Star Wars Amplifone deflection mod and issues

Brad303

Member
Joined
Jun 19, 2024
Messages
58
Reaction score
51
Does anyone recognize the hack/mod someone made to this Amplifone deflection board? I know there are some recommended upgrades, but this doesn't look like anything I've read about.

I'm considering taking it back to stock, but figured I'd check with the experts, first.

The image definitely has issues, but I haven't tested anything yet (sorry for the lousy pic - will get a better vid later today):
 

Attachments

  • 2025-02-14-14-31-24-446.jpg
    2025-02-14-14-31-24-446.jpg
    368.4 KB · Views: 61
  • 2025-02-14-14-31-33-729.jpg
    2025-02-14-14-31-33-729.jpg
    539.4 KB · Views: 58
  • 2025-02-14-14-31-47-409.jpg
    2025-02-14-14-31-47-409.jpg
    526.6 KB · Views: 60
  • 1739736899789.png
    1739736899789.png
    1.7 MB · Views: 62
What HudsonArcade said.

Remove all of it. None of the mods for these boards are actually useful or worth doing. Just return the board to stock, rebuild it properly, add some extra solder to all of the header pin joints, and you'll be fine.
 
I was going to do that, but then I came across the Atari Tech Tip that said to add some diodes. I haven't yet figured out if that's a version of what was done here, but you'd suggest avoiding any mods altogether?

Any chance that mod is causing the problem with my video output? It would appear that the problem only occurs after the system is restarted after being on for at least 15 mins.
 
Ignore the tech tip. Atari was not always right. And yes, don't do any mods.

I repair these boards, and have done tons of them for people here, over many years. I've looked into and experimented with all of these mods, and none of them are necessary or beneficial.

And yes, it's possible for mods to cause problems. (Though not the only possibility in your case, but it's not impossible.)

What problem are you having?
 
Ignore the tech tip. Atari was not always right. And yes, don't do any mods.

I repair these boards, and have done tons of them for people here, over many years. I've looked into and experimented with all of these mods, and none of them are necessary or beneficial.

Your expertise is legendary, so I've no problem freely accepting and following your advice, my friend.

And yes, it's possible for mods to cause problems. (Though not the only possibility in your case, but it's not impossible.)

What problem are you having?

The last pic shows the issue, but here's a link to a 1m vid:
 
Last edited:
I should add that when the game is first turned on after being off for a day or so, it flickers a bit, but then seems to be relatively stable with a full screen image. After about 15 or so minutes of playing it, if I turn it off and then back on, I get what you see in the video.

Flickering seems to be the lower half of the image compressing to a single line, like the constant left half of the image being compressed as seen in the video.
 
First thing, you'd want to make sure the problem isn't being caused by the game board, since half-screen collapse can be either game board or monitor. Ideally you'd need a scope to do that. However you could also try measuring the DC voltages on the XOUT and YOUT test points on the game board, with a DMM. Normally the XOUT should be close to zero volts (or hovering around it). However if the left half of the image is missing, you're going to see XOUT be sitting a a couple of volts or more.

In this case, my guess is it looks a little more like a monitor issue. It could be a cracked solder joint on Q14 or Q15, or one of the other parts in that area. (Q11/12/13 as well).

The traces and pads are far too thin on these boards, and they also didn't use plated holes. That means there is no solder IN the holes, and it's only the blob on the bottom of the board that holds each joint together. That allows them to crack very easily, as the solder fatigues over time. That would also correlate with why the issue comes and goes as it warms up, as cracked joints can go open or closed as they heat up.

Inspect the deflection board under a bright light, and try to find one or more cracked joints. Ideally you can find it first, and not just reflow everything blindly. (Though if you can't visually find anything, you can always do that.) But it's better to find issues first, before trying to fix them.

Also, you REALLY don't want to be operating that monitor that way, as you can very easily end up blowing a lot more stuff, and have a completely non-working monitor, instead of one that is working with collapse.
 
Thanks, Andrew.

I don't have a scope.. yet. That's on my very short wish list, but I've got to find an old analog one that fits the bill.

I have a 100MS/s Digilent USB scope, if that'd work as a substitute. And a decent Brymen DVM if not.

Either way, I'll check out the solder joints under magnification and a bright light. Reflowing everything wouldn't be the end of the world, just time consuming. And if it restores this technological work of art, absolutely worth it!

I'm not an EE, but wish I'd gone that route instead of broadcast comms, so forgive my ignorance.

Can I test the existing deflection transistors in the circuit, or do I need to remove them?

I just ordered a dozen each of what looks to be Mot original 2N5686/2N5684, just to have them on hand, but I don't know if that's the problem yet or not. This board looks like it has mica pads under the transistors, so maybe they were replaced at some point before I got it.
 
You can use the scope you have, if it has two channels and XY mode (to generate an XY image). Otherwise you could use it in volts/time mode, but you'd need to be familiar with what the waveforms should look like ahead of time, e.g., by looking at some on a known-good board, to know if any signals you look at are bad or not.

Where did you buy the transistors? I don't recommend using alternate transistor substitutes in these. They are not 'upgrades', counter to what anyone might try to tell you. If the board is working properly, the original 2N3716/92's should work fine.

You can test most transistors in-circuit, using the 2-diode test. However it's not a silver bullet. Some of the transistors will measure low voltage drops, due to the other components around them. However if you have a good board to compare to, you can know which ones should act this way, and know what to expect.

But basically for almost any transistor in circuit, you should get two voltage drops in one direction (either to or from the base, depending on transistor type), and either zero or one drop in the opposite direction (one being the case when there are other components interfering with the measurement, and usually that drop will be low). But if you don't see that, or any pins are shorted to any others, you know the transistor is bad. However you should always remove and test out of circuit to be 100% sure.
 
I found a broken solder joint on C13, but reflowing didn't fix the problem. I'm in the middle of restoring it to stock.

I got the Mot transistors from 2 small shops that indicated they had new-old-stock. Date codes and appearance looks legit, but of course, there's always the chance of counterfeits, especially on old components. Finding genuine 6809s has been no small feat either, although I'm planning on using those to build a Vectrex.

FWIW, 2N5684/2N5686 are have the same Hfe & Ft as the originals, but have higher voltage rating (120V v 80V), currently handling (25A v 20A), and power dissipation (200W v 150W). I figured they would be "sturdier" in the case where another component is out of spec, and less likely to go kablooey.
 
FWIW, 2N5684/2N5686 are have the same Hfe & Ft as the originals, but have higher voltage rating (120V v 80V), currently handling (25A v 20A), and power dissipation (200W v 150W). I figured they would be "sturdier" in the case where another component is out of spec, and less likely to go kablooey.

What's the bandwidth, and how does it compare to a 3716/92?
 
What's the bandwidth, and how does it compare to a 3716/92?
My original research indicated the Ft on the 5684/86 was the same 4 MHz as the originals, but pulling some Mot/ON Semi datasheets shows it's actually 2 MHz. That still should be more than fast enough for the deflection board that works in KHz, but may generate more heat.

As it turns out, I pulled all the transistors on the heat sink (Q4-Q7 & Q14-Q17) , and they check out, so it looks like I have a couple of boxes of transistors I don't need (not the end of the world). Maybe once I get this thing up and running, for grins and giggles I'll swap out the originals for the new ones and see how they work.

Board is back to stock, and reflowed (not resoldered) all the joints, but I'm still having the issue.

I also nabbed a Vector Wars which exhibits the same problem, so that indicates it's in the monitor, not the AVG/game boards.

So what's next? Throwing parts (caps) at it?

I'd prefer to diagnose the issue, but not sure where to start. I'm assuming it'd be on the right side of the board, correct? I'm inferring that from the fact that the X Size pot (R76) is there.

Another thought I had: since C13 had a broken solder joint, but fixing that made no difference, might that indicate it's the culprit?

Thanks for all the help. I'm thoroughly enjoying this journey. (I wish said journey was shorter, in this case, but I'm enjoying the scenery along the way!)
 
Caps won't help you here. This is most likely a cracked trace or bad joint. Or a bad transistor.

Yes, it's going to be on the X side, which is the right half of the board (i.e., the side with Q11/Q12).

I'm still suspecting a bad connection between parts. Tone out all point-to-point connections between transistors. Probe from the transistor legs, for the ones where you can access them. Focus on the paths between Q11/12, Q14/15, and Q16/17.

I think Q14/16 control the -X direction, which is the half you are missing. Worst case, you could swap Q14 with Q4, if you don't have any spare MPS-U57's. You can also replace Q11/12/13 with MPSA06 (which is what I use, as they're a little beefier than 2N3904's), in case one of those is bad.
 
I'm still suspecting a bad connection between parts. Tone out all point-to-point connections between transistors. Probe from the transistor legs, for the ones where you can access them. Focus on the paths between Q11/12, Q14/15, and Q16/17.

I think Q14/16 control the -X direction, which is the half you are missing. Worst case, you could swap Q14 with Q4, if you don't have any spare MPS-U57's. You can also replace Q11/12/13 with MPSA06 (which is what I use, as they're a little beefier than 2N3904's), in case one of those is bad.
Roger. I'll try that and report back.
 
UPDATE: It's working!

I swapped the X & Y MPSU07/57s and got a partial vertical collapse. I had some TIP31C/32Cs, which had similar enough specs that I figured I could use them for short-term testing. In theory they would have worked if I'd bothered to check their pinout, which is BCE instead of CBE. I got too anxious to see a real image that I checked the screen first. After a few seconds of nothing, I returned to the back of the machine to see one of my resistors (R7) glowing white before it popped. It may have even actually flamed out, but I was scrambling to get to the power switch and I'm not sure. Another resistor (R12) burned out, too, along with F1.

Whoops.

So I replaced the resistors and put the original MPSU07/57s back in to verify operation. Same problem, but it was working that far. Then I grabbed some BD139/140s, which have been recommended as replacements here on the forum. I *swear* I put the pins in the right order, but after putting those in and powering on, BAM!, same 2 resistors popped. This time it blew F3 and F4.

Yay! More mistakes to learn from!

So I ordered some original replacements from Andy's Arcade on eBay. While I was waiting, I replaced the fuses and put the originals back in, to verify it was working as expected (with the 1/2 horizontal collapse). At some point I replaced the 0-Ohm resistors with jumper wires. Several of them tested bad when the game was live, but replacing them didn't have any effect that I could see in the board's current state.

New transistors arrived today and after getting them in, the game woke up working great. I went through the size and alignment procedure and got it dialed in. I had to bump the brightness quite a bit, but I don't know if the HV was with this Amp or not.

There's some pretty bad pincushioning on the test screens, but it wasn't really noticeable during the game.

The raster test screen was also skewing a block of red lines to the right as it got lower. The manual says it's used to adjust white tracking, and the picture shows a rectangle, not a trapezoid. What could cause that?

Lastly, some of the lines were "bowed", which is especially noticeable with text. There was Tech Tip that mentioned something similar, but I'm not sure if it's the same thing.

Even so, the display was absolutely stunning and an utter joy to view.

Thanks everyone for the help. I'm open to any further suggestions.

I am curious, @andrewb, what are the downsides of the Atari and/or Michael Kelley (et al) upgrades? I haven't replaced the 2N3904s in Q11/12/13 & Y equivalents. I'll have to order them.

I've half a mind to fab a new board with the fixes/upgrades as jumper-selectable options to be able to A/B them.
 
I am curious, @andrewb, what are the downsides of the Atari and/or Michael Kelley (et al) upgrades? I haven't replaced the 2N3904s in Q11/12/13 & Y equivalents. I'll have to order them.

Sorry, I missed this post.

The Atari/FAQ mods are not necessary. They either don't work as proposed, or are workarounds that cover up other problems, where you should just fix the root cause. None of them will produce any improvement in the picture. Also, the things that actually cause these boards to fail are not addressed by any of the 'mods'.

The key to making these reliable is to just fix them properly, and beef up all of the solder joints with QUALITY solder, because the non-plated through holes and crap solder they used at the factory are really the Achilles' Heel of these boards.

The only real mod you should do is replacing all the 0-ohm resistors with jumpers. Those have been seen to go bad, and are an easy fix. I don't do the Zener mods, or adding diodes to the bottlecaps. You do want to clean the pads where the screws that bolt the bottlecaps to the heatsink contact the PCB, as the screw is part of the electrical connection, and you can get oxidation between the screw and the tin on the PCB, causing a bad connection. I remove all of the screws and hit the pads with a Dremel wire wheel brush, to get them shiny again.

For the 2N3904's, I replace them with MPSA06's, which are a slightly beefier sub, but that's mainly so I don't have to stock both transistors. (Just like I don't stock all 1N400x diodes, I just use 1N4007 for everything.) The design of all Atari vector deflection boards (color and b/w) are similar on the front ends where you see six small-signal transistors (3 per axis). You can use MPSA06's for these on 6100's, Amps, v2000's, and G05-802/805's, they're all basically the same design.
 
Back
Top Bottom