Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
7912AD Fails start-up self-test (perhaps)
|
So as of a few hours ago I became the proud owner of a 7912AD with a matched pair of 624 & 634 monitors.
The inevitable puff of smoke led me to a hot 10R resistor and thus the equally inevitable shorted blue tant (why is it nearly always the blue ones that die but only rarely the yellow ones?) on the +15V rail on A38 Scan board. Once that was fixed things seemed ok power-wise, or at least all voltages looked ok on the PSU test points, the fan runs up but we don't get much further. I believe I should see the 'Remote' and 'Digital' lights come on for two seconds followed by the 'TV' and 'Local' lights coming on but all I see is two brief flashes of the 'Reduce Intensity' LED, a short pause, a third flash then nothing. My question is: Does the self-test actually flash that LED (it is not mentioned in the manual as far as I can see) or does the fact that it comes on at all mean that the system is sensing a high intensity somehow and aborting the start-up? - and yes, the two intensity pots are set at minimum. Right now I'm thinking it could be the A10 front panel and A22 'Translator' boards need looking at ..... but I'm still wading through 400+pages of manual...... Adrian |
|
Adrian,
The 'reduce intensity' light can be triggered either by current in the write gun measured via the A1 electrode (diagram <7>, A40 board, signal is labelled 'Protect Enable' and drives 'reduce Intensity' via U114 on diagram <12>) or by the CPU via U830 on diagram <13>. In fixing my 7912AD I was fooled by the Reduce Intensity light making me think there was write current when in fact the 10kV anode voltage to the write gun had failed. The symptoms you describe suggest that the 6800 MPU is not working properly and if the DC level of +5.1V is correct then check also for ripple. The installation of the many plugin boards into the backplanes makes for difficult testing at the board level without extenders (I have never seen any advertised). I would be tempted to solder in a few flying leads so that you can check CPU clock and busses (and reset) and work from there. There is a handy trick to check if the read gun and video circuitry are OK - it can be found under the calibration section of the manual. Reverse the current through the focussing coil of the read gun and you should get a bright ring around the edge of the display. This is independent of anything written to the target. Regards, Roger |
|
On Sun, Jul 1, 2018 at 02:46 pm, Adrian wrote:
So as of a few hours ago I became the proud owner of a 7912AD with a matchedCongratulations and welcome to the club. I believe I should see the 'Remote' and 'Digital' lights come on for twoI've just made a couple of short movies of my 7912AD waking from its slumber. The lights change as you describe and the 'reduce intensity' LED does flash in the pattern you mention. First a view of the whole thing (sorry about the horrid little LCD monitor, but it's all that will fit on the bench): Then a closeup of the buttons and lights: In my experience failure to start up has most usually been caused by missing/incorrect power supply rails. There's a circuit which monitors most of them and holds the 6800 CPU in reset until they're right. Bringing a wire out from the 6800's reset pin and watching what happens to it might be interesting. If the self-test is failing for other reasons (RAM or ROM test error) then that gets shown on the TV's readout. The last time my 7912AD stopped working, it was because there was a poor contact on a harmonica connector on the PSU regulator board which fed -2V to the power distribution board. That was enough to stop play. Chris |
|
Hi folks and thanks to all for the input, the video was really reassuring as the LED activity matched what I'm seeing. It also might tie in with the service guide description of what's going on in firmware during the power-up sequence where "one of the first things to happen is the Main and Graticule intensity levels are cleared (two flashes?) then writes the value at the address (one more flash?)"
toggle quoted message
Show quoted text
So reassured the processor was doing something at least I started thinking what else could stop the buttons lighting up? Then came the light bulb moment (so to speak) there are lamps, not LEDs in them thar switches, you don't suppose....? (embarrassed cough) oh yes, the filaments of the 'TV' and 'Local' lamps both open circuit although the others were fine. Changed them out, reassembled the panel and ALL the lights now worked in the expected sequence! I haven't traced the connections yet but either the processor is reading back that the lamps are good - unlikely that we have aerospace type BIT in this box? - or taking things apart and putting them back got the other button-lights going somehow? Very odd but set that aside (for now, but I know it's going to nag at me in the still dark watches of the night). So on to step 2. I fired up the 634 video monitor and yup a burning smell and yup another blue tant gone! Do they really hate me that much?? However after fixing that I now see a raster and the readouts from the plugins - 7A29 and 7B90P - which is some progress.? Advancing the Graticule and Main intensities produces screen activity but very much rolling and out of sync in X & Y (although the readout displays are rock solid) so I guess I now need to read different bits of manual but any pointers will be very welcome! Dare I turn on the 624 X-Y display and look at the digital images? Probably, but I'm checking ohms on the power rails first this time! I'll keep you posted and thanks again, Adrian On 7/2/2018 12:01 PM, cmjones01 wrote:
I've just made a couple of short movies of my 7912AD waking from its slumber. The lights change as you describe and the 'reduce intensity' LED does flash in the pattern you mention. |
|
Adrian,
There is a jumper on the A28 board for 525/625 operation of the composite video but it should affect the 7000 plugin readout (CPU generated) and the read gun scanning consistently. There are multiple jumpers for normal or fast scan and these need to be consistent. Pages 4-9 onwards of service manual 2 give all the jumper settings. If my memory is correct the digital display does not show the readout from the plugins. Regards, Roger |
|
On Mon, Jul 2, 2018 at 11:32 am, Adrian wrote:
Then came the light bulb moment (so to speak) there are lamps, not LEDsGreat! Even if we don't quite know why, that's still progress. So on to step 2. I fired up the 634 video monitor and yup a burningMine showed similar symptoms. The readout was clear, but the 'analogue' image was just a flickery, jumpy mess. The problem turned out to be a shorted tant (surprise) on the -15V rail on the A38 scan output board which was preventing the read scan of the target working properly. The result was that the video was showing a highly magnified image of the centre line of the target. You can infer whether a lot of things are working by playing with the intensity controls. With the timebase free-running or in AUTO mode (otherwise nothing, not even the graticule, gets written on to the target) try turning up the graticule intensity. Does the screen brighten? Try also turning up the trace intensity. Near the top of its range the 'reduce intensity' LED should come on. Before that the trace should bloom dramatically. Leave the intensity set to a level below which the 'reduce intensity' LED lights and try twiddling the vertical position through its range. You may find there's a position at which the trace passes through whatever read scan is going on, and the video output goes all white. It helps a lot to visualise the write scan and read scan when trying to reason about what you're seeing. It's normal for the video output to be a bit flickery when the timebase is auto-triggering. Because the graticule only gets written when the timebase triggers, and the timebase's auto-trigger is not synchronised with the video output, there's a constant tussle between the two timings. Give the timebase a trigger signal and it all gets much better. Dare I turn on the 624 X-Y display and look at the digital images?Give it a try. Remember that you'll only get X-Y output when you press the 'DIGITAL' button and the timebase has triggered. The X-Y output will show the trace frozen. It will only update with a new trace on each press of the 'DIGITAL' button. The setting of the intensity control that gives good results in digital mode is likely to be somewhat different from the optimum setting in TV mode. Chris |
|
On Tue, Jul 3, 2018 at 12:12 am, Roger Evans wrote:
There is a jumper on the A28 board for 525/625 operation of the compositeIf the 7912AD is a 525-line model and you set the jumper on the A28 board for 625 lines, the readout gets repeated at the bottom of the TV display because the readout board's counters aren't set up to deal with the extra lines. There exists a modified readout board (I've got one here as a spare) which has a load of extra logic bodged on to it dead-bug style to handle the readout properly in 625 line mode. If my memory is correct the digital displayYou mean the X-Y output or the 'digital' video output? I think the latter does show the readout, but the X-Y output certainly doesn't. Chris |
|
Hi Chris & Roger
toggle quoted message
Show quoted text
Yes, "jumpy mess" about describes what I see too! Stills don't do it justice really, but you can see images here (ignore the apparent lack of focus, that's shutter speed combined with my 'three-score years and ten' hands!): /g/TekScopes/album?id=62098 As you can see from the pix, I just got the XY display running and while the display is odd in that it seems to be 'seeing double' (PSU ripple somewhere?) it is stable at least. Off to investigate A38 (already changed one tant) power lines... I had checked jumpers as I looked at various boards but you are right, I should go through methodically, with the manual, and make sure I didn't miss one. The link on A28 is set at 625 lines, changing it to 512 shrank the image vertically as you would expect but made no other difference. One difference between the manual I have (Artek) and this unit is the A54 MPU, mine is a single board and there is no A52 MPU memory board fitted (or location for it that I can see). I've uploaded a pic of that as well. On 7/3/2018 9:38 AM, cmjones01 wrote:
Mine showed similar symptoms. The readout was clear, but the 'analogue' image was just a flickery, jumpy mess. The problem turned out to be a shorted tant (surprise) on the -15V rail on the A38 scan output board |
|
On Tue, Jul 3, 2018 at 02:29 am, Adrian wrote:
Yes, "jumpy mess" about describes what I see too! Stills don't do itThe graticule display on the X-Y output looks good to me. Mine's exactly the same. You can reduce the size of the little circles by turning down the graticule intensity a whisker. The scan image processing produces one dot where each bright area of the trace starts, and another where it finishes (sort of like the modulus of the differential of the image), so you tend to see the graticule dots and the trace doubled unless the intensity setting is exactly right. Apparently the GPIB data download looks for the midpoint between the dots to give the position of the trace, but I've never tried that. Off to investigate A38 (already changed one tant) power lines...It would probably make the lower readouts drop off the bottom of the screen, if you have a 625-line readout board and plugins that display anything in the lower readout fields. One difference between the manual I have (Artek) and this unit is theInteresting. I can't remember what mine looks like! Chris |
|
On Tue, Jul 3, 2018 at 02:29 am, Adrian wrote:
/g/TekScopes/album?id=62098Another thought: the trace output on the X-Y display might also be correct, but as if the intensity is set high so you're seeing the edges of a trace which has bloomed to fill half the screen. Do those two dominant horizontal lines move if you adjust the vertical position? Do they get closer together if you turn the intensity down? That's what I'd expect. Chris |
|
Adrian and Chris,
The CPU on the CPU board is labelled 6802, the earlier boards had a 6800 processor. My Artek manual shows the 6800 and 6802 interrupt circuitry separately but doesn't have the 6802 MPU board. Sorry for the confusion over the digital vs XY outputs. The analogue output is all analogue so you will see the trace bloom at high intensity, while the XY output is thresholded so you just see the edges as Chris says. The other major difference is that the analogue and XY scans are 'rotated' by 90 degrees, the analogue scan is just a TV scan while the XY scan is vertical so that the thresholding logic can detect edges for each horizontal (time) position. It can store several on-off pairs for each X position to cope with reading the graticule position. Since the graticule is just a sequence of spots from the write beam, and it looks perfect to Chris and me, there isn't much wrong in XY mode. Can you input a much simpler waveform, but triggered as Chris says, also use a faster timebase to reduce the effective intensity of the write beam. Regards, Roger The analogue image of the graticule appears rotated. Was that from an earlier sequence and reseating boards and cables has now changed things. I have used the GPIB interface quite a bit, mimicing the original intention of the 7912! There is a choice of reading the dot pairs or the mean value, also automatic centroiding of the graticule and other niceties. I didn't go as far as measuring the geometric distortion from the graticule and using that to correct signal readouts - that was part of the minicomputer software that went with the 7912s |
|
Roger
toggle quoted message
Show quoted text
I think the 6802/MPU info is? in the Change section at the end of the manual. Why Tek did not include an updated schematic I? cant say unless the schematic didnt change ? Dave manuals@... On 7/3/2018 6:56 AM, Roger Evans via Groups.Io wrote:
Adrian and Chris, --
Dave Manuals@... www.ArtekManuals.com |
|
Dave,
It IS all there in the schematics, I just managed to miss the crucial one! Schematic <14> is the updated 6802 processor board, it uses 4k EPROMS instead of 1k EPROMS and the SRAMs are also four times the density so the memory fits on the CPU board. Aside to Adrian and Chris: The right hand side of the TV image that Adrian uploaded shows what looks like the beginning of a 'sensible' trace at the top of the screen. That adds to the feeling that the 7912 is trying to display a very strange signal. When the display blooms due to charge spreading in the target (and some bending of the read beam) it rapidly loses any fine scale detail and I wouldn't expect it to look like Adrian's XY images even after thresholding. Roger |
|
Thanks Roger, I'll look at that, the 7A29/7B90P and generators (one of leo Bodnars fast pulsers and an HP 8657B) produce sensible looking traces in a 7854 but that doesn't mean they aren't getting mangled in the 7912!
toggle quoted message
Show quoted text
On 7/3/2018 3:44 PM, Roger Evans via Groups.Io wrote:
Dave, |
|
Adrian,
I don't know what the rep rate of your pulser is but if you aim for one or two cycles of a square wave or just two edges if the mark space ratio is very large, that should be an unambiguous trace. It looks like you have at present 1usec/div and many pulses per usec so a very detailed trace. Regards, Rpger |
|
Hmm, I think we are getting there! See 'Getting there'* in album:
/g/TekScopes/album?id=62098&p=Name,,,20,1,0,0 So what did I do? Well not a lot except move the stuff to a different bench and, more importantly I suspect, switched things on early today before the sun moved round - well, ok it stayed still(ish) but the planet rotated - and started warming up the barn. It ran happily for about 15 minutes then collapsed in a heap again, let it cool off for 5 minutes and it ran for a couple more then failed again. So something is getting hotter than it likes. There is no cooling in this thing other than in the PSU, I wonder if it was assumed it would be mounted in a blown rack? I've put it up on blocks with a desk fan blowing across it and we'll see if that helps but I'm going to swap out all the tant caps I can see on A38 and A28 for a start! * Sorry about the Non-Tek scopery lurking in the background - hope I don't get struck off! Thanks again for all the help and advice, Adrian |
|
Adrian,
My 7912AD runs happily with no extra cooling, I think the interior layout is designed to pull air past the major boards. I have had it running for a couple of hours with no problems. The graticule and main intensities change markedly in the first 10 minutes or so but this is mentioned in the manual. Most of your images that look 'wrong' have multiple spots rather than continuous traces and I notice you still have the graticule turned on even though the brightness is fairly minimal. One image shows a graticule with rotated and misplaced spots. Try turning the graticule intensity down to minimum, that should turn the graticule generator off and remove one of the possible culprits. When it is working it looks really good! Roger |
|
Craig Sawyers
Here's what may seem an odd question - does the rear fan pull air in, or push it out?
toggle quoted message
Show quoted text
Craig My 7912AD runs happily with no extra cooling, I think the interior layout is designed to pull air |
|
Hi Craig,
toggle quoted message
Show quoted text
That's not an odd question at all, it's one that I should have asked myself before thinking about blown racks! The fan pulls air in (quite a lot of it), through the PSU module which then exhausts through a wide horizontal slot low down on the module front face. From there it effectively blows between a gap under the 'widthways' cardframe carrying the PSU interface and MPU cards and thence into the 'open' end of the 'longways' cardframe carrying the other seven cards among which, perhaps significantly, is the video processor and scan control card. The chassis vents at the top, front half of each side panel so the airflow is along the cardframe backplanes, up between the cards and across the top front then out the sides. So by running with no top panel I may well have short circuited the flow by allowing most of the flow to escape to atmosphere from the top of the first frame and prevented any effective cooling of the second? I will replace the panel, test and report - I was contemplating doing a temperature test under controlled conditions but turns out this beast is too long for my environmental chamber! Adrian On 7/4/2018 5:26 PM, Craig Sawyers wrote:
Here's what may seem an odd question - does the rear fan pull air in, or push it out? |
|
I remember seeing 7912ADs and HBs in final test and Cal. They would use
toggle quoted message
Show quoted text
them lead that had holes drilled in it for all the adjustments and test points. today at wouldn't be hard to use a piece of luck Sanders just a sheet of scrap aluminum. But what the 7912 you do want to make sure there's air moving across everything On Wed, Jul 4, 2018, 10:44 AM Adrian <Adrian@...> wrote:
Hi Craig, |
to navigate to use esc to dismiss