|
I tried to pull CR130 off, in case one of the diodes is damaged, but no change.
Then I checked (rapidly I admit) all resistor values: they're within tolerances, and with very symmetrical values. For some, I must pull out the ICs since their presence alters the measured value. I will do in the weekend with more calm and patience.
All the small power resistors appear to be carbon film rather than carbon composition. It seems quite reasonable to me since the p.c.b. is a latest version: in fact it carries the S/N 670-2323-08 with the "08" applied with a sticker.
In the meanwhile, I checked carefully the resistors in the input source follower. I found R133, a very little composition resistor, marked 16 ohm 5% (brown-blue-black-gold).
On the schematic is marked as selected (selected and added if necessary). The measured value is 20.3 ohm.
It's clearly out of tolerance, but I'm not sure it affects the signal.
Or maybe yes: I think it affects the polarization of the Q150A fet.
What do you think?
Should I replace it with a modern metal film resistor or must I necessarily use a carbon comp. one?
Max
|
On Thu, May 12, 2022 at 11:32 PM, unclebanjoman wrote:
In the meanwhile, I checked carefully the resistors in the input source
follower. I found R133, a very little composition resistor, marked 16 ohm 5%
(brown-blue-black-gold).
On the schematic is marked as selected (selected and added if necessary). The
measured value is 20.3 ohm.
It's clearly out of tolerance, but I'm not sure it affects the signal.
In my schematics, no R133 is present, so I can't be specific. Adding a low-value resistor if necessary, else leave open, which is the implication of what you write, seems strange to me, unless it's across an L. Just to check, you could temporarily tack an 82 Ohm R across R133 (on the solder side of the PCB) and be within 2% of nominal 16 Ohm. Raymond
|
After a carefully check of all resistors, I decided to examine the signal path in more detail using the 2 GHz Philips PM3340. From the last two pics I've posted, it looks like the jfet Q150 is the culprit. Pulling out Q150 and examining the signal at the gate pin (R132/R134) I obtain this: /g/TekScopes/photo/262075/3430113With Q150 inserted, the signal become this: /g/TekScopes/photo/262075/3430114very similar to the waveform observed with the 7854 using the defective 7A16A plugged in. So? I think the Q150 dual jfet has failed. Max
|
On Sat, May 14, 2022 at 10:43 PM, unclebanjoman wrote:
So? I think the Q150 dual jfet has failed.
Could be. OTOH, remember that the gates of these FETs are Hi-R only as long as their Gate-to-D/S junctions are reverse-biased. As an example, a shorted C158 (dipped tantalum?) would thoroughly disturb things. It my be a good idea to measure some DC levels with Q150 in -or just replace Q150 if you have another sample available. Raymond
|
Hi Raymond, I measured voltages in the source follower stage. I posted a screenshot of the partial schematic with measured voltage values written in red. There are also some diodes added by Tek, not present in the original schematic. I draw these extra diodes in red also. I was unable to recognize the type.I think they are for protection purposes. The schematic with my annotations is here: /g/TekScopes/photo/262075/3430589I tried also paralleling R133 with a 82 ohm resistor to bring it back to the correct value of about 16 ohms but with no visible effect. I tried to short completely R133 but also without noticeable effects. I really don't know its purpose. Just for curiosity I measured the signal on pin 14 of U350. It is visible here: /g/TekScopes/photo/262075/3430590I don't really know what else to try Max
|
Perhaps one of the capacitors, C142, C156 or C158 has gone open or nearly so. Try paralleling them with known good capacitors (with short leads) of the same value.
Stephen Menasian
|
On Sun, May 15, 2022 at 08:07 AM, unclebanjoman wrote: ...
I measured voltages in the source follower stage.
I posted a screenshot of the partial schematic with measured voltage values
written in red. All the DC voltages look fine. Unless there is a subtle failure mechanism that doesn't show up in DC operating point JFET looks good. If you probe left side of R132, is rising edge clean? If left side of R132 looks good R132 may have went up in value. If you don't want to lift one side of R132 to measure you can short its right side to ground and measure resistance of parallel combination of R132 and R130 =~ 320k. Stephen's suggestion of checking caps is also good. Ozan
|
Just replaced all 1000 pF caps, C132, C142, C144, C156. No luck. No relevant changes detected. While changing caps, I measured R132 = 467.5 kohm. R130 reads practically 1 Megaohm Then, following Ozan's suggestion, i checked the input waveforms. Results: Waveform immediately after R10 (right side of R10 on the schematic) with the input selector on GND position: /g/TekScopes/photo/262075/3430811Same as above, but with selector on DC position: /g/TekScopes/photo/262075/3430812Waveform on the left side of R132 (junction of R130/R132): /g/TekScopes/photo/262075/3430813I must admit that the last waveform is somewhat lousy. I'm more and more confused. Max
|
On Sun, May 15, 2022 at 05:07 PM, unclebanjoman wrote:
I don't really know what else to try
Hi Max, Currently, I'm not where I have my equipment, doc and parts so I'll do without: I'm pretty sure that Q150 is an often-used FET (151-1032-00?), so it would be easy to replace. I'm almost certain I have a few. You're in Italy, right? I'm in NL, so replacing it wouldn't be a real problem, even if you don't have a replacement. I can check if I have one this Tuesday. Since these FETs are basically symmetrical, you could swap both by rotating the case 180 degrees, depending on pinout, just to try, also depending on how the case is connected. R133 isn't in my schematics, so I didn't know where it was until now. It's not surprising that it doesn't make any difference re. your problem. The DC voltages in the circuit seem fine. I agree with you that the red diodes and - CR130 - are for protection only. Are you sure that continuity from the BNC input to R132 is OK? You can't really trust your 500 Ohm 'scope probe input at the gate of Q150A. Still, I don't quite understand your different observations at the gate of Q150A with Q150 in vs out, although of course, there's Q150A's gate capacitance. I have a feeling the problem isn't in the area around Q150. What do you see at U350 pin 14 if you first pull the IC? Again, don't worry, easily replaceable IC, not expensive. Carefully lift each pin (spider leg) a little in sequence and go around until the IC comes free. Do you have another 155-0078-xx available for testing? Sorry, can't be of much help ATM. Raymond
|
On Sun, May 15, 2022 at 12:52 PM, unclebanjoman wrote:
Just replaced all 1000 pF caps, C132, C142, C144, C156. No luck. No relevant
changes detected.
While changing caps, I measured R132 = 467.5 kohm.
R130 reads practically 1 Megaohm
Then, following Ozan's suggestion, i checked the input waveforms. Results:
Waveform immediately after R10 (right side of R10 on the schematic) with the
input selector on GND position:
/g/TekScopes/photo/262075/3430811
Same as above, but with selector on DC position:
/g/TekScopes/photo/262075/3430812
Waveform on the left side of R132 (junction of R130/R132):
/g/TekScopes/photo/262075/3430813
Waveforms look like reflections coming from the cable between signal source and the plugin. If you double the length of the coax you use, does the place where first step happens in the middle picture change? In the setup coax coming from the signal source needs to be terminated to 50-ohms at the point it enters the plugin. Either a feedthrough 50-ohm termination (ideal) or a BNC-T with one port terminated to 50-ohm is needed. Ozan
|
Ozan:
yes, I'm always using a x10 50 ohm attenuator and a 50 ohm feed-thru termination at the BNC plugin. The same pieces I use with a 7A16 (not A) without problem.
Using a cable 50 cm more long, the width of the first step increases a little. Remember that I measure the waveform after R10. Measuring directly at the BNC pin of the plugin, the waveform changes, in the sense that the middle step is a the 80% of the final value, instead of the former 50%.
Raymond:
I did not check the continuity between the input BNC and R132. The form of the signal gives me strong suspicions that there may be something anomalous in the signal path even if, being the input selector au 10, V / div, all the "A" contacts should be closed ... giving a direct path from the BNC to the R130/R132 junction.
I'll check the continuity with a DVOM tomorrow.
And yes, I'm from Italy.
I already tried to reverse the Q150 since it's symmetrical. No changes whatsoever.
Unfortunately I only have the 500 ohm probe original from Philips, given together with the 3340 having type N connectors. I have a good 7A19 but not a suitable probe (I'm searching one at a reasonable price).
I've already tried to swap U350 and U450 some days ago, with no results, so I doubt it's the culprit.
I will try to unplug U350 and measuring the signal at pin 14. At this point we might as well try all the possibilities ...
I have tried also a 155-0078-01 pulled out from a 465 I use for parts, still with no noticeable results.
Max
Max
|
On Sun, May 15, 2022 at 10:17 PM, Ozan wrote:
Same as above, but with selector on DC position:
/g/TekScopes/photo/262075/3430812
There's definitely a reflection there. AFAIR, the OP mentioned that suddenly, things went wrong after being OK before, so I assumed what Ozan mentions: In the setup coax coming from the signal source needs to be terminated to 50-ohms at the point it enters the plugin. Either a feedthrough 50-ohm
termination (ideal) or a BNC-T with one port terminated to 50-ohm is needed. was already in the setup. Come to think of it, I see a few more strange things in the photographs: The curves' slow start of the PG506, as shown on the PM3340, is surprising and the images inside the 7A16A seem very "rough". Neither look like my observations with my PM3340. Together with Ozan's remark, could user error be involved? Raymond
|
I use the same cables and terminations with my 7A16 and 7A19 with no problems. I've checked and tried different feed-thru terminations and cables, with no difference.
The same cable and termination (they are labeled and distinguishable) I used to calbrate this 7A16 about one year ago.
The problem had probably been dormant for some time. I realized it a cople of weeks ago while I was making measurements on an RF oscillator output using the offending 7A16A.
As the frequency changed between 100MHz and 200 MHz, there was a point where the amplitude of the generated sinewave decreased and then rose again, something that did not happen to me with the 7A19.
Suspecting something anomalous in the bandwidth in the 7A16A, a quick check with the PG506 revealed (to my regret) the problem I am now trying to solve.
The slow start of PG506 is due to the fact that, when i took that photo, its output signal was very low. For very low signals my PG506 always has a slow start, that decresases with increasing amplitude. In fact I usually use a x10 attenuator to circumvent this idiosyncrasy when calibrating the 5/10 mV range. Which I didn't when I took the photos with the 3340.
I remember that it was talked about in other topics a long time ago.
Anyway, this evening I will try some other experiments.
Max
|
Thinking of, you should note that the input signal is clean with no aberrations when the input selector is in the GND position. In this case the input signal is grounded via the series combination R102/R104 (1 Mohm + 56 ohm) and all things are fine.
Positioning the selector in the DC (or AC, has the same effect) things changes completely. Seems that some strange thing happens inside the input stage... while theoretically should see always 1 Mohm.... or not?
Max
|
On Sun, May 15, 2022 at 11:49 PM, unclebanjoman wrote:
Thinking of, you should note that the input signal is clean with no
aberrations when the input selector is in the GND position. In this case the
input signal is grounded via the series combination R102/R104 (1 Mohm + 56
ohm) and all things are fine.
Positioning the selector in the DC (or AC, has the same effect) things changes
completely. Seems that some strange thing happens inside the input stage...
while theoretically should see always 1 Mohm.... or not?
Because of C132 high frequencies (or faster edges) will see a lower impedance. However, it should still be relatively higher impedance than the 50-ohm source. Do you put the attenuator at source side or at the plugin side? Does removing Q150 still clean up the waveforms? If you keep Q150 in place and remove U350 does it change anything? This experiment is in case somehow amplifier stage is loading Q150 output and it is not able to function as source follower. I think you swapped U350 but there could be a fault in the components around U350. The problem with this theory is final settled amplitude is correct. Otherwise cam switches of the attenuator (e.g. if one of the attenuator sections is not completely isolated) and attenuator ground connections are possible suspects. However, it doesn't explain behavior with Q150 removed. Ozan
|
On Mon, May 16, 2022 at 09:12 AM, Ozan wrote:
On Sun, May 15, 2022 at 11:49 PM, unclebanjoman wrote:
Do you put the attenuator at source side or at the plugin side?
Yeah, I always use this signal path: PG506 -> 80 cm coax -> x10 50 ohm attenuator -> 50 ohm feed-thru termination -> BNC of 7A16A Does removing Q150 still clean up the waveforms? If you keep Q150 in place and
remove U350 does it change anything? This experiment is in case somehow
amplifier stage is loading Q150 output and it is not able to function as
source follower. I think you swapped U350 but there could be a fault in the
components around U350. The problem with this theory is final settled
amplitude is correct. O.K. I will try with mooore calm this evening.
Otherwise cam switches of the attenuator (e.g. if one of the attenuator
sections is not completely isolated) and attenuator ground connections are
possible suspects. However, it doesn't explain behavior with Q150 removed.
I'm thinking the same. One test at a time and I will report to you ASAP. Max
|
Some good news finally. After more and more reflections I decided to bypass entirely the input BNC and the attenuators' section. I put the V/div switch in the 20 mV position. Pulled out the X2 attenuator, setting the input switch to GND (just to be absolutely safe). Then I placed a female BNC, firmly fixed with a screw to a post (so assuming a secure GND contact) and the central pin of the BNC soldered directly to the union point of R130 / R132 on the pcb with a piece of nude wire. The result of the hack can be seen here: /g/TekScopes/photo/262075/3431632After a few quick adjustments of the trimmers, the waveform is shown as in this figure: /g/TekScopes/photo/262075/3431633Since I'm using the flex extender, there are some ringing and reflections, but I know from experience that they disappear once the plugin is inserted directly into the manframe. The risetime is practically the same as before the failure. The next step is now to understand what is wrong with the attenuators and the input BNC section. Since normally in the DC position and 5 or 10mV/div all the attenuators are shorted (all "A" contacts closed) and the signal should go straight to the R130/R132 junction, so by instinct I tend to ipotyze: 1) A false contact on the DC path. 2) A non optimal ground connection, perhaps due to oxidation or a loose screw. 3) Something of absolutely unexpected (who knows?) At this point I will probably have to remove the attenuator pcb with its cam (groan). Max
|
On Mon, May 16, 2022 at 12:02 PM, unclebanjoman wrote: ...
Then I placed a female BNC, firmly fixed with a screw to a post (so assuming a
secure GND contact) and the central pin of the BNC soldered directly to the
union point of R130 / R132 on the pcb with a piece of nude wire.
The result of the hack can be seen here:
/g/TekScopes/photo/262075/3431632 Very impressive setup. ...
At this point I will probably have to remove the attenuator pcb with its cam
(groan).
On similar cam based switches sliding a thick paper soaked in IPA between the contacts worked well. I believe this is also what Tek recommends. Ozan
|
Unfortunately I am still struggling with the problem.
Entering directly with the signal on the BNC applied as in the photo, the behavior is perfect.
I would like to point out that I have removed all the attenuators, and the signal remains perfect in the 20 mV / div position, in which the contacts "A" are all closed except number 20, in which "A" is open and "B" is closed. Since the attenuators are not present, this guarantees that the attenuator section is completely isolated.
At this point I noticed a strange thing: by varying the Volt / div from 20 mV / div upwards, the signal displayed on the mainframe remains almost unchanged.
Switching on the 10 or 5 mV / div, the signal becomes distorted, almost as if applied to the BNC on the front panel (initial situation): it ramps well, then at approx 50% amplitude slows down.
I was stunned.
Proceeding by exclusion, I tried to:
- remove C10 (thinking of a dispersion)
- connect R10 directly to the attenuators pcb: no change
- put jumpers on contacts "A": no or imperceptible changes
I haven't disassembled the attenuators pcb yet, but I removed the readout pcb to check if there was any oddity. Examined with a magnifying glass. No traces of dirt or oxidation, contacts are clean.
Then II noticed that excluding the input BNC on the front panel (removing the central connection to the attenuator's pcb) the signal improved, not by much but the rise-time went from almost 3 ns to about 2.5 ns and the waveform was less distorted.
After many hours of cross-checking, exasperated I removed my BNC pinned to the input source follower, and pinned it firmly to a post near the original input BNC. With a piece of wire I connected the central of my BNC to the input of the attenuator section.
Result: the rise-time dropped to 2 ns but with some oscillations.
It seems that the whole attenuator pcb has some kind of dispersion. It appears to act as a capacitor/inductor combination. I cannot locate a faulty spot.
The only thing I think to do is to disassemble and remove the attenuator's pcb, remove the cam and then immerse it completely first in IPA, then wash it with distilled water and a little neutral soap, rinse it, let it dry for 48h possibly in the sun and then reassemble everything after another IPA cleaning.
I don't see other alternatives. It is the first time that such a case has occurred to me in 30 years of Tek repairs.
I am really discouraged, also because this plugin is one of my favorites, and I use it regularly.
Comments and suggestions are welcome.
Max
|
On Wed, May 18, 2022 at 02:08 PM, unclebanjoman wrote: ....
At this point I noticed a strange thing: by varying the Volt / div from 20 mV
/ div upwards, the signal displayed on the mainframe remains almost unchanged.
Switching on the 10 or 5 mV / div, the signal becomes distorted, almost as if
applied to the BNC on the front panel (initial situation): it ramps well, then
at approx 50% amplitude slows down. 5mV and 10mV are the only settings in which there is a direct path from gate of the source follower to BNC. Probably this says something is happening at or to the left side of switch 5. Did AC/GND/DC switch change anything when you had the direct connection to source follower setup? Because signal dips to ~ 50% initially with 50-ohm driver perhaps R104 switch is not opening completely and showing up as a small cap in series with R104. I have seen a case where bolts of the plastic case holding the camshaft were loose and cams didn't push the contacts far enough. The only thing I think to do is to disassemble and remove the attenuator's
pcb, remove the cam and then immerse it completely first in IPA, then wash it
with distilled water and a little neutral soap, rinse it, let it dry for 48h
possibly in the sun and then reassemble everything after another IPA cleaning. For the attenuator PCB only IPA is recommended by Tek. I would not apply water or soap. IPA did a good job on three similar cases for me (two on time base plugins, and one on 485 attenuator). I put IPA in a spray bottle so I could rinse the board with slight pressure. Ozan
|
Raymond Domp Frank