Odd CRT failure mode

[grumpydoc]

I'm currently trying to fix a Phillips/Fluke PM3394B 'scope and I have come across what seems to be a very odd failure mode for the CRT.

Basically the deflection sensitivity seems to have gone down in both X and Y planes. The Phillips manual says X should be 8.5V/div although the data sheet says 7.7V/div. However even with the final X amp gain turned up full traces are squashed horizontally (ditto for Y). I have confirmed that the voltage on the plates is identical to a PM3394A which works fine and swapping another CRT into the 'scope solved the deflection problem (but brought even more issues for various reasons).

I haven't measured all of the voltages but the cathode supply which is supposed to be -2200V reads about -2230 which is within the tolerance of my HV probe. Focus and astigmatism controls work as expected so the voltages on those plates should be OK and if the -2200V is good then the final anode should be at the correct voltage as it is generated off the same winding on the EHT transformer (only the 2200V is regulated anyway).

Does anyone know if it is even possible for a CRT to "fail" like this and what the mechanism might be?

[tautech]

I won't even try to tell you how to fix a CRO grumpy, but are you sure you haven't missed something?

Back to basic checks? PSU?
DC value checks throughout the schematic?

Will the position pots allow full trace movement?

[TerraHertz]

Interesting. I had a CRT in a video waveform monitor where the deflection was screwed up. But that was clearly due to the instrument being dropped heavily in shipping, resulting in the inner structures in the tube bending out of position.

As I'm sure you're aware, possible causes of reduced electrostatic deflection are:
 * Below-spec voltages on the deflector plates,
 * Electron beam with higher than spec acceleration voltage,
 * Wrong shape of post-deflection plates electric field.

Can't you measure the anode EHT? If it's too high maybe that could be the problem?

Another idea: some tubes seem to have a kind of electrostatic lens in the wide portion of the tube, formed by a resistive spiral trace going round and round inside the glass. If that's to provide more deflection 'gain', than maybe if that trace went open circuit the trace would appear squashed. Or... there would have to be an electrical connection to the end of that spiral closest to the gun. Perhaps that went open? It would be a finger contact to the inside surface of the tube, probably. Maybe try giving the tube a good hard thump? (with a soft object, after all else fails!)

Ha ha. I can't believe I'm suggesting that.

[aargee]

The tube hasn't gone 'gassy' has it? i.e. partial loss of vacuum. This can happen with a failure around the glass envelope on CRT pins.

[grumpydoc]

Thanks guys - good thoughts.

but are you sure you haven't missed something?

No and I have a nagging doubt that I am missing something, hence the post.

The problem affects both X and Y deflection so it is likely to be something "global".

Back to basic checks? PSU?

Can't you measure the anode EHT? If it's too high maybe that could be the problem?

The PSU low voltage rails all check out - the 10V ref might be microscopically low at 9.99V (+/- 5mV spec - another 5mV would be within the DMM inaccuracy).

Yes, too high an EHT voltage was one of my thoughts. I haven't measured the final anode voltage directly - I do have a 40kV probe but it would be impossible to get it to the CRT connector with the tube in situ. There is a spec for a dummy load in the service manual but I don't have a 14kV rated 290Mohm resistor handy

As a surrogate for measuring the EHT I checked the cathode -2200V supply which is OK at -2230 or so (well within the 2% accuracy of the HV probe). Both supplies come off the same winding of the EHT transformer and the regulation is feedback from the -2200V rail anyway.

Intensity, focus and astigmatism controls adjust as expected so, although I haven't measured all the tube voltages I'm not currently suspicious that they are out of spec.

DC value checks throughout the schematic?
Will the position pots allow full trace movement?

* Below-spec voltages on the deflector plates,

DC checks throughout the schematic whould be a tall order and I'm not sure how much light it would shed. Strictly there aren't any pots - it's all DAC controlled but there are good ranges of trace movement.

I have checked the deflection plate voltages - the final X amplifiers (two in antiphase) give a swing of between 80 and 96V p-p which is the same as a 3394A which works and in that 'scope about 88V of swing is needed.

I have checked that a signal is affected and that this is not just the vector generator in the 'scope and it is - I can't quite stretch a 10Vp-p 1kHz square wave to read correctly in either X or Y. I haven't checked the timebase frequency but I don't think that's it as the vector generator is entirely voltage driven and anyway that would not explain Y axis problems and the cursor system measures the signal perfectly.

* Wrong shape of post-deflection plates electric field.

Another idea: some tubes seem to have a kind of electrostatic lens

Mechanical damage isn't out of the question but to produce a perfectly symmetrical problem seems odd. The tube has some internal magnetic elements to help geometry.

The tube hasn't gone 'gassy' has it? i.e. partial loss of vacuum. This can happen with a failure around the glass envelope on CRT pins.

I don't believe so. EDIT: At least I don't believe the 1st CRT (the one with funny deflection) is gassy, the dim one with low cathode voltage on the other hand is a different kettle of bananas.

The plan was to replace the CRT as a diagnostic test as this has become a two 'scopes into one kind of exercise. Unfortunately the EHT connector on the 2nd CRT was not the same as the first so I needed to bring the multiplier over with it which in turn meant bringing the whole PSU.

That did "fix" the deflection problem but at present that is not terribly useful. The 2nd PSU is making "I am not a happy PSU" noises and the cathode supply is definitely too low by 10-15% (about -1900 - 1980V) so the final anode voltage will be too low as well. This is likely to make this CRT too sensitive to deflection voltages.

I'm going to swap the EHT multiplier this evening and see where that gets me. If the voltages are OK but I still have the deflection problem with the replacement CRT I will know to look elsewhere, if the voltage is still low then I will know that the 2nd CRT or EHT multiplier have problems (though that won't help much with the original problem).

[Ian.M]

There's always the old beerglass trick for getting at the EHT or patching in a temp. EHT lead with a more suitable connector.  Obviously its as dangerous as ****, so be careful.

Take one straight sided pint glass.  Wash out thoroughly, and rinse with IPA (97% alcohol, not export beer), hot air dry and stand it on a good high voltage insulator (e.g. a chunk of perspex) next to the chassis.  Attach the bare end of the temp EHT lead to the clip on the original lead, and put the joint inside the beerglass right at the bottom.   You may get a little corona when you power up, but the glass surface usually charges up enough to stop that.  DO NOT attempt to touch or move the glass while power is on, or after poweroff before discharging it through an EHT probe or meter.   

That was our go-to method for checking a TV with a suspected gassy tube that was overloading the EHT and triggering the protection circuits, or if we had to bench test a chassis without a tube but had a good yoke from a scrap set.

[PA0PBZ]

Just my thoughts...

- Can you adjust the original PSU/CRT combo so that the cathode is at the same too low level? Does that change anything?
- Is the display still linear?

[grumpydoc]

Just my thoughts...

- Can you adjust the original PSU/CRT combo so that the cathode is at the same too low level? Does that change anything?

It would take some component changes.

- Is the display still linear?

As far as I can tell.

[PA0PBZ]

It would take some component changes.

I know, but it still seems a very weird problem, and one you have one working with too low voltages and one not working with the correct voltages.
Hmm...

[calexanian]

I have seen things like this before. The CRT mat have a conductivity leakage issue on the deflection plates. This happens in scopes that have been on forever every day. Over time remaining things break down inside the tube and adhere to the internal structures causing electrical leakages internally. Additionally barium can be evolving from the electron gun and causing problems. Its simply new tube, or new scope time if this is what has happened.

[Artlav]

Try waving a magnet nearby (but not too close), and see if any field direction would make it improve a little.
I had a problem that sounds similar from the internals of a CRT being magnetized (someone was putting a magnet too close to it), and it got solved by degaussing the gun part of the tube.

[T3sl4co1l]

Try waving a magnet nearby (but not too close), and see if any field direction would make it improve a little.
I had a problem that sounds similar from the internals of a CRT being magnetized (someone was putting a magnet too close to it), and it got solved by degaussing the gun part of the tube.

Heh, reminds me of an old problem: misalignment of the ion trap magnet.  That was a common sight on magnetic deflection (and occasionally magnetic focus also) CRTs in the 50s.  If it's not quite right, the picture gets cropped by internal geometry (probably hitting the 2nd anode inside the gun, before making it out to the phosphor screen).

Demagnetization (of a CRT that's not supposed to be in the first place!) would be helpful, if that's what's going on.

Nickel and iron were commonly used in CRTs, and are magnetizable.  I believe Tektronix used stainless steel, which isn't as magnetizable, and probably doesn't suffer from this problem.  YMMV.

If it has been magnetized, I'd also be concerned about the integrity and magnetization of the mu-metal shield (if present).

Tim

[grumpydoc]

Well,  the CRT is exonerated!

Swapping the EHT multiplier to the "good" PSU and I get a bright trace on the previously dim CRT, and the fault is back - the trace is slightly squashed horizontally and I can't stretch it out enough with the available gain on the X amplifier. Y will be the same looking at the traces but I haven't formally checked.

One CRT with a bizarre failure mode is enough, two is implausible - it has to be something else.

The grid bias (or "dark level") won't adjust on this CRT which is different and might be relevant.

My initial instinct was that the EHT was too high, I'm still not sure about that but it is a good candidate. The next step is clearly to measure all of the CRT voltages whether they are tricky to get at or not.

Now, just to find that 290Mhom 20kV resistor.......

[Artlav]

Now, just to find that 290Mhom 20kV resistor.......

Why the odd value?
I use a 50V panel meter (which have 50kOhm) with a 50MOhm resistor on the positive lead.
Hook it up, get away, turn things on, observe the voltage, turn things off.

There are plenty of HV 50Meg resistors, from cheap USSR NOS to tingly-priced new ones on Digikey.

[grumpydoc]

Now, just to find that 290Mhom 20kV resistor.......

Why the odd value?
I use a 50V panel meter (which have 50kOhm) with a 50MOhm resistor on the positive lead.
Hook it up, get away, turn things on, observe the voltage, turn things off.

There are plenty of HV 50Meg resistors, from cheap USSR NOS to tingly-priced new ones on Digikey.

That's what the service manual suggests as the dummy load - there is a table of currents and suitable resistor combinations which for the EHT is 50uA and 290Mohm (implying 14.5kV which is still low for the tube spec of 16kV)

50uA is probably just a bit more than the tube will draw under normal conditions.

50Mohm might load the supply down too much.

Anyway Farnell sell 3.5kV 47Mohm resistors for pennies - 6 of those in series would do the trick.

[grumpydoc]

There's always the old beerglass trick for getting at the EHT or patching in a temp. EHT lead with a more suitable connector.  Obviously its as dangerous as ****, so be careful.

Take one straight sided pint glass.....

I was thinking about this as I walked to work - I now have a mental image of a pint glass full of angry electrons fizzing around and wanting to bite something

[Ian.M]

That's an excellent mental image - it will keep you safe if you ever have to put an EHT lead in a beer glass.

[grumpydoc]

Just my thoughts...

- Can you adjust the original PSU/CRT combo so that the cathode is at the same too low level? Does that change anything?
- Is the display still linear?

I am still trying to figure out the best way to build the dummy load specified for the 14kV supply - 4x 100Mohm 7.5kV resistors in parallel with the 1Gohm of the HV probe should be OK at 285Mohm but they are tending to be expensive (about £3.50 each is the cheapest I found). Plus I want to combine the order with some other stuff and it looks like DigiKey, Farnell and Mouser each have part of what I want but not everything

So, I was thinking about PA0PBZ's suggestion of tweaking the voltage and looking at the EHT supply schematic and have realised that there are a few things I don't understand.

Here is the schematic

That's a bit small to see what's going on so click on the image for the larger, readable, version.

I understand broadly what's going on, there's a self resonant flyback oscillator, the duty cycle of which can be tweaked slightly by the output of the op amp N1101 however the exact operation is confusing me a bit.

Feedback is via the 31.6Mohm resistor R1103 which forms a voltage divider with R1123 and R1101 - the top of the chain is the 10V reference voltage. The non-inverting input is held at 0V via R1106 - I presume that the inverting input should be about 0V when the -2200V line is at the correct voltage but I get about -1.6V by calculation (it's pretty close to 0V when measured).

Also given that the -2200V supply is DC (well, nominally) I don't understand what the role of C1104, C1103, R1104 or C1102 is.

Could anyone explain in detail what is going on with this supply?

To further add to the confusion in the working 3394A the -2200V rail is about -2015V which is quite a bit off -2200V (but slightly more in line with the CRT datasheet which suggests -2kV is optional for the first accelerator), but I have the 3382 as well which is running at -2245V. The 3382 does have a different CRT (D14-374GH-123) which I can't find a datasheet for but the service manual does not show a different PSU in the 100MHz 'scope.

I'm starting to feel a bit  and  trying to get my head around this problem, maybe knowing exactly what the EHT voltage is in the working/non working 'scopes will help but I'm not sure - it's going to take me a while to get the bits to do the measurement though.

[T3sl4co1l]

Looks like V1101 (who came up with these ridiculous designators) is simply controlling DC bias into the oscillator.  Something about V1103-4 (seriously, who?) pulling down on the base bias, when the signal swings sufficiently low, turning it off and allowing the cycle to proceed.  It should be roughly something like this,



where the "swing under and catch" behavior is permitted by V1108 (...gah!).

Tim

[grumpydoc]

Looks like V1101 (who came up with these ridiculous designators) is simply controlling DC bias into the oscillator.  Something about V1103-4 (seriously, who?) pulling down on the base bias, when the signal swings sufficiently low, turning it off and allowing the cycle to proceed.  It should be roughly something like this,

where the "swing under and catch" behavior is permitted by V1108 (...gah!).

Tim

Yes and your trace matches the one in the notes so that's fine.

I'm less clear on the details of the regulation and why the RC networks around the op amp, especially before trying to tweak the value of R1103 to lower the voltage a bit and see what happens..

[Ian.M]

Re Vnnnn - the drafting department was obviously still classifying all discrete semiconductors as 'valves' 

I wouldn't touch R1103 (other than cold testing).  Adding 15K in series with R1123 and R1101  would probably be a better bet to drop the 2200V supply a bit.  Use a 22K trimmer with the wiper wired to one end of the track so it *CANNOT* go open. and you should be able to 'dial an EHT'.

[T3sl4co1l]

The RCs are for filtering and compensation.  Without (or with incorrect values), it will most likely oscillate rather than deliver stable regulated DC.  Or be overdamped and too slow, which is probably the side it's erring on.

Tim

[TerraHertz]

I'm not sure you really got what I meant by this:

* Wrong shape of post-deflection plates electric field.
Some tubes seem to have a kind of electrostatic lens in the wide portion of the tube, formed by a resistive spiral trace going round and round inside the glass. If that's to provide more deflection 'gain', than maybe if that trace went open circuit the trace would appear squashed. Or... there would have to be an electrical connection to the end of that spiral closest to the gun. Perhaps that went open?

Here's an example of what the 'spiral' I referred to looks like. I've seen this on various tubes, not just this rather old one (a GEC type 1074H, dual beam. No idea what it was from. Just happened to have it handy to take pics.)


One end of the spiral is connected to the EHT (anode) aquadag. The other end has it's own aquadag area down at the tube neck, with finger contacts and an independent connection pin. The final set of beam deflection plates are inside that neck aquadag area, so the beam emerges from the deflection plates directly into the graded E-field created by the spiral resistor.


Here's your schematic, edited with what I *think* is going on, *if* your CRT has a similar resistive field shaper electrode.


The CRT-internal resistor provides both the EHT voltage feedback to the regulator circuit, and also increases the overall CRT deflection gain. It's a bit tricky visualizing how that works. Simplest way to think about it, is the electrons spend more time being deflected radially towards the tube sides, while they haven't yet picked up much extra velocity due to the axial EHT acceleration field.

There is deflection gain due to the Anode EHT, but grading the field via that resistor increases the overall deflection gain.

My crude attempt to illustrate the 'gradient', doesn't have anything to do with E-field direction. It's just E-field intensity. Effectively the tube interior sees field strength proportional to voltage along the spiral resistor on the inside wall of the tube. I have a hard time trying to mentally model the field direction, and I don't have any modeling tools. But I do recall seeing some writeup (long ago) of how that spiral system increased defection gain.

Annnnyway... if that CRT spiral resistor went open circuit at some point, the whole system would be screwed in much the way described. Poor deflection, EHT too high (further decreasing deflection gain), -2200V rail wrong, etc.

Edit to add: try measuring the current at the CRT pin of the -2200V supply. It should not vary (much?) with beam intensity, but if there is an internal resistor in the CRT (many hundreds of Mohms) between Anode and -2200V, you'll be able to see that current. Hence whether it's open circuit on the bad tube.

[T3sl4co1l]

No, that curve is going the wrong way.  The beam is being accelerated towards the face, increasingly so as it drifts along the lens region.  It should be more of a parabolic curve, starting at a wider deflection angle and straightening out towards the face.

I think it's more a matter of, you get more deflection for a given spot intensity.  In the simple CRT case, the anode is at the same potential as the deflection plates (average), so whatever intensity you get, it's determined by cathode voltage only.

You can't simply amp up the anode in a simple CRT, because the field between the last deflection plates (usually the horizontal pair) and 2nd anode, has a warped shape because it's mostly defined by the edges of the plates, and whatever the aperture of the 2nd anode looks like, at the gun end (usually a proper cylindrical electrode, otherwise internal aquadag).  Indeed, anode voltage is usually variable, near 0V, to provide astigmatism control.

The only way you can smoothly go from the potential on the deflection plates, to a higher potential at the screen, is by adding a shield, or changing the field gradually.

Later Tektronix tubes used the shield: a fine mesh screen, following the last deflection plates, biased by the ASTIG control.  The beam's position, as it passes the screen, is a significant fraction of the total deflection: the path straightens out as it passes down the acceleration region.  The screen is visible by its effect on the "texture" of the display; it necessarily absorbs some beam current, and makes the beam a bit more diffuse.  On the upside, these often used quite high voltages (up to 13kV I think?).

Tubes as pictured, implement a similar gradient, at the expense of DC bias.  The lack of shielding between accelerating region and deflection plates (maybe I'm making some assumptions here, as I don't have an x-ray view handy!) means it won't be perfect (some residual distortion from the deflection plate tips should be present), but maybe the linearity and dot size consistency will be good enough for an o-scope.

Tim

[TerraHertz]

No, that curve is going the wrong way.  The beam is being accelerated towards the face, increasingly so as it drifts along the lens region.  It should be more of a parabolic curve, starting at a wider deflection angle and straightening out towards the face.

True in a simple case where the beam sees only a linear field between the anode and the deflector plates. My diagram would be more correct if the beam path approaching the anode was as you say, bent the other way. You try drawing subtle beam path curves in photoshop on a small diagram.
Mainly trying to make the point that with the graded field produced by the spiral resistor, the beam in the early part of the tube doesn't see the final anode EHT field yet. So, it spends longer with about it's original deflection (whether increasing or decreasing by the radial field, I'm not confident to argue about that) and so has an overall greater deflection in the graded field.

I think it's more a matter of, you get more deflection for a given spot intensity.  In the simple CRT case, the anode is at the same potential as the deflection plates (average), so whatever intensity you get, it's determined by cathode voltage only.

What? Deflection is not affected by spot (beam) intensity. Thankfully. Imagine how messed up trace intensify highlighting modes would be if that were the case.
Edit to add: Oh wait, by 'spot intensity' you meant EHT voltage (higher EHT allows maintaining spot intensity despite faster sweep speeds.) Not variations in beam intensity during operation.

Also "anode is at the same potential as the deflection plates"... What? Is that a typo, or did you mean it?

You can't simply amp up the anode in a simple CRT, because the field between the last deflection plates (usually the horizontal pair) and 2nd anode, has a warped shape because it's mostly defined by the edges of the plates, and whatever the aperture of the 2nd anode looks like, at the gun end (usually a proper cylindrical electrode, otherwise internal aquadag).  Indeed, anode voltage is usually variable, near 0V, to provide astigmatism control.

Wait, by 'anode' you're referring to the final focusing electrodes? Strictly speaking, the 'anode' is the EHT terminal and CRT face, since that's where the electrons end up. Other 'high positive voltage' elements are numbered anodes, but don't take any beam current.

The only way you can smoothly go from the potential on near the deflection plates, to a higher potential at the screen, is by adding a shield, or changing the field gradually.

Which is what I'm describing.

Tubes as pictured, implement a similar gradient, at the expense of DC bias.  The lack of shielding between accelerating region and deflection plates (maybe I'm making some assumptions here, as I don't have an x-ray view handy!) means it won't be perfect (some residual distortion from the deflection plate tips should be present), but maybe the linearity and dot size consistency will be good enough for an o-scope.

On that tube, the internals are visible and there's no shielding. The final deflection plates (definitely the horizontal, since the Y plates are twinned due to being dual beam) are in the middle of the spiral low-end aquadag area.
But any distortions are less than you'd get with exposure to the full EHT field, since they are inside that much lower voltage shell. There are some internal planar elements at the same potential as the aquadag, but they are well back from the exit aperture of the H-deflection plates.

But this is all kind of irrelevant. Point is, does grumpydoc's CRT have that kind of internal resistive EHT field shaping? (Which is primarily to achieve higher deflection gain, regardless of fine details of which way the beam curves.)
And if his CRT does have that, is the resistor open circuit?