CPD-1604S Repair

[vol.2]

Picked up this Sony Monitor which was mostly used for CAD work and medical imaging back in it's day. Totally non working for $75. I knew it was going to be a huge PITA when I made the offer, so it's all my own fault of course. The hope was that it just had a couple of bad caps in the power supply or whatever and that I'd be able to just do that and clean it.



Yes, it looks like dog shit, but I just wanted a thing to tinker with and have to keep in the basement for playing Doom. But it did clean up pretty nicely:

Spoiler

Unfortunately, after getting to the insides, I discovered it has this little deflection control board which was lousy with those early type SMD capacitors which always leak their guts on everything. Sony seems to have been kind of obsessed with them in the 90s and I've had to do complete recaps on a variety of devices from that time.



To make matters worse, the electrolyte had time to corrode things on the board, including these two 20 pin JST board to board connectors that are impossible to get without buying them in bulk from Japan. I would need to get $50 worth of crap just to purchase these 2 little plastic connectors.

I decided to just use some wires to connect the two sides together. I purchased some wires precrimped with posts on the ends so that I can permanently wire on the card side and then solder the posts into the main PCB. That way at least I won't destroy the wires if I have to remove it multiple times (hopefully not, but you never know). If everything works well, I might go ahead and spring for the correct edge connectors, but meh.

In order to mitigate the damage from the leaked caps, I soaked the board in vinegar for awhile, washed it down with soap and the soaked it in 99% IPA and let it dry out completely. There are many exposed traces, but I don't see any broken or missing traces, everything is intact and I traced out all the worst looking spots to check for continuity. It looks good.



I tested the regular caps on the DA board (little deflection guy) and found them to be all horribly bad, and I remembered all of a sudden that the Nichicon PF series is prone to drying out. After that, I realized I needed to check the rest of the set out and I tested every single cap in the whole thing. I found that about 95% of the caps where bad, most of them terribly bad with ESR in the 500 kOhm range and wildly out of spec. Some of them where just marginal (but still more than 20% out), and others were a-ok.

Behold the main board:



I have caps coming in the mail in a couple days, so hopefully it will all work out and turn on. I like the controls and features on this monitor; it has a separate flyback and focus pack in order to really dial in the focus in both the H and V. The deflection control board offers granular control over each of the scanning frequencies.



I'll update here when I make any progress. Hopefully someone finds this somewhat interesting. I do this kind of work pretty often, but also a lot of audio equipment like tape decks and old amps.

[vol.2]

I'm still working on this project, and while I have made progress, it's not working properly yet.

Basically what's happened so far is that I did the recap, only replacing the bad ones, and then I fired it up and I had no Vertical. I managed to trace the Vertical back to the 24V power supply and a blown fusible resistor. Replacing the resistor restored my Vert, but I then was left with a severe Pincushion issue.

For those not familiar, the Pincushion circuit is what's responsible for the Horizontal geometry adjustments on a CRT. All of these circuits in this monitor are on the small DA PCB that saw some electrolyte damage from the SMD electrolytic caps, so it's a bit of a slog to find the issue.

In the meantime, I did manage to domestically source and replace the JST connectors that run between the main PCB and the DA board. This was sort of necessary because I am having to take the DA board on and off a lot to troubleshoot, and I already managed to ruin a trace on the underside by doing that multiple times. I bodged the trace of course, and now I don't ever have to mess with it again because I have the correct edge connector on there. I bought 2 sets just in case because they were only like 2 bucks a pop, so nbd.

Another piece of good news is that the tube is very strong in this set. When I opened it up, the PCB was super filthy and I was pretty worried about the health of the tube, but I realized that there are Sony Repair Center "Refurbished" stickers all over the place on the inside, and also that the tube looked pretty clean. The date on the tube is also much later than the rest of the chassis, so I think this tube must have been an NOS replacement towards the end of the 90s when it was being used in the CNC shop. Luckily (to some degree) for me, the rest of the chassis must have self destructed before they had a chance to put much wear on the new tube, because there was almost no dust on it at all. (a little bit). My guess is the 24V fusible blew and they just said screw it.

If anyone feels like throwing in, this is the SM. I've more or less confirmed or denied all relevant voltages in the circuit and I can't find any smoking guns, but I can see where the breakdown is.

This set uses a Pincushion IC (IC303) to manage all of that. It takes a sawtooth on Pin 1 and compares it to another current on Pin 2 to produce a Parabola which is the balanced Horizontal geometry output. I am not seeing the correct sawtooth on Pin 1 of IC303. Replacing the IC didn't help. I see the sawtooth on the other side of the 39K resistor that feeds Pin 1 (Pin 2 has an equal resistor).

All pots have been removed from the board and checked for function and value. All resistors in the circuit have been measured and are all dead nuts except for those which would be subject to the effects of parallel resistance. All electrolytics have been replaced and quadruple checked for value and orientation by measuring continuity to adjacent components in the schematic.

The only "bad" waveforms I can find are Pin 7 of IC303 and Pin 5 of IC303. 7 is supposed to be the parabola that is generated by comparing the current of Pin 1 and 2, and it's not a parabola, it's messed up. Pin 5 is a square wave with a long, flat top, and it's supposed to respond to the Pincushion controls by shifting the long, flat top portion from left to right until it's symmetrical; the flat top portion is "stuck" on the left side of the square wave, reflecting whatever is going on with the raster.

If someone is interested and wants to see any waveforms or details from the SM, I'll gladly post them.

[vol.2]

Unfortunately, I ended up having to remove a large number of surface mount components to clean underneath of them and inspect things for damage. I found a couple of possible causes for the issue including a cracked resistor (that was still reading the correct value, but might have been experiencing issues under load), and a more heavily damaged area of the board that two transistors with corroded legs which kind of crumbled off when I removed them; they were still in circuit and voltages were mostly correct when I pulled them out, but they were clearly corroded on the inside somewhat and therefore could have been creating some weird effects.

The issue with this repair is that it isn't a "smoking gun" kind of situation where I can find a single bad part and match it to the symptom. Because the circuit is "mostly" working, it's just one specific area that is having issues, and the issues are related to current and not voltage. The only bad waveform I can find is Pin 7 on IC303, which suggests that the Sawtooth on Pin 1 is current limited, and indeed there is a slight amplitude issue on the waveform. It's not enough for me to have thought anything of it, but I had someone point out that it's a crucial spot to the effect of the circuit. If there is enough parasitics in the circuit leading up to Pin 1, the current balance will be off and the Pincushion controls will not function and viola.

The assessment was that there is 1) enough residual electrolyte on the surface of the PCB to make it conductive and create effects between the pins of transistors in the Pincushion circuit and/or 2) enough components have a small amount of damage from the electrolyte to create extra resistance in the circuit and current limit Pin 1.

I removed all the components from the effected areas including the areas involved in the Pincushion stuff, washed the board with warm water for a few hours and dried it thoroughly including a second wash with IPA to drive moisture out of nooks and a heat gun.

I've got replacement transistors and a bunch of high accuracy resistors to replace almost everything, so I'll just do that and cross my fingers.

If I do that and the fault is the same, at least I'll know what it isn't. I suppose if the fault is the same, it will probably turn out to be a broken trace I missed somewhere, or a broken via given that it's a double sided board.

[vol.2]

If anyone is following this at all, I finally fixed the problem, more or less.

The issue turned out to be the 24V line failure in general. I knew this was an issue from the beginning, but because the Multisync PCB had all the electrolyte damage from the SMD caps, I wasn't really focused on that, I was focused on cleaning up the board and broken traces.

It turns out that in addition to the fusible resistor blowing up on the 24V line, there were two other culprits. There is a connection between the Pin 2 of the Pincushion control IC (IC303), and the "Vertical Position Signal" on the main board, deriving from Pin 12 of the Vertical output IC.

On the main board, this is Q202 and Q203. They are a matched pair PNP/NPN set of transistors that, the best I can figure, are providing the balancing feedback for the current-derived parabola that the Pincushion IC produces. In any case, the way that IC works is that it takes current inputs on Pins 1&2, and it produces a parabola; the Pincushion controls tweak the current of Pin 1 to match the current produced by the Vertical Position signal coming in on Pin 2.

Q202 and 203 are coupled to the 24V source and were clearly weakened by whatever happened previously. When I checked them, they were producing 1.3V at their output and it's supposed to be 8.3V; when I pulled and checked them, Q203 measured very off. Replacing the pair restored the 8.3V, and the input voltages across Pins 1&2 of IC303 stabilized.



However, this didn't get me back Pincushion control. In fact, it was now worse than ever.

I set about replacing every transistor that was in any effected area or connected to the Pincushion control at all. I also bought and replaced three of the logic ICs used for the frequency selection circuits. The more I replaced, the tighter all the voltages around the board became, but the Pincushion still didn't work.

Finally, I went back over the voltages and noticed that IC304, the shunt regulator which feeds Vcc of IC303, had a slightly too high voltage. It was supposed to be 20.6 and it was 22. This didn't seem like a big deal to me, but it absolutely was. Replacing the IC restored Pincushion control and they basically work now.



I still think there are probably some areas on the Multisync PCB that need to be addressed, and I would probably get less fussy geometry controls if I did that, but I can make it more or less look pretty straight now and that's what counts. I might go back to this and either just clean up the board a little more by replacing the remaining transistors (I have them all to do already) and the rest of the logic ICs.

I also started making a copy of the PCB in KiCAD, so I might finish that and just have a brand-spanking new one made, but with tantalums or something instead of the SMD 'lytics.

So I got it back together working, and then I noticed that the yoke was out of alignment, twisted a little bit. This is a pretty common thing to happen in shipping; yokes get knocked free. In this case though, I think it was just sloppiness on the part of the Sony Service Center people. I think I mentioned before that this monitor had been refurbished late in its service life (it was used in a CNC shop), and they had a new tube installed in it. Which is great; the tube is bright and strong and calibrates much higher than the SM stated cdm2 level. However, I can tell that it wasn't installed too carefully. The twist in the yoke had to have been there when they did the setup because, not only was it caulked to hell and back, but the static convergence had clearly been performed with the yoke twisted.

Typically when a yoke twists in transit, the convergence rings will also be off because they are tuned to the spot the the yoke is supposed to be at before it got rotated. In this case, the convergence was good with the yoke twisted, and now it's off that I went in and corrected it.

Additionally, the yoke is now pointing upwards a little bit so the picture has a slight vertical bow in it on the top and bottom.

So that's pretty much it for now. Next I'll go in and fix the yoke, and probably have to do a static convergence and redo the color balance and geometry yet again. I'll update here when I finish and include some after pics.

[jd213]

Glad you got it fixed, sorry if you felt you were screaming into the void, I didn't really feel like I had anything to contribute to the thread, but I was at least following it.

You've also given me a little more motivation to fix my 20L5 (I also need to get a lull in my day job and stop procrastinating...)

[vol.2]

Glad you got it fixed, sorry if you felt you were screaming into the void, I didn't really feel like I had anything to contribute to the thread, but I was at least following it.

Thanks! Yeah, I know this stuff is pretty difficult to help out on, especially remotely in this way. I have been lucky enough to get some pointers on other forums when I ran into a wall, but nothing that specifically pointed the way to the problem. Mostly people have helped me when I couldn't figure out how a circuit worked or whether or not a specific wrong voltage was something to worry about.

I actually learned quite a lot about analog controlled digital logic circuits doing this repair. This particular monitor is from 1992, which is a really weird spot because it was designed just before the big monolithic control ICs came into being. Literally 2 years later, all the Sony CPD monitors were using vastly larger ICs that took care of all the Multisync stuff. It was probably only about 5 years this style of XGA thing was even made. It's pretty cool as it can do 1024x768 interlaced, if I can ever find a source that can drive it at that.

You've also given me a little more motivation to fix my 20L5 (I also need to get a lull in my day job and stop procrastinating...)

You should!

[vol.2]

Well, I ended up having to completely remove all the yoke wedges and glue and start over from scratch. Which sucked. But on the plus side, I was able to just pretty much dial it in with no permalloy strips or anything. The convergence isn't 100% perfect, but it's pretty damn good, and the sides are straight with the exception of a tiny distortion at the top of the image that isn't terribly noticeable unless you're really looking for it.

I might end up going back in to see if I can improve the Pincushion circuit more at some point, but maybe not. It's close to dead-nuts as it is, and I would have to risk messing up the fragile PCB if I messed it again. More likely I might make a replacement PCB and see if that improves it.

Here's Dark Forces

[tongshadow]

Good job, I love troubleshooting topics like these!

That's the challenge of working on older hardware, the dammned SMD electrolytics, they all inevitably fail at some point.