Tektronix 465 fat trace problem

[skautkurt]

Hej!

First: this forum is a great information resource, thanks for that!
Also, this is my first post here, so please have a bit of patience with me

I bought a Tektronix 465 for a very good price on the local craigslist equivalent, and I am absolutely in love with it. The tactile knobs, the whole buildquality... amazing. It even came with the tek scope mobile!
After a frantic unpacking of the giant package I plugged it in and pulled the power knob, and tada, we got two traces!
A huge sigh of relief was expressed, as the pictures on the offer were not really good and detailed.

But then a (hopefully) small problem became obvious: Without the 20mhz bandwidth limiter on, the roofs of the calibration square wave are really fat and fuzzy. This happens on both channels. I am attaching some pictures below. With the limiter pulled/activated the signal is very clean and I get a sharp trace.
I know that big bandwidth might bring some noise, but I don´t think that this amount of noise is normal.





(sorry for the picture quality, I only had my smartphone on hand)

I opened it up (after careful reading of the Users Manual) and measured the Voltage testpoints for +55, +15, +5, -8 and +110 V. All of these were right in spec, only 110v was on the upper scale of the limit.

While the scope was open, I also did a thorough visual inspection, and found some gunk around the bases of the big silver caps behind the trigger circuit board, but I am not sure if it is leakage (the board beneath looks fine) or if it´s just flux. Does anybody know?







I am pretty new to the oscilloscope game, and so I decided to ask my questions here:
Where would one start to check/test to maybe fix this issue? Is it an issue? What might cause the problem? Might a normal calibration fix it or is it more likely to be an ill-aged part (like the silver caps)? If the voltages are in spec, does a ripple measurement make sense? I do have access to a digital Oscilloscope, so that would be an option.

That are a lot questions, but maybe someone here had similar problems. I would really love to get the scope working completely.

Thanks in advance, and sorry for the long post
Cheers,
Timo


Oh and P.S.: is there a serial number anywhere inside the device? The one from the front panel is missing on my unit...

[tggzzz]

Where would one start to check/test to maybe fix this issue? Is it an issue? What might cause the problem? Might a normal calibration fix it or is it more likely to be an ill-aged part (like the silver caps)? If the voltages are in spec, does a ripple measurement make sense? I do have access to a digital Oscilloscope, so that would be an option.

The ripple spec is vital; a large ripple indicates damaged capacitors. Basically any ripple will feed through and be visible on the trace. First get the ripple right, then get the voltages right; be aware that if you change the 50V line then all calibrations will have to be done.

Quick test for damaged PSU ripple filter capacitors. Take another capacitor (equal or higher voltage, capacitance 1/10 to 1/2 of the nominal capacitance) and temporarily tack solder it in parallel with the suspect capacitor. If the ripple reduces proportionately, the capacitor needs replacing.

Be very very careful using a scope anywhere near the mains or on an SMPS; the scope probe's shield is grounded but the various bits of the SMPS might not be. Make a mistake and sparks will appear. Best to use a differential probe with appropriate voltage rating.

[BreakingOhmsLaw]

First, hook it up to a trusted signal generator. Just to avoid digging into the front end, only to find out that the cal signal is bad.

[woodchips]

Even simpler, is the trace correct with no signal in at all? Does the thickness of the trace vary as the volts/div switch is rotated from one end to the other? Put your finger on the channel signal input, around 1V/div, to get a mains 50/60Hz hum, what does it look like, finely focused or not? Does the trace width vary as the timebase speed is changed, no input?

That the vertical part of the signal is finely focused probably means the EHT voltages are correct, good.

[David Hess]

What you show is normal when using the build in calibration signal which has an unshielded connection to the oscilloscope probe.  With a 100 MHz bandwidth, a standard x10 probe connected to the calibrator output will pick up all kinds of extraneous EMI.

Some Tektronix oscilloscopes used a BNC connector for the calibrator output and with a coaxial connection to the x10 probe using an adapter, the trace was clean.  Like BreakingOhmsLaw suggests, either connect an external signal generator using a coaxial cable or use the x10 probe with a probe tip adapter connected to the BNC output of an external signal generator.

[skautkurt]

Thank you all for your quick replies, I will test all these things tomorrow.

cheers,
Timo

[bd139]

Lazy test. Stick your DMM on AC volts and measure the rails with that. The ripple is likely 120Hz which is going to be within the AC bandwidth of your meter. If anything is way out like the caps it will be a big AC voltage floating on the DC line.

Got a 465 too of same vintage for ref. Lovely scopes mine had original caps in it still bad a couple of tants.

[David Hess]

The power supply input capacitors will always show ripple (1) when measured but the output capacitors should have essentially zero ripple.

You can check if AC line ripple is affecting the trace by changing the trigger source to LINE and the sweep speed to 1ms/div or slower.  With the sweep synchronized to the power line, signals at the vertical input average out somewhat to the eye.

(1) Up to a couple volts of ripple on an average or RMS responding AC meter is typical.  What really matters is if the dip in the input voltage allows the regulator to drop out.  In theory some multimeters can measure the low peak in min/max mode but I have never used one which can actually make this measurement at even power line frequencies.

[skautkurt]

Hej,

I put the my probes onto a different, external signal generator and also the calibration output of another oscilloscope (siemens oscillarzet d
d1010). I got the same, fuzzy result. I am going to try putting my probes onto the 1x mode, and will look into the results.

I tried measuring the ripple, but I am not too sure if I did it right. I put my probe onto the test point given in manual, and put the ground clamp onto a ground test point. Then I put the digital oscilloscope into a low mv setting, and started measuring the peak to peak voltage. Is that the right way? (All of this while the tek is turned on)

To the other Tek 465 owners: is there a serial number inside of the scope? As mine is missing its serial number I don't know if I am using the right manual as reference, as the pdf I found online is for sn > 250k. The date code on some of the ics on the inside seems to indicate `74.

Cheers,
Timo

[bd139]

Occasionally they hand write the SN on the label in the shield over the front end of the CRT. Otherwise it's just on the front.

My SN is 104xxx and is 1974 dated.

[deepskyridge]

I just repaired a tek 465, replaced all 5 big can caps in the power supply. 110v rail had 22 volts of ripple. -8v rail had 9 volts of ripple. Works great now. I would replace these first.

I have a PDF document that shows how to do the cap replacement but the forum says it's to large to attach.

I can provide it if someone want's it.

Good Luck
Gary 

[IanMacdonald]

What you show is normal when using the build in calibration signal which has an unshielded connection to the oscilloscope probe.  With a 100 MHz bandwidth, a standard x10 probe connected to the calibrator output will pick up all kinds of extraneous EMI.

My Kenwood CS5175 shows no such effect. The cal trace is clean even with the probe ground off. Or even if a piece of unshielded wire is is used to connect the cal. Of course if the probe is left open circuit it will generate all kinds of 50Hz waveforms. If I set it to 1us/div and free run/no trigger the result looks a little like a 'fat trace' but I had to seriously misadjust the controls to achieve that. It will not produce a 'fat trace' under any normal conditions.

[skautkurt]

I just repaired a tek 465, replaced all 5 big can caps in the power supply. 110v rail had 22 volts of ripple. -8v rail had 9 volts of ripple. Works great now. I would replace these first.

I have a PDF document that shows how to do the cap replacement but the forum says it's to large to attach.

I can provide it if someone want's it.

Good Luck
Gary 

That would be great! I think I found a similar pdf as well. Can you tell me, what kind of caps you used? do you have a link?

Cheers,
Timo

[David Hess]

What you show is normal when using the build in calibration signal which has an unshielded connection to the oscilloscope probe.  With a 100 MHz bandwidth, a standard x10 probe connected to the calibrator output will pick up all kinds of extraneous EMI.

My Kenwood CS5175 shows no such effect. The cal trace is clean even with the probe ground off. Or even if a piece of unshielded wire is is used to connect the cal. Of course if the probe is left open circuit it will generate all kinds of 50Hz waveforms. If I set it to 1us/div and free run/no trigger the result looks a little like a 'fat trace' but I had to seriously misadjust the controls to achieve that. It will not produce a 'fat trace' under any normal conditions.

Skautkurt did not give enough details to know exactly what is going on (1) but every 100 MHz oscilloscope with a x10 100 MHz probe should show significantly more noise without the 20 MHz bandwidth limit or a coaxial connection made to the probe.  If it does not, then something is broken.

On a 100 MHz oscilloscope with x10 probe, connecting the probe ground lead to the tip and waving it around actually works pretty well as an EMI sniffer.

The Kenwood CS5175 will show about 1/3rd the noise from its calibration signal simply because its calibration signal has about 3 times the amplitude of the Tektronix 465 calibration signal.

(1) Skautkurt's photograph shows 20mV or 200mV per division depending on if a x1 or x10 probe is being used but the 300mV peak-to-peak calibration signal rules out 200mV per division which is inconsistent with what he wrote so it is not entirely clear what he measured or how.

[deepskyridge]

I just repaired a tek 465, replaced all 5 big can caps in the power supply. 110v rail had 22 volts of ripple. -8v rail had 9 volts of ripple. Works great now. I would replace these first.

I have a PDF document that shows how to do the cap replacement but the forum says it's to large to attach.

I can provide it if someone want's it.

Good Luck
Gary 

That would be great! I think I found a similar pdf as well. Can you tell me, what kind of caps you used? do you have a link?

Cheers,
Timo

Here is where I found the document:

http://docplayer.net/35447553-Tektronix-465-oscilloscope-power-supply-capacitor-replacement-ron-childress-october-2008.html

I used Nichicon caps from Mouser.


Good Luck
Gary

[skautkurt]

Skautkurt did not give enough details to know exactly what is going on (1) but every 100 MHz oscilloscope with a x10 100 MHz probe should show significantly more noise without the 20 MHz bandwidth limit or a coaxial connection made to the probe.  If it does not, then something is broken.

On a 100 MHz oscilloscope with x10 probe, connecting the probe ground lead to the tip and waving it around actually works pretty well as an EMI sniffer.

The Kenwood CS5175 will show about 1/3rd the noise from its calibration signal simply because its calibration signal has about 3 times the amplitude of the Tektronix 465 calibration signal.

(1) Skautkurt's photograph shows 20mV or 200mV per division depending on if a x1 or x10 probe is being used but the 300mV peak-to-peak calibration signal rules out 200mV per division which is inconsistent with what he wrote so it is not entirely clear what he measured or how.

Hej, thanks for the input. I will set it up again in the afternoon, giving more specific information, to clear up the current situation.
Timo

[james_s]

Lovely instrument isn't it? I rarely use my 465B anymore but I keep it around because it's just so nice. Also occasionally there are applications where an old fashioned analog scope still has advantages.

[David Hess]

I have a 465 that I am slowly refurbishing but my go-to oscilloscope is a 2230/2232 which I think has a nicer looking CRT image.

I have looked for a modern upgrade but none have any additional features beyond a 2232 or 2440 that I want and the features that I do want they lack.  In some cases, they are a step backwards.

[skautkurt]

Update time:

I didn't get around to posting this yesterday. I did some further testing, especially addressing the points David mentioned.
So, I went ahead, set the scope to 20mV and 50µs per division, and hooked channel one via a 10x probe onto the calibration output of another (trusted) oscilloscope. The calibration wave is a 1khz 1v p to p Square. It was a cheap probe, but I also tried the same with a tektronix p6010 (which I don't completely trust, as is shows its age), and it showed identical results.



The result was the trace on the photo below.



I think, I really have to check the ripple.

I tried measuring the ripple, but I am not too sure if I did it right. I put my probe onto the test point given in manual, and put the ground clamp onto a ground test point. Then I put the digital oscilloscope into a low mv setting, and started measuring the peak to peak voltage. Is that the right way? (All of this while the tek is turned on)

Would this be the right way?

I hope this clears up some of the questions.
have a nice weekend,
Timo

[skautkurt]

Even simpler, is the trace correct with no signal in at all? Does the thickness of the trace vary as the volts/div switch is rotated from one end to the other? Put your finger on the channel signal input, around 1V/div, to get a mains 50/60Hz hum, what does it look like, finely focused or not? Does the trace width vary as the timebase speed is changed, no input?

That the vertical part of the signal is finely focused probably means the EHT voltages are correct, good.

I tried the finger thing at full bandwith, and the trace looks sharp and clear and showed a 50hz signal.
So that seems to work.