with great patience.
If other resistor's values are needed I will make a single order, the only
place near Italy where I can buy them (carbon composition, 5% 1/4 W) at decent
price is in G.B....
Otherwise I can get SMD resistors but I don't know if they are suitable. The
seller (in Italy) guarantees them up to 1GHz only.

Max

unclebanjoman
 

Work still in progress.
Yesterday I put a normal metal film resistor in parallel with the other two (R65/R68). Measured total resistance now is 1,33 ohm. OK.
Behavior is almost unchanged. With +15 V on the positive supply the main pulse appears only with R90 fully CW.
A strange thing I noticed: with the unit just powered up the pulse width initially is approx 500 ns or so.
While the unit warms up, the pulse width gradually decreases until it disappears completely after 1 or two minutes.
To make it reappear I have to decrease the supply to 14.5 V.
Suspecting some thermal drift, I tried to blow (with my breath) over the zone around R109 ( the resistor in series with the 5.1 V zener): the pulse reappears except to shrink in duration and then disappear again within 30 sec.
Oh yeah, in that pcb zone there are the resistors R106,R107 and R108 which generates the -3 V and - 2.5 V reference voltages. Measured, they are 3.1 and -2.6 respectively.

I don't think Tek made such a critical temperature circuit. I suppose that the optimal operating conditions are reached after a few minutes of warm-up and that the engineers have taken into account the heating produced by the 85 ohm resistor for the zener polarization.
Resistor values in the circuit area around Q86, Q88, Q80 are all in the range.

I will patiently check all the waveforms by comparing them with those shown on the diagram and I will check the voltage values on the basis of Q86 and Q88 also.

Varying the voltage between 13 and 15.5 V I noticed another phenomenon: the DURATION of the main pulse first decreases, then abruptly increases and then decreases again until it disappears completely at 15.1 / 15.2 V. Seems to be a timing problem, just like Albert said.
To better describe this phenomenon, a video would be needed. If I can do it, I'll upload it on the TekScopes forum.

Max

unclebanjoman
 

While checking the circuit I just discover that SN74L93 and SN74LS93 are NOT pin-to-pin compatible!
I was using the "classical" 7493 datasheet from Texas Instruments datasheet and wondering how it worked ...
I never imagined that the 74L93 had a different pinout! I found the correct pin-out in a very old National databook.

Max

Carl Hallberg
 

That is true for other 74 series.  Should always check pinout.  Example, 74L,LS,W: 95, 85, 78, 72, 74 and maybe others.
Carl Hallberg (W9CJH)

On Saturday, July 4, 2020, 1:21:11 PM CDT, unclebanjoman <mmazza@...> wrote:





While checking the circuit I just discover that SN74L93 and SN74LS93 are NOT pin-to-pin compatible!
I was using the "classical" 7493 datasheet from Texas Instruments datasheet and wondering how it worked ...
I never imagined that the 74L93 had a different pinout! I found the correct pin-out in a very old National databook.

Max

Harvey White
 

74L series of chips should be checked for pin compatibility with anything else.  Not sure why they did that, but that's been a design problem for any number of years when substituting.

Also, not all 7400 series chips have power and ground at 14/7 or 16/8.  Surprise....

Harvey

On 7/4/2020 2:21 PM, unclebanjoman wrote:
While checking the circuit I just discover that SN74L93 and SN74LS93 are NOT pin-to-pin compatible!
I was using the "classical" 7493 datasheet from Texas Instruments datasheet and wondering how it worked ...
I never imagined that the 74L93 had a different pinout! I found the correct pin-out in a very old National databook.

Max


Jim Ford
 

Probably to decrease parasitic inductance on the power and ground pins.  Wirebonds make great inductors, and as the edge speeds increased, ground and power bounce became a problem.  V = L*di/dt.  I assume the power and ground pins went to the center of the package.  Corner pins are the worst possible choice for inductance.Jim Ford Sent from my Verizon, Samsung Galaxy smartphone

-------- Original message --------From: Harvey White <madyn@...> Date: 7/4/20 12:47 PM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] S-52 only works with lowered +15 V power supply 74L series of chips should be checked for pin compatibility with anything else.  Not sure why they did that, but that's been a design problem for any number of years when substituting.Also, not all 7400 series chips have power and ground at 14/7 or 16/8.  Surprise....HarveyOn 7/4/2020 2:21 PM, unclebanjoman wrote:> While checking the circuit I just discover that SN74L93 and SN74LS93 are NOT pin-to-pin compatible!> I was using the "classical" 7493 datasheet from Texas Instruments datasheet and wondering how it worked ...> I never imagined that the 74L93 had a different pinout! I found the correct pin-out in a very old National databook.>> Max>> >>

unclebanjoman
 

Just discovered that R27 on the schematic is 510 ohm: on the board there a 2.0 kohm is mounted instead.
My timing board has the code number 670-1320-02 so I don't have the latest schematic. I bought mine from Dave's Artekmedia and it references the 670-1320-00 version of the pcb with only two schematic changes/additions (R60 and CR10).
Moreover: the voltages -2.5V and -3V, they vary if the + 15V varies, changing the operating point of BJTs. The -5.1 V is rock- steady.
The waveform at the point "E" is very critical.
Blowing some compressed air in the zone around R26 / 27 / 30/37 the pulse width at pin "E" (positive pulse at right of waveform 10) changes noticeably, influencing the width of TD pulse.
I'm reaaaaally puzzled.

Max

Dan G
 

On Sat, Jul 4, 2020 at 07:10 PM, unclebanjoman wrote:

Just discovered that R27 on the schematic is 510 ohm: on the board there a 2.0
kohm is mounted instead.
My timing board has the code number 670-1320-02 so I don't have the latest
schematic.
The timing board A1 rolled to revision -01 at s/n B080000,
then to revision -02 at s/n B091450.

R12 changed from 4.3K to 3K at s/n B091450.
R26 changed from 15K to 13K at s/n B080000.
R27 changed from 510 Ohm to 2K at s/n B080000.

On the trigger board, the snap-off diode CR66 changed
to Tek p/n 152-0252-01 (GHZ Devices, Inc. p/n GC-2534-15)
at s/n B080000. C65/R64 probably had to change at the same
time to compensate for different diode characteristics.


dan

Albert Otten
 

Hi Max,
The waveform at the point "E" is very critical.
Blowing some compressed air in the zone around R26 / 27 / 30/37 the pulse width at pin "E"
(positive pulse at right of waveform 10) changes noticeably, influencing the width of TD pulse.
I'm reaaaaally puzzled.
At the more negative level Pin E resets the output to true zero in periods 3-6 and 9, by switching the pair Q86/Q88 such that Q86 conducts and Q88 is off. Pin E is the only Timer board output which can force zero output. We saw already that zeroing also occurs at high "+15V". Now zeroing also seems to occur before the official end of the output pulse in period 8 because of temperature effects, likely also via the observed changes at Pin E.
How locally concentrated is your pressed air? The Control board (with the comparator pairs) is probably not much affected, so the cause of your problem might well be in the Timing board. I'd still like to see the base voltages of Q86/Q88 to confirm things.

Albert

Albert Otten
 

Hi Max,

For comparison I added a photo of the waveform at timing pin E to your album.
Because of both the "+15V" effect and the temperature effect also Q32 might be suspect. As far as I can see Q32 serves only for protection and normally should never conduct. In my unit the base of Q32 is around -4 V with some spikes op to -3.5 V while the emitter is -3 V. When the "+15V" increases the base voltage will also become less negative. When Q32 conducts for some reason timing then pin E is pulled down, causing zeroing the pulse output.

Albert