Author Topic: HP 8568A Spectrum Analyzer display fault REPAIRED ? (Read 10774 times)

Simon Spiers · « **on:** October 18, 2016, 03:16:06 pm »

Hi all, wonder if you can help>
I have just moved my 8568a from one side of the workshop to the other and now it has a problem.
When the unit has been on for a while the CRT starts to show scrambled characters and the graticule seems to vanish.
I have been over both units cleaning edge connectors and checking ESR of the tantalum caps but still no joy.
All DC rails are in reg and smooth.
It also shows the check instrument 1 led one this fault starts. Which is instrument 1?
The actual waveform displayed looks un corrupted, so I am guessing the fault must lie in the display units character generation section?

https://photos.google.com/share/AF1QipPwu8Yr7UkM9UGuLLGe5SrZ2RFt5LeYJk4sdtKy8xAPbNNcVCMSXMjlDGU7yTtJ7g/photo/AF1QipMAmmWk_jaYrVG7oUhUBULcp1q0FOQ_iwrI_vI8?key=TUxDM0VpWEUxdXgxVy12V1VBMThLbGc0QU9HeHFR
Thanks in advance. Simon

Simon

tautech · « **Reply #1 on:** October 18, 2016, 07:12:59 pm »

Quote from: Simon Spiers on October 18, 2016, 03:16:06 pm

Hi all, wonder if you can help>
I have just moved my 8568a from one side of the workshop to the other and now it has a problem.
When the unit has been on for a while the CRT starts to show scrambled characters and the graticule seems to vanish.
I have been over both units cleaning edge connectors and checking ESR of the tantalum caps but still no joy.
All DC rails are in reg and smooth.
It also shows the check instrument 1 led one this fault starts. Which is instrument 1?
The actual waveform displayed looks un corrupted, so I am guessing the fault must lie in the display units character generation section?

https://photos.google.com/share/AF1QipPwu8Yr7UkM9UGuLLGe5SrZ2RFt5LeYJk4sdtKy8xAPbNNcVCMSXMjlDGU7yTtJ7g/photo/AF1QipMAmmWk_jaYrVG7oUhUBULcp1q0FOQ_iwrI_vI8?key=TUxDM0VpWEUxdXgxVy12V1VBMThLbGc0QU9HeHFR
Thanks in advance. Simon

Simon

You'd think heat related, well I would. Do all rails maintain acceptable voltage and ripple levels after it's been on for a similar time ?
I guess with the covers removed there's going to be better cooling so things that come to my mind to identify where the fault might be are a thermal camera or introduction of a heat gun to get the fault to show itself.

Simon Spiers · « **Reply #2 on:** October 18, 2016, 07:51:32 pm »

Hi thanks for the repl .
Yes agreed it looks like a thermal related problem as I'm watching it now in a cold workshop and all is well.
All the rails stay stable during fault conditions but I am now monitoring an aux supply on the A3A4 board fed from a linear regulator. This feeds the memory circuits on that board. That looks fine for now but time will tell.

Richard Head · « **Reply #3 on:** October 18, 2016, 08:28:35 pm »

You may have pushed the instrument right up against the wall and blocked or at least limited the flow rate of the fan. Is the fan filter clean?

Simon Spiers · « **Reply #4 on:** October 18, 2016, 08:51:43 pm »

Hi Richard yes the filters clean. Its really annoying that it has performed without fault this evening.
Ihave been removing caps to check but other than that nothing else. I'm sure its memory related as the service manual sugguests. But why it's ok now is anyone's guess.

tautech · « **Reply #5 on:** October 18, 2016, 09:02:32 pm »

Then you might suspect any inter PCB connectors or even any socketed IC's.
Re-socket everything ?

siggi · « **Reply #6 on:** October 18, 2016, 10:05:44 pm »

Quote from: Simon Spiers on October 18, 2016, 03:16:06 pm

The actual waveform displayed looks un corrupted, so I am guessing the fault must lie in the display units character generation section?

I don't think there's such a thing as "character generation" in the 85662A. I believe the entire on-screen display is written to a display instruction list, which includes the individual strokes for everything - including each character (though there may be indirection to re-use characters) ... time passes ... and I'm wrong. There is a ROM containing character stroke memory.
From your video, it looks like the graticule display is wonky, but the character display is OK? If so, then you probably have a bad display memory RAM chip. Unfortunately my service manual doesn't have the A3A4 schematic - for some reason. IIRC there's a description of how the video memory is used and interpreted in the programming manual - and it's not as plain as my nose.

In any case get the service manual (I have the Artek Manuals scan for HP 8566B). Page 3-1 says this:

Quote

INSTRUMENT CHECK LED: The left INSTR CHECK LED (CHECK LED I) remaining on after
Instrument Preset is pushed indicates that a bad bit was detected when the analyzer performed its self
check routine. This check routine writes and reads to the read-write memory (RAM) in Digital Storage. To
activate this test, press Instrument Preset.

It then goes on to e.g.

Quote

DIGITAL STORAGE TEST PATTERN: The test pattern shown in Figure 1 can be used to verify the
existence of a failure. This pattern can be generated independently of the main processor in the RF Section
(A15 Controller). To enable the test pattern:
a. Jumper A3A6TP3 and A3A6TP6. (ThIns on the test ROMs.)
b. Push A3A7S1 to initialize.

simmconn · « **Reply #7 on:** October 19, 2016, 01:23:25 am »

The fact that the annotations and graticules are screwed up but characters are still legible indicates the vector drawing engine and everything after it is most likely okay. It looks like the 3rd page (address 2048 and up) of display memory has problem. The hp logo is still in the correct location and look okay means at least part of the memory in the 3rd page (address 2168 thru 2175) is working. It is possible to track down the memory IC that went bad, but replacing the A3A4 assembly would be much easier. There should be quite a few of them out there that have been swapped out during the LCD display retrofit.

Simon Spiers · « **Reply #8 on:** October 19, 2016, 01:22:48 pm »

Hi Simon thanks very much. The manuals have had me looking around a3a4 board area and it seems to have corrected the fault once I removed a few tant capacitors across a 12v regulator on the board. I am wondering if these wet tantalums are suffering silver migration? Anyway since tinkering with this board the display appears fine?
So at present it seems like a possible shorting capacitor across a TO220 style regulator may have been the fault, putting the device into current limit maybe?

Simon Spiers · « **Reply #9 on:** October 19, 2016, 04:34:26 pm »

Sigi, thanks for looking that up. I have got the manuals and will have a read if the fault returns. At least i'm in on the correct board.

Simon Spiers · « **Reply #10 on:** October 29, 2016, 08:39:43 am »

I have to report the fault has returned and this time it looks permanent.
I have also noticed that recalling memory 8 or 9 that i believe are in ROM the waveform vanishes and there is just a line of garbled text on the screen.
I am starting to fear the Eproms are corrupt now.
Will remove the cover and have a poke about but access is really difficult without riser boards.
This is now what i'm getting

MadTux · « **Reply #11 on:** October 29, 2016, 05:31:45 pm »

These symptoms rather look like corrupted EPROMs in the RF section.
85662A only has 2kB of mask programmed character ROM, which seems to be fine in your 85662A
8568A has 2x 27C256 EPROM,
See if you can find them online, otherwise I'll rip open one of my 8568As and dump their EPROMs
BTW, 8568A is not compatible with 8568B, 8568A has funny custom HP processor while 8568B uses 68k CPU.

Simon Spiers · « **Reply #12 on:** October 29, 2016, 05:46:10 pm »

Hi Madtux thanks for replyin .
I am currently looking at the a3a3 board thats the line generator.
Even though the displayseems the problem , recalling 8 or 9 memory seems to brick the while thing.
Is this why you are suspicious of the rom? It would make senceto my limited understanding of this machine.
I will have another look for a rom iso but last time had no luck.
Cheers again.

MadTux · « **Reply #13 on:** October 29, 2016, 06:01:27 pm »

Display ROM has character X/Y coordinates which the display section writes onto screen, just like in the Tek7000 series.
If that would be broken, you would get garbled characters "chinese" on screen. But since the shape of the characters is fine (nonsense with good characters), I would strongly suggest that A3A4 board is fine. So is A3A3, which makes characters from coordinates stored in A3A4 ROM.
BTW, there is a test button on 85662A to test just that, RTFM.

I'm quite sure the problem is somewhere in the 8568A digital part, most likely ROM or bad PSU caps.

Simon Spiers · « **Reply #14 on:** October 29, 2016, 06:36:02 pm »

I will take a look at the manual regarding the testing display but have just found that recalling setup 8 from memory locks the wwhole thing up.
Recalling 9 which is used to to set zero span seens to work fine despite not loading the correct parameters ast week.
So i need new eprom image then?
I really don't like digital technology. ...

MadTux · « **Reply #15 on:** October 29, 2016, 06:45:26 pm »

I just had a look inside my 8568A. And there are like 16 funny HP (Mask)-ROMs on the ROM board. Not even socketed, so most likely mask programmed. Since they aren't socketed, I can't remove them easily, which means I can't read them on my Tl866. Sorry

Maybe you could check whether you perhaps have HP 8568AB, which is 8568A with 68k CPU retrofit and new EPROM, which basically makes it a 8568B.

If you want to test your 85662A in standalone mode, make jumpers between:
A1A8TP1 and A1A8TP2 => turns PSU on
A3A6TP3 and A3A6TP6 => next to each other, starts self test mode
Then insert mains cable and press reset button: A3A7S1

This is how it should look like, just tested on mine:

Simon Spiers · « **Reply #16 on:** October 29, 2016, 07:25:47 pm »

Just had a look inside mine. It has the dedicated hp chip by the looks of things and the same unsocketed Roms as yours i think.
Looks like its off to the scrap for this old girl.

MadTux · « **Reply #17 on:** October 29, 2016, 07:27:58 pm »

Please don't scrap it

I'd take it for shipping + small cost and try my luck.
Or try to find a 8568B for parts somewhere, they sometimes can be found quite cheaply

Simon Spiers · « **Reply #18 on:** October 29, 2016, 07:52:45 pm »

I won't its just ffrustration on my part.
I will look out for spares and hhopefullyfind some spares.
I need to just be dure where the problem lies for sure.
But everything points to a corrupted Rom.

MadTux · « **Reply #19 on:** October 29, 2016, 09:20:52 pm »

If you donate me this sucker:
http://www.ebay.com/itm/AMERICAN-RELIANCE-9001-28-DIP-IC-28-PIN-TEST-CLIP-MODIFIED-FOR-0-300-ICs-/221296011486?hash=item33864528de:g:6oIAAOxygLxSVYXj

I'll try whether I can read these chips "in circuit" as 2716 EPROMs.

You would then need to stitch 2x8 of these 2kB images together in right order and burn them into two 27C128, each of them gives 8bit data on 16bit data bus. Grab the address bus before it gets multiplexed and wire it directly onto both 27C128s. LROMEM most probably goes to /OE, perhaps needs to be inverted before.

Forgot to mention, have you already tried to resocket and clean that 9825 CPU and it's socket?

Simon Spiers · « **Reply #20 on:** October 30, 2016, 06:44:11 pm »

Madtux i have just tried the stand alone test on the display unit and get nothing on the display?
i followed your instructions verbatim .

MadTux · « **Reply #21 on:** October 30, 2016, 11:34:18 pm »

That's weird.

Perhaps buy another, preferably working, complete 8568 system or find somebody in UK who has a complete, working 8568 and start swapping cards. At least that's what I usually do before I get completely desperate on something like that. Because HP instruments are usually build really modular, you can start off very coarse (swap display/RF sections) and narrow the fault down to individual cards. Then you can use good and bad card to compare signals on individual parts. Very nice/convinient way to find faults without needing to understand everything.

Simon Spiers · « **Reply #22 on:** November 17, 2016, 09:43:54 pm »

Just to tie up loss ends i have repaired it.
The fault was the s3a4 momory board as some of you suggested.
Thanks to a very kind ex HP tec from the States who sent me a completeset of boards foc!
Thank you all for your help suggestions and insight .


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: HP 8568A Spectrum Analyzer display fault REPAIRED ? (Read 10774 times)

Share me