HP85662A graticule out of focus but everything else is fine

[tkamiya]

On HP85662A which is a display section of HP8568B spectrum analyzers, I have several that shows graticule is out of focus but characters and traces are sharp in contrast.

Turning focus on front panel didn't improve things much.

Is there a seprate focus on graticule???  I know trace and the rest is generated differently but it doesn't make sense focus is not consistent.  The display does not seem to have burn-in, although I know it's old and tired. 

Again, the question here is NOT why my CRT fuzzy.  WHY the graticule ONLY is fuzzy.

[tkamiya]

I've already tried the focus adjustment on Z axis card.  This particular one makes very little difference.

[DaJMasta]

I don't have familiarity with this unit, but have you tried scoping the output of the digital signal generation for the graticule and text and such?  If you can trace down the source of that, I would suspect there's some noise on that signal, so that when combined with the analog signal and actually drawn on the screen, it appears fuzzy while other parts are clear.

Pictures could help, though.  It also could be something that is effecting the whole screen and is only obvious on the graticule for whatever reason.  There are probably a bunch of people here who are familiar with HP CRT display sections similar to this one too.

[cvanc]

Watching this closely; I have an old Leader analog scope that does the same thing.  I can adjust focus of both the traces and the overlays (cursors, etc), but they don't remotely focus at the same position of the focus knob.  Never have figured it out.

[tkamiya]

Thank you.  Yes, that makes sense.  I recall vaguely thinking about that then I promptly forgot!  I also wanted to shift the image slightly for phosphor may be damaged or degraded for having activated for its entire lifespan.

[KE5FX]

Sounds like a soft CRT.  Take a look at the photos and videos here and see how they compare to what you're seeing?

If so, you can 'zap' it to get some more life out of it.  But as my note on the bottom of the page suggests, I'd recommend a 1024x768 NewScope LCD instead if the budget permits.

[tkamiya]

Yes, I know that page.  It kind of looks similar but mine is more subtle.  Also, my characters are sharp, which is odd, if it was the weak CRT.

Those LCDs look very nice indeed!  I don't know if I want to invest that much.  I have more than a few like this.

[tkamiya]

Just this afternoon, I found someone's YouTube video about DIY LCD replacement for HP-141T (I think?) 

He used an inexpensive Touch screen-LCD/micro-controller, a generic one that's available for few 10 bucks everywhere, and developed a new display.  I was thinking somewhere along the line but with different logic.  He digitizes half the buffer memory, display that, while the last half fills.  Brilliant design.  I was thinking of X-Y display but of course refresh rate is a problem.  I'm start to think I may be able to roll my own. 


I think there are two companies who make replacement LCD and interface.  Those are very nice but they cost somewhere around $500. 

[KE5FX]

Just this afternoon, I found someone's YouTube video about DIY LCD replacement for HP-141T (I think?) 

He used an inexpensive Touch screen-LCD/micro-controller, a generic one that's available for few 10 bucks everywhere, and developed a new display.  I was thinking somewhere along the line but with different logic.  He digitizes half the buffer memory, display that, while the last half fills.  Brilliant design.  I was thinking of X-Y display but of course refresh rate is a problem.  I'm start to think I may be able to roll my own. 

Yes, I've seen VK2SEB's video where he used the STMicro Discovery board with his 141T.  That was a great project if you have a 141T, but the 8566's display couldn't possibly be more different.   Nothing in that video will be of any use if you want to go this route.

I think there are two companies who make replacement LCD and interface.  Those are very nice but they cost somewhere around $500.

SimmConn's replacement is the only one you should consider, trust me on this.  Or don't. 

[tkamiya]

Why do you say that?  Of course I may be on "beginner's euphoria" drug, but he took his output from basically an XYZ output from back.  I have the same output on back of my display. 

I can display the same way he did onto my oscilloscope. 

At this point, it's just an XY (and Z) image.  Only thing particular to both his and mine compared to regular XY is that our X moves at consistent rate and only Y is a variable.  (consistent per scan, of course)  I sample Y output at certain rate, display periodically, and back. 

What would be SO different?

Yes, I am aware HIS was totally analog.  Mine is digitally processed.  But when it becomes XY, does it matter at all?  After all, a simple 20Mhz analog scope can display it (in XYZ mode). 

Seems to me, the same technique will work.  No?

[KE5FX]

What would be SO different?  Yes, I am aware HIS was totally analog.  Mine is digitally processed.  But when it becomes XY, does it matter at all?  After all, a simple 20Mhz analog scope can display it (in XYZ mode).  Seems to me, the same technique will work.  No?

The difference in display content and quality between an 8566 and a 141T is massive.   Taking full advantage of the 8566's vector display requires a 1024x768 panel.  I think (don't quote me on this) that VK2SEB's project is 320x240, which is almost sorta close to reasonable for emulating the analog storage tube on the 141T.  So the best you can do is something that's not a good fit for the quality of the rest of the instrument. 

Not 100% sure if the XYZ outputs on the back provide the graticule and/or readout vectors, or just the trace.  Hopefully all of the above, but if not, that would be another drawback.

[tkamiya]

XYZ on back does NOT provide the graticule.  It's trace only.

Now I see your argument!  You are right.  It's probably not a good match.

[tkamiya]

I read somewhere, what I see is just a symptom of weakening CRT tube.  The article explains the reason characters are sharp and graticules are not that it's a function of length of the "draw".  It can focus on short vector but not the long one.  It was a "late night reading" (I couldn't sleep last night) so I don't know which article said that.


I'm thinking of an external box that houses a set of A/D converters for X/Y and somehow take in Z input.  That should feed into a PC.  Details are still fuzzy.  (I didn't sleep last night!)

[KE5FX]

I read somewhere, what I see is just a symptom of weakening CRT tube.  The article explains the reason characters are sharp and graticules are not that it's a function of length of the "draw".  It can focus on short vector but not the long one.  It was a "late night reading" (I couldn't sleep last night) so I don't know which article said that.


I'm thinking of an external box that houses a set of A/D converters for X/Y and somehow take in Z input.  That should feed into a PC.  Details are still fuzzy.  (I didn't sleep last night!)

It's basically a thermal problem from what I understand.  As the 'poisoned' cathode heats up during a given vector stroke, the surface inconsistencies have the effect of changing the size, shape, and/or location of the area from which the beam originates.  If you tweak the focus to favor the small strokes used for the readout characters, it will degrade the appearance of longer strokes used for the graticule lines.   Obviously the intensity comes into play as well, as different Z-axis drive levels are used for the various display elements.

Not all of the tubes look exactly the same as they degrade.  But in my experience, whenever it becomes impossible to achieve uniform focus at different vector lengths, the same basic phenomenon is at work, and it can be temporarily 'fixed' the same way.  A well-functioning 85662A CRT really looks great, much sharper than any other CRT-based spectrum analyzer or any with a VGA LCD.  That's why I was never 100% happy with the 640x480 LCD replacements being sold before the SimmConn displays came out. 

All that said, I definitely don't mean to discourage you from trying.   Just tempering your expectations.  It takes a lot of work to do it right, but it's by no means impossible, and anyone who tries it will gain a lot of useful experience.

[G0HZU]

I had this issue with both my HP8568B and HP8566B analysers about 4 or 5 years ago and I eventually got fed up with it and tried to see if I could improve it.

See the old eevblog thread below to see some before and after screenshots of my analysers. Quite a difference! The HP8566B display is the best and is crisp and pin sharp and stunning to look at and it is still great 5 years later. The pictures in the link don't do it justice as they make it look a bit dull and fuzzy. As KE5FX says, a healthy 85662A CRT display is a joy to look at. 

https://www.eevblog.com/forum/repair/hp-8568b-display-problem/msg1321167/#msg1321167

In both cases I managed to improve things a lot by replacing some resistors on the Z axis board. Sadly, I can't remember which trimpot I changed but it was the same one on both and it had gone noisy and erratic and had drifted away from the proper setting. Also a fixed resistor had aged a bit and I swapped it out. But I don't think this actually mattered. Before I fixed it the grid would appear first (and was sharply focussed) as I turned up the brightness but there would be no trace. But by the time the brightness was up enough to show the trace, the grid had bloomed to a fat fuzzy line as you can see in the images in the link above. The grid dominated the display trace and hid it behind the grid. But now the grid stays pin sharp along with the trace and the trace dominates the grid. Much better!

[tkamiya]

I actually have 4 or 5 display sections here and two of them are PIN SHARP!.  So I know what you are talking about.  I exchanged the X axis board and results were the same.  I am guessing either a soft tube or how grid is generated is causing my issue. 

If any of you need parts, let me know.  I have a junk piece(s) that I can harvest stuff from. 

[G0HZU]

Thanks for the offer of spares! I'm in the UK so probably not viable to ship stuff to me otherwise I'd probably buy quite a few things from you. I've been using my HP8568B since 1990 at work and I bought it from the company a number of years ago. It has a lot of sentimental value because it was one of the company's most prized assets back in the early 1990s when the company was still quite small. The company couldn't afford to buy it so it was donated (new) by one of our customers under a contract and it was treated like royalty for many years. However, it has seen a lot of use in the time it was at work. It must have been used  every working day for over 15 years as it was in constant demand. The CRT looks good in my linked images but in reality it isn't as crisp as the HP8566B CRT and this is probably because of the high number of hours it has been used for.

One thing that would be interesting to try across all your HP8568B RF sections would be to see how low the -10.7MHz spurious term is that is reported in this link below:

https://community.keysight.com/thread/5220

You can see that some analysers have quite a high leakage level of this -10.7MHz term. I think it is probably something to do with maladjustment of the 2nd IF filter or maybe an LO3 drive level or mixer 3 issue. But I think the most likely reason is an issue with the 2nd IF filter.

[tkamiya]

I'll try that later.  I *just* moved all of mine out of my lab, and don't really want to hook that all up again.  I have a plan of having up 5 at a time and do some comparisons, so I'll do that as one of my tests.

If you need some oddball parts, if it's small, it may be feasible to ship that part to you.  Keeping an equipment of this vintage takes a community, so I'll be happy to offer whatever I can.  I'm not really thinking of charging beyond shipping and enough for my supplies, so please - let me know if I can help.  If you want that transformer, however, you are on your own.   

One of my monitors is stupid sharp.  That one is labeled "OPT 462 Impulse Bandwidth Spectrum Analyzer" where others are labeled "Spectrum Analyzer Display".  I was going to investigate this but I haven't yet.  I *think* it has something to do with EMI testing.

I'm thinking your 8568B is quite pleased he/she is in such a demand.  I have an equipment like that but mine is quite a basic stuff.  Analog VOM.  I bought it when I was in 7th grade.  I still have it.  It is still here despite at one occasion, a user (ME) almost died working with it in tube equipment.  When I worked in Japan as a tech, the company borrowed an Anritsu SP+TG unit from one of its clients.  I don't recall the model.  Royal treatment is right.  It was quite needed but the company didn't have enough to buy one of its own.

That got me wanting one and I finally have one.  (a FAMILY of HPs!)


I also have an HP8591EM.  I've been wanting to do a comparison with HP8568B.  I have a pair that's calibrated, so that would be the one to use.  I'm kind of liking 8568B better.  I quite don't like those push here, cycle through and push this and that - kind of UI. 

Maybe you know this one.  I took out two kinds of OCXO.  One is by OVENAIRE and model OSC 49-61C.  The other is HP10811 OSC series 3010.  It also has a sticker that reads "UPGRADED TO SERIES 3010".  I wonder what this mean?  Simon who has a blog says they are about equal.  But seems "upgraded" to certain series is something special.  By the way, former is damaged.  So I'm about to open it up and see what's going on.  Something is lose inside and I can hear it by shaking it.

[cncjerry]

I had one with just the grid fuzzy and I was able to tweak the adjustments to make it much better, almost perfect.  I started with G0hzu's original post and kept playing around until I got it.   It had something to do with lowering the front brightness control before tweaking some of the pots but I cant remember which ones.  It ended up the best of my three.

Jerry

[tkamiya]

I've tested like 6 of my boxes.  Out of it, 2 shows 10.7Mhz signal.  One was 52db below fundamental, and another was 62db below fundamental.

[KE5FX]

I've tested like 6 of my boxes.  Out of it, 2 shows 10.7Mhz signal.  One was 52db below fundamental, and another was 62db below fundamental.

Any correlation with the serial # prefix?

[G0HZU]

FWIW I studied the design of the HP8568B analyser back in about 1990 as I was involved in the design of various small high performance RF converters that spanned up into the GHz region. I was studying the art of frequency planning and also filter design and synthesiser design back then. So I looked at various receiver designs including several spectrum analysers. The HP8568B was one of them and we had a new one at work to play with. I do remember that the 2nd IF in the HP8568B is surprisingly cheap and basic in design and the bandpass filter is a mix of handwound coils and wide range trimmer caps and gimmick caps on a PCB. From memory, the pass/fail spec on this filter in the service info is very wide so my guess is that there could be a wide variation in how selective the 2nd IF filter is. Because the spurious term is a 2nd order term, every dB of extra rejection in the 2nd IF filter means a possible 2dB improvement in the suppression of this spurious term. It might be worth comparing how the hand wound inductors look when you look at a good example and one with a poor spurious response because the coil dimensions will affect the IF2 filter bandwidth and stopband performance slightly.
I think that even a subtle change in dimensions/inductance will affect the selectivity once aligned with the trimmer caps.

However, there were also some odd design choices to do with the 3rd mixer and its post mixer amplifier and I never really understood why they did some things the way they did. So something might be marginal here too. I never asked HP about any of this at the time and it will be way too late now. It was probably too late to ask way back in 1990. But if it is to do with IF2 selectivity, this would mean the analyser was always like this from new and -52dBc does seem quite grim. So there might also be an ageing issue here where something (3rd mixer balance? 3rd LO harmonic level?) causes the spurious to slowly get worse over time?

[tkamiya]

I'm sorry, I don't have serial number/brightness information.  I didn't keep any record.

I found something interesting the other day in service manual.  Apparently, trace and numbers are written by one circuit with it's own focus control, and graticule is done by something else. There is a calibration process for relative brightness and absolute focus for each.  I'll have re-read it to fully understand this.  What an odd way to do this.