[Scope is working!]LCD display flaw - TDS210

[neil]

It's that same board-corroded TDS210, and it's still here. And mostly dead.

But in the words of Miracle Max: "There's a big difference between mostly dead and all dead. Mostly dead is slightly alive."

I've arrived at a point where it boots (and, of course, fails acquisition) and can cycle through all the menu's, and the front panel passes the onboard testing. I had to hard-wire the contrast to make it visible, since it doesn't respond to the contrast menu (or the service menu version either). The probe comp. signal is present, but no traces are visible.

But this question has to do with the display. It seems that every fourth vertical row is not painted. It's especially obvious if you look at the menu item values. Since the screen is black-on-white, the pixels in question are lit, when they should be dark, so I guess that means they're not dead pixels? The same problem is plainly visible during the screen test, and shows that the problem affects every fourth column, full height of screen.

So, is this flaw likely to originate from the ASIC (which undoubtedly has some troubles anyway), or does the every-fourth-column sound more like something that would come from the screen assembly, or perhaps the actual cabling/attachment to the LCD? 

It feels like the pattern of the failure should be enough to answer that, but I'm just not that familiar with LCD systems.

[james_s]

I'm thinking that looks like a problem with the video RAM, or the logic driving it. I have no idea how the Tektronix hardware works but typical video hardware will have a RAM or ROM dumping 8 bits at a time into a parallel to serial shift register which then clocks each pixel out to the display. If one of these bits is stuck then you will get vertical lines through the display. It may be that this hardware clocks out chunks of 4 bits at a time and one of those lines is stuck. Can you find the pinout for the LCD? If you can work out what lines carry the pixel data you should be able to use another scope or logic analyzer to verify the gap in the pixels being clocked to the display.

[neil]

james_s: Thanks for your reply. I found this documentation for the contents of the flex cable to the display, and in fact there are four data lines from the ASIC, so I'll look to see if one of those is dead.

BTW the first line (contrast) was running about -20 v and the screen was solidly white; it looks like it's meant to be regulated by a transistor shunting to ground, driven by a line from the ASIC. Apparently that's a dead output from the ASIC, and I've had to add a resistor voltage divider to bring the line down to -17v which gives a decent screen display.

[james_s]

Ah that's definitely promising then, here's hoping it's a connection rather than a dead line from the ASIC, otherwise it's pretty much game over until you find another scope, but then you might find a cosmetically trashed one that has a good board in it, or at least a good ASIC. Not that a TDS210 is worth heroic efforts to repair but I do hate to let a fault get the best of me. What happened to yours to get it in this state?

[neil]

This was a relatively low-cost eBay "parts or repair" deal. I've never had much hope for this scope since I opened the case and saw the extent of the corrosion - I didn't expect that it would ever be usable, but I decided to learn as much as I could from it, and it is serving well in that regard.  Input from you and others here has helped a great deal, as have other internet resources.

[james_s]

Yuck! That's a mess! I wonder what the hell happened to that thing? Is that before cleaning? It's possible that you have some rotted out vias or damaged traces somewhere, the ASIC may be ok, it's hard to say.

[neil]

Oh, there are definitely some rotted-out vias and traces - I've bodged around them in several places to get the machine to this stage. I think the acquisition ASIC is probably toast since the VR that should be creating a -2.6v rail for that shows +0.6v on its emitter! That must be back-fed from downstream, since there is -4.6 v going into the VR! I haven't looked into that yet...

[james_s]

I wouldn't be so quick to condemn it, I would be much more concerned if the voltage had shot up to -4.6V or whatever. I think there's a reasonable chance the ICs are all fine.

[TERRA Operative]

If the board is ultimately unrepairable, I wonder if it is worth salvaging the chips (assuming they are ok) for future repairs?..

[neil]

Well, this has been a good two days for the little scope that could.

Yesterday I checked the four video lines in the display flex, and found one hanging at 0v., accounting for the missing every-fourth vertical line on the screen. Tracing it back, I found yet another bad-via-that-looks-OK. Reworking this restored the full screen display!

Today I tackled the bad -2.6v rail problem (there are actually two rails). I removed the power transistors, and tied the rails to an external power supply. When I booted the scope, it came up fully live!  I was astounded that the ASICs and other chips were still functional (which had been predicted by wiser voices here). Everything now works, including the contrast setting, which allowed me to remove my bodge from yesterday. 

So I need to figure out how to reconstruct the -2.6v power supply (which has at least a few bad vias).  I have shamelessly used the image JFJ posted in an earlier thread because it's a great reference photo of a nice clean board. But identifying the components is a problem. The transistors are not a big issue, and I can tell PNP from NPN easily enough.

The main mystery is U150 (code "A00"). It seems to be  directly driving the bases of both Q153 and Q151 in parallel with a single output pin (which seems odd) and I thought maybe an op amp? Some sort of voltage regulator chip? It's a very small 5-terminal package.  I'm not sure that it's alive, since the output voltage seems to nearly match the input -4.65 voltage, while I would expect it to be nearer the target voltage of -2.6v. 

I measured the current drawn by the two -2.6v rails while using the external supply. The one served by Q153 peaked at about 35 mA and the Q151 rail peaked at about 160 mA.

I suppose that I have no real objection to just wiring up a regulator on a little daughter card, and that might actually be the easiest way to go. Plenty of space in the plastic case. Any favorite devices?

[james_s]

Nice job! It's always a good feeling to bring something back from the dead, even if it's not really a good investment in terms of time spent relative to the value of the equipment. Regarding the dead rails, can you find the regulator for the -2.6V and just wire it up with jumpers made of wire wrap wire, bypassing any faults in the board? Otherwise assuming the current is low you could just glue a LDO to the board somewhere convenient, solder some bypass caps right to the leads and then run a wire from that to the load.

[JFJ]

... The main mystery is U150 (code "A00") ...

On my board, U150 is marked A00B, which is a Low-Power Operational Amplifier With Rail-to-Rail Input and Output (LMC7101BIM5).

[neil]

Thanks. Being retired makes it easier to justify the time spent!

I think the regulation is by an op amp driving a pass transistor. That's what is used for the other two large devices in the picture - a dual op amp U110 driving the bases of Q110, Q141, and also Q143.  So this is probably also mirrored here for the negative rail.

The pass transistors are from On Semi. Initially I mistook what is apparently the date code (R13 and R11 in the photo) for the ID, so couldn't find them. But looking under "P1F" and "CE" yields likely identification as pretty normal NPN (PZT2222AT1) and PNP (BCP69T1G) power transistors.

I may have identified my mysterious "A00" component too:  I think it's a TI rail-to-rail op amp LMC7101, which has the right package, a pinout that seems to make sense with what I know about the PCB traces, and the doc's say it has device markings of "A00A" or "A00B".

I need to read up about negative voltage regulator circuits, and trace this out a bit more. I'm not sure if there's a zener hiding under the board, for instance. Fortunately, I think either restoring it or adding a standalone regulator are both good solutions to get the scope finished and buttoned up, but I'll let you know how it turns out. Thanks again for all comments and help from the blog.

[neil]

JFJ - Thanks. Same conclusion I came too. They seem to be readily available from the usual sources, so that's one less mystery to deal with!

I need to trace out the circuit a bit more to see where it may be failing - probably more bad via's (in addition to the ones I already know about) although I also found Q153 with an open B-E junction, and have no easy way to evaluate U150 other than just replacing it.

Thanks again for your comments and especially the board photo which put me on the right track in terms of rail voltages.

[james_s]

I still can't figure out what happened to this thing. Do they have an internal battery that leaked? Did somebody drop it in the ocean? Maybe it was damaged in a flood?

[JFJ]

... have no easy way to evaluate U150 other than just replacing it.

I had the lid off a TDS210 with a backlight problem, today, and made a note of the U150 pin voltages:

1: -3.32V
2: 4.97V
3: 0V
4: -0.001V
5: -4.54V

Pin numbering is as per LMC7101 datasheet:

I still can't figure out what happened to this thing ...

Maybe someone followed Dave's example and took it on a charity mud run. WARNING! - the following video contains scenes that some forum members may find distressing 
https://youtu.be/yAahT8_P-_E

[TERRA Operative]

I remember that... Was for a good cause though so I can't be too sad. 

Imagine what he could have sold those knobs for! Too many missing knobs on too many scopes being sold.

[coromonadalix]

Poor scope    had tears in my eyes  loll

[neil]

JFJ:  Thank you again. Those voltages are similar to what I saw, but don't make sense to me. I don't see how those voltages will give me -2.6v out of the pass transistor (which inputs -4.65v).

If I traced the circuit correctly (easy to miss things with bad vias, however) it seems like a typical op-amp driving a PNP transistor (two, actually, Q151 and Q153). The non-inverting input is tied to ground, and the inverting input was tied via 1K resistor to the pass transistor emitter (i.e. output voltage) to provide feedback, but was also connected through about a 10K resistor to the +4.30v zener voltage reference (U155). (That reference is also fed to the dual op-amp driving the pass transistors for the two  positive rails.)

Typically, I thought a reference voltage was applied to (In+), and a feedback sample to (In-). I'm struggling to understand this different situation, but then I haven't looked at a negative regulator before.

Does this make sense? I'm guessing that the VRef reference voltage is fed to (In-) instead of (In+) because it's a positive voltage and we want to use it to set a negative rail, so it needs inverting. And the feedback from the transistor output has to also come to that same (In-) input, so the two are sort of summed. Since (In+) is hard-wired to 0v, the op amp would try to bring (In-) to near zero as well, which means the output voltage, via feedback, would have to cancel out VRef. So it seems like that should force the output voltage to around -4.3v (which is what I was seeing on my board).

I tried that circuit on a breadboard, juggled resistor values, and got transistor output voltages no closer to -2.6v than -3.2v. But I was able to get -2.6v out of the pass transistor if I changed the circuit a bit: I disconnected the +4.3v reference from (In-), lifted (In+) off ground and put -2.6v on (In+) as a reference. Doing that seems to give a nice, stable -2.6v output with decent regulation.  Unless I can understand the problem with the original circuit, I think I'll go with this modification (with the addition of a 2.6v zener to provide the reference voltage).

[gkmaia]

I had a similar problem a while ago and the issue was the BU6132F IC on the LCD board. It is unlikely you have a problem with your display ASIC or motherboard.

[JFJ]

... I'm guessing that the VRef reference voltage is fed to (In-) instead of (In+) because it's a positive voltage and we want to use it to set a negative rail, so it needs inverting. And the feedback from the transistor output has to also come to that same (In-) input ...

That is correct. Tracing out the circuit on my board was straightforward, because it has no bad vias:



The 2.5V reference is applied to U150's inverting input via a 9.5k resistor (two 4.75k resistors in series). As the inverting input is a virtual ground (maintained by the 10k feedback resistor), the current through the 9.5K input resistance is 2.5/9500 = 0.263mA. That current has to pass through the 10k resistor - causing a voltage drop of 2.63V, below the virtual ground.

[neil]

Thank you for taking the time to trace and explain that circuit! 

I found I hadn't updated my notes after fixing a bad via for the zener reference U155, so I was considering the original 4.3v value I had measured, rather than the correct 2.5v that it now shows (which matches yours).

So they were able to use one zener +2.5v reference (by means of a single op amp and a dual op amp) to stabilize four rails: +3.3v (Q110),  +2.8v (Q141), and two -2.6 rails through Q151 and Q153.  That's pretty elegant.

When my strip of LMC7101 op amps arrives, I should be able to restore the original circuit after a few via and pad repairs.

[neil]

So they were able to use one zener +2.5v reference (by means of a single op amp and a dual op amp) to stabilize four rails: +3.3v (Q110),  +2.8v (Q141), and two -2.6 rails through Q151 and Q153.  That's pretty elegant.

Actually, FIVE rails from one reference. I missed the small one through Q143 that shares the drive with Q141.

[neil]

The TDS-210 rust-bucket is now buttoned up and appears to be fully working! Yes, it's not much of a scope, but it still feels great to see it alive.

For those playing along at home, the final tabulation of problems encountered:

- two bias resistors rotted out of a quad pack caused the display to be turned off.

- corrosion of pins of one of the ASICs resulted in no clock reaching the CPU and RST sitting at active low.

- bad via blocked one of the four display data lines between ASIC and flex connector, blacking out every fourth vertical column of the display.

- bad via to ground disabled U155, a 2.5v zener reference, making 3.3v and 2.8v rails too high (around 4.2v).

- bad vias to U150 (possibly killed U150) op amp causing -2.6v rail to be too high ( around +0.6v) to ASIC.

- open B-E junction of Q153 killed another -2.6v rail to ASIC, leaving it floating.

Fixing this scope has been a wonderful educational experience, and I am very grateful for all of the responses and suggestions that I received here on eevblog. Thank you to all!

[james_s]

That's really impressive you were able to get through all those problems and make it work again, I suspect most people would have given up and scrapped it long ago. I'd still love to know what the heck happened to that thing before it came into your hands because the only time I've ever seen anything like that was when a battery leaked on a PCB. Here's hoping whatever process did all that damage is not continuing to rot anything out.

[neil]

Yeah, it'd be satisfying to know. Whatever it was, it was pretty active - it even corroded the metallized interior coating of the plastic case  over a several square-inch area in the left rear corner (i.e. the side away from the display and power supply).

Well, no battery to explain it. The caps in the power supply look good, although I guess it could be a replacement power supply. Still, the distribution doesn't seem right for even an exploded cap, as the metallized lining is clean over the entire power supply area.

The only (maybe) clue is a small brownish dried-liquid stain on the inside of the top of the front-panel/case. I suppose that might indicate something spilled onto the top of the scope and drizzled inside - the outside was clean but of course that could have been wiped off. If I was a real detective, I guess I could lick the stain to see if it tasted like coffee or coke, but...