Tempest Picture Centering/Size

You normally shouldn't have exactly zero DC. There should be some.

And the AC should be higher than a volt. You can doublecheck in attract more, but unless your XY size pots are dialed all the way down, it sounds like something isn't right.
 
You normally shouldn't have exactly zero DC. There should be some.

And the AC should be higher than a volt. You can doublecheck in attract more, but unless your XY size pots are dialed all the way down, it sounds like something isn't right.
Yep. I suspect the pots at least as a starting point. Will have a closer look hopefully this weekend.
 
Okay, first look. Pots all test out as functional, and none of them are visually separated from the board. The forward 6 pots (everything but the X and YBIPs) appear to have been replaced in the past. Only visible damage is the 1K R160 which has either been clipped intentionally or came loose at some point in the life of this board. There's a scorch on the underside of the board at VR1, but it looks like it has been replaced; it appears functional via an in-circuit DMM test.

IMG_4897.JPG

IMG_4898.JPG

IMG_4899.JPG
 
Regarding R160, you see that occasionally on Atari boards. These boards were wave-soldered, meaning they didn't solder each component individually. The boards are populated with parts, and then passed over a table of molten solder, and the solder sticks to everything that isn't covered with mask.

What happens sometimes is parts will fall off during that process, or get knocked loose, so they don't seat completely. And unless it was a part that was critical to the operation of the board (i.e., that would cause the board to not work), there's a good chance it would not have been caught by the QA inspection, as I think they just plugged these in and made sure they played ok. Their QA was not that thorough.

That resistor was probably barely connected with just the solder holding it down, and came loose. You want to resolder it back securely in the hole. It may not fix your issue, but it's worth doing anyway.

Everything else looks normal.
 
Repaired the resistor. Spot killer still on.

YOUT 1.8V AC, -1.5V DC
XOUT 1.3V AC, 1.2V DC

All test mode.
 
You can use XCTR and YCTR to adjust those DC values down closer to zero. But you may have something else wrong with the board. Measuring the AC/DC only checks to make sure you're safe to connect to the monitor. But to really see if you're outputting valid video signals, you need a scope.

You might also have separate monitor issues, too. You can get monitor issues any time there is a game board issue, so never assume your monitor is still working. Always go through my guide to test the transistors, etc, to sanity check yourself (or plug in a known-good board).
 
Q102 is for the LV power supply. That could be due to many things.

However frame transistors don't usually go bad on their own. They are almost always a symptom of something else being wrong. Your LV board might also be bad, even if it visually looks ok.
 
Okay. Get ready KLOV, here comes a heaping helping of ignorance on display. You've been warned.

I've pulled the deflection board for a visual inspection and spot check of components. No obvious burned parts. All fuses, R808 and R809 check out correctly. Looking at the schematic (standard disclaimer, I am not an EE, so take any electrical analysis on my part with the contents of the Great Salt Lake), Q103 feeds into the low voltage subsection, which is no longer present; this board has an Arcadeshop LV-2K that was installed (not by me, by somebody who knew what he was doing) last summer, along with new caps and reflow on all connectors (also done on the HV board).

Both LV-2K LEDs are green when the monitor is powered up. So current status is, spot killer on, no picture, game plays blind, XOUT and YOUT AC and DC voltages are nominal (the readings I was getting before were thanks to loose lead connections at my DMM--d'oh) and a bad Q103. That in itself would set off the Spot Killer, correct? Could an over voltage from the game board when I was flailing around with the pots cause that particular failure? What's the failure analysis path with an LV "fixer" board installed?
 
Q102 and Q103 are still part of the LV subsection. The LV subsection isn't completely replaced by the LV2000 board, and Q102 and Q103 are still active parts of it.

You either have a bad LV board, and/or something else is wrong on the deflection board, causing Q103 to fry.
 
Replaced Q103, pulled the Arcadeshop LV and replaced it with one of Jeff's LV2000s. Much OCD futzing to check voltages and connections (following The Guide), and... success. Spot killer is off, picture is back. Size is way off as might be expected, but with a thunderstorm rolling in that tweaking will have to wait until tomorrow.
 
Last follow-up: settings dialed in, fully functional. The vector weirdness I was seeing before the monitor went out is gone, could well have been due to the (presumed) failing LV2K and/or transistor.

First time I've ever repaired a monitor on my own. Insert joke about old dogs and new tricks at your leisure...
 
Last follow-up: settings dialed in, fully functional. The vector weirdness I was seeing before the monitor went out is gone, could well have been due to the (presumed) failing LV2K and/or transistor.

First time I've ever repaired a monitor on my own. Insert joke about old dogs and new tricks at your leisure...
Congrats. Gotta love the feeling of accomplishment.

As for old dogs... We're never to old to learn.
 
@andrewb can you confirm some transistor theory from what I believe is an accurate recollection. I can't find my Veatch book. Or a replacement yet.

The voltages on the X and Y outputs: the DC voltage is present to keep the transistor biased on while the AC voltage, riding on the DC, generates the signal.

To high a DC could send the transistor into clipping, as the DC bias voltage is actually 0 as far as the AC signal is concerned.

I think I blew the dust off of this accurately.

Thanks


Edit: OP, sorry for the thread jack
 
@andrewb can you confirm some transistor theory from what I believe is an accurate recollection. I can't find my Veatch book. Or a replacement yet.

The voltages on the X and Y outputs: the DC voltage is present to keep the transistor biased on while the AC voltage, riding on the DC, generates the signal.

To high a DC could send the transistor into clipping, as the DC bias voltage is actually 0 as far as the AC signal is concerned.

I think I blew the dust off of this accurately.

Thanks

You're remembering your BJT transistor theory correctly. However that isn't what's going on here.

The X and Y outputs are drawing vectors to the screen/tube. And those vectors all center around the center point of the screen, which represents zero beam deflection. If the monitor was drawing a straight vertical (or horizontal) line, centered at the center of the tube, then the signal would be pure AC, as there's no offset (i.e., the vector is perfectly balanced around the center of the tube.)

However, in a real example, the monitor is drawing whatever image is happening at the moment, which is a complex combination of vectors. And because those vectors are not all perfectly centered and balanced around the center point, there is some net DC voltage at any given time. This is natural, and just comes out of whatever is being drawn at that given moment, and how balanced it is around the center. That's why the worst point in the Tempest attract cycle is when it's drawing the scrolling TEMPEST logo. During that sequence, the monitor is basically drawing tons of vectors, but ONLY on the top half of the screen. So you see the Y DC value shoot up to about 3-4 volts during that part of the attract mode.

The other part of the offset comes from the adjustments. You can control the X or Y DC value somewhat by adjusting the XCTR and YCTR pots. This will shift the image up/down or left/right, forcing an offset. (And if you force both pots to the extremes, you can blow your monitor just with the adjustments, which is why you want to always measure the XY outputs before connecting the monitor.) But basically any time you deflect the beam away from the center of the screen, it takes power. And the further you bend the beam away, and the longer you hold it there, the more power it requires.

So if there is a problem with the adjustment pots or circuitry on the game board, OR if the game board is malfunctioning in a way where it isn't drawing the normal vectors (and instead is stuck at one of the maxed out X or Y levels), that will force the beam off the edge of the tube permanently, which will blow the deflection system.

The deflection circuits are designed to be able to deflect the beam to the edge of the screen, but only for a relatively short time (e.g., long enough to draw a vector, which is a very, very small fraction of a second). If you try to bend the beam to the edge and hold it there permanently, that takes too much current, and the deflection transistors burn up in seconds (usually causing a string of other things to blow along with them, like the LV section, and potentially the HV as well, which can fail when the LV fails.)
 
Thanks for that. It's gonna take a bit to digest completely, but if I'm grasping this correctly; it almost sounds as if the DC could infer relative position on the screen. I'd think it would have to travel into negative to get to the other half of the screen on whichever axis. Or does 0 vdc represent the opposite extreme of deflection?
 
Back
Top Bottom