TEKTRONIX TDS220

[JFJ]

During the boot-up of a TDS200 series oscilloscope, the frequency at the Probe Comp output terminal changes from 5 kHz to 1 kHz, in discrete steps (rather like a countdown timer). When it reaches 1 kHz, the status screen should be displayed.

If the Probe Comp frequency of your oscilloscope gets to 1 kHz, then the fault is likely to be with the LCD panel or its driver circuitry on the main board (in an area that isn't monitored by the boot-up self test). However, if the Prob Comp frequency fails to reach 1 kHz, then your oscilloscope is not completing its boot-up process.

Failing at 5 kHz, for example, might suggest that a faulty ROM or RAM chip is causing the microprocessor to crash. If there's no frequency present, then there may be a fault with the microprocessor, its power supply, its system clock, the reset line, the ASIC that provides the Probe Comp signal or the buffer device that drives the output.

[lionelkjh]

Hello; Great surprise here the oscillogram that I get out of the test point for the probes! The motherboard seems defective .....

[JFJ]

That signal is just the noise that appears on the Probe Comp terminal as the oscilloscope begins to start up. It's half the frequency of the processor clock (~ 7.6 MHz), and has a DC offset of about +2.5 V:



The noise starts about 1 ms after the Reset line (green trace, below), which connects to the processor and the display ASIC, goes high:



The noise should only persist for just over a second, when it is normally replaced with the 5 kHz, 4 kHz, 2 kHz and 1 kHz sequence of square wave signals (evident within the shading of the yellow trace):



To initialise the processor, its Reset and Halt lines (blue trace, above) have to be externally asserted - at the same time. If the processor encounters a problem, it may assert the Halt line, itself, and make all of its three-state lines go high impedance. Monitoring the Halt line might help to track down the cause of the problem with your oscilloscope. While there, you may as well check the processor's Vcc, CLK and Reset lines:

[tacpaf69]

Hi All,

Looks like I'm the "lucky" winner of a TDS 220 with similar problems (*).
At first, the scope would not even start. A wire was simply broken on the power supply board. Easy fix.

Then I got into more serious troubles, so I asked some advice to Google and ended up here.

As per Armadillo and JFJ's suggestions, here's what I found:
- the probe comp does show the bootup pattern. It indeed starts with some noise with a 2.5VDC bias at first, then the 5kHz, 4Khz, ... to end with the expected 1kHz square test signal.
HALT and RESET signals are both positive, which is OK.
To me, the µP boots up fine.

- I cheked power supplies on different ICs, everything seems OK.

- More interestingly, it seems the ASIC in charge of the display is out of synch:

• where I'm supposed to get 1.89MHz clock, I get 1.40MHz
• where I'm supposed to get 19.3kHz clock, I get 17.4kHz
• where I'm supposed to get 75.5Hz clock, I get 72.7Hz

See attachments. Sorry for the poor quality pictures, this is all I can get from my old (but trustworthy) pre Fluke PM97.

I had a look to the service manual of this scope, 6-42 more specifically.
As per service manual recommendation, the only noticeable difference whether the LCD module is connected (or is not), is the contrast bias: 0.7VDC on J202 Pin 1 when LCD module is connected. -16VDC when the LCD module is disconnected from the main board.
Clock signals are unchanged whether the LCD module is connected (or is not).


My conclusions:
- the ASIC is fried, hence the entire main board is to be replaced. Does anyone have a pretty clever idea to troubleshoot and fix the ASIC?
Otherwise, I'll probably relist the scope for parts on eBay
- actually,  I'd be interested more interested in a reply to this question: how do you explain an ASIC would fail this way (out of synch)?

(*): believe it or not, I bought my failed Tektronix TDS 220 on eBay a few days ago from a seller named lionelkjh, but who sweared he never attempted to fix the scope. Serial number from "my" main board is the same as the one on picture from post #22 ... But the seller stated he never open the unit ...
Yeah ... Whatever ...

[Armadillo]

Hi;
I don't have the schematic for this unit.
But as in repair, find clues because 1 failure may lead you to the root of the problems.
I would suggest you start at the 12 pin J202, Pin No 12. Vsync/Start. Previous owner
measure Zero signal here, I would advise you to trace this copper track, hopefully it lead you to some Gates before it reach to the IC.
Find out why there is broken signal. This may or may not lead you to further positive discovery. Hope it does.
Don't look at the ASIC yet, we look simple road first.

Regards;

Armadillo;   

[tacpaf69]

Hi Armadillo,

Thanks for attempting to help.
This is what I get from J202. For pin 12 to 10, I recorded some screenshots of my measures in my previous message.
For easier reading, I underlined the unexpected readings.
- pin 12 (Vsync/Start as you name it): supposed to be a clock @75.5Hz, 0.4% duty cycle, ... I get 72.7Hz. Close ... But yet not exactly what is expected.
- pin 11 : supposed to be a clock @19.3kHz, 16% duty cycle, ... I get 17.4Hz. Nearly 10% shift compared to expected ... meeehh ...
- pin 10 (a clock for data transfer to LCD I guess): supposed to be @1.89MHz, 50% duty cycle, ... I get 1.40MHz, 50% duty cycle. Significantly different  .
- pin 9: expected 5VDC. Got 4.89VDC. Should do the job.
- pin 7 & 8 : GND. OK
- pin 3 to 6: looks like a bus. Unchecked for the moment. A logic analyzer might be better to do the job, but I don't have any.
- pin 2: -23 to -25 VDC (LCD reference for contrast I guess). Can't remember exactly the measure, but it was within tolerance
- pin 1 (most likely the LCD contrast adjustment). Expected to be [-10;-22] VDC. I'm getting +0.7VDC when LCD is connected. -16VDC when LCD is NOT connected. When disconnected, on the LCD ribbon, there doesn't seem to be any shortcut between pin 1 and other pins of the ribbon cable. (*)

Since the ASIC doesn't seem to be providing proper clocking to the LCD, I've been thinking that the main XTAL may have drifted.
Quite unlikely of course.
But problem is : XTAL on the mainboard is 60MHz (60.606MHz if I remember correctly from another post). Sadly, my working scope is limited to 50MHz bandwidth. I attempted to make a measure still, but got garbage on my screen. I'll try to get my hand on another scope with higher bandwidth next week.

Another clue which could confirm or discard a problem from the the main XTAL: the µP clock. I read 7.4MHz on pin 16 from the *P. However, I don't know the exact expected µP clock frequency.

(*) I attempted something very crude here  :
got an external DC power supply set to -16VDC. DC power supply was current limited to 200mA for precaution sake.
I hooked it up to the PCB on pin1 from J202 while LCD was connected to the mainboard, TDS being also powered up. I wanted to see if I could overcome the 0.7VDC to get some contrast on the LCD display and see if something was happening on the screen.
The current protection from the external power supply didn't kick in, confirming there's no short-circuit (I can't remember precisely how much current was drawn).
By adjusting DC from the external PSU, I could tune the contrast.
Definitely the mainboard was trying to display something as there were things going on with the pixels, but since clocks from the mainboard are out of tune, there was nothing to be read on the display. Still I'd say that something was happening. (and of course, I forgot to take pictures  )

[Armadillo]

Good thing that Pin 12 is alive, because national ASIC ADG322B chip is not available anywhere.
Funny why previous owner measure nothing there.
But Signal logic level is also important with the frequencies.
When you get your higher BW scope next week, I trust you are technical able to troubleshoot the scope much more better.
But if Pin 12 is alive, then the LCD would also come into the suspect.
We can all learn from your findings.
 

[Armadillo]

- pin 1 (most likely the LCD contrast adjustment). Expected to be [-10;-22] VDC. I'm getting +0.7VDC when LCD is connected. -16VDC when LCD is NOT connected. When disconnected, on the LCD ribbon, there doesn't seem to be any shortcut between pin 1 and other pins of the ribbon cable. (*)

I hooked it up to the PCB on pin1 from J202 while LCD was connected to the mainboard, TDS being also powered up. I wanted to see if I could overcome the 0.7VDC to get some contrast on the LCD display and see if something was happening on the screen.
The current protection from the external power supply didn't kick in, confirming there's no short-circuit (I can't remember precisely how much current was drawn).
By adjusting DC from the external PSU, I could tune the contrast.

That is a good lead there, don't let it slip. Troubleshoot why phantom voltage at Pin 1 from the main board.

[tacpaf69]

Just found the datasheet for the LCD module:
http://www.beyondinfinite.com/lcd/Library/Sharp/LM32P07.pdf

Vertical clock (called S in the datasheet) and horizontal clock (called CP1) match together on my scope: CP1 is 17.4 Hz <> 57.4µs.
S is 72.7Hz <> 13.7ms.
As per datasheet, there should be 240 pulses from CP1 to get a pulse from S. Let's check that out:
57.4*240 = 13.7ms.
 

By the way, compared to the LCD datasheet, the timing I'm getting from the ASIC for the start clock (13.8ms) is well within expected range [12.5; 14.3]


I'm also expecting a relationship between CP1 and CP2.
Again, based on the LCD datasheet, horizontal pixels are addressed 4 by 4 (hence the digital bus of 4 bits, from pins 3 to 6 on J202).
I should be having a 320/4 ratio between CP1 and CP2, but there's a trick: there's a one pulse from CP2 clock which is used for sending "invalid" data. As a result, there should be a 81 ratio between CP1 and CP2.
Let's check this with my measurements:
CP1 is 17.4 Hz <> 57.4µs.
CP2 is 1.4MHz <> 714µs.
714*81 = 57.83µs. Compared to the actual measure, that's pretty close!
 


Funny thing: if you apply the same formulae with the figures provided by Tektronix in the service manual:
CP1 is 19.3kHz <> 51.8µs.
CP2 is 1.89MHz <> 528µs.
528*81=42.7 => not exactly the expected 51.8µs
Or maybe, I'm missing something because it is getting late here

Anyway, the important things is that the clock signals I get from the ASIC seems to be in line with the expected behaviour. 



I thought I had an ASIC issue to drive the LCD, but now I'm starting to think it might be an LCD module issue.
If only I had a spare unit to check if I'm close to the end!

[tacpaf69]

...
But Signal logic level is also important with the frequencies.
...

As per previous screen captures, sensitivity on my working scope was set to 2V/div.
Signals reaches roughly 2.5div, so that seems nice.

[Armadillo]

I thought I had an ASIC issue to drive the LCD, but now I'm starting to think it might be an LCD module issue.
If only I had a spare unit to check if I'm close to the end!

The LCD is a static dump display. It means that they is no Self RAM refresh ability or feedback signal to the uP.
Thus we should expect the uP to execute nominally until Probe Comp output 1KHz 5V Square Wave Signal.
If output is > 1.2Khz, it indicate some power supply failure. It should also be square wave.
The Pin 1 Contrast adjust input impedance is very high, thus the main board contrast voltage should not drop by so much.
Hence you should load test the Pin 1 by shorting 1Kohm resistor from Pin 1 to Ground and measure the voltage and troubleshoot it also. 
I would advise not to buy replacement LCD as yet.

Regarding the logic level, sorry I didn't accustom myself to the green screen display.

There are a lot more discoveries and leads from your troubleshooting.... looking solid good. 
French mirage or French exocet [Britain knows it well], powerful..... 

[Armadillo]

Talking about Phantom Voltage and While you are at it, it is worth to check ripples of all the power supply lines.
Somewhere < 20mVpp.

[tacpaf69]

I would advise not to buy replacement LCD as yet.

Before buying any LCD, I'm seriously thinking about buying a cheap logic analyzer and try to rebuild the picture by exporting captures from the logic analyzer. Sounds like a bad idea to someone?

I've been trying to do some sort of the same by trying to acquire clocks and data from the scope with an Arduino Uno.
But due to my lack of experience with Arduinos and coding, I failed acquiring the lowest frequency clock so I won't waste more time on this solution.

[Armadillo]

I would advise not to buy replacement LCD as yet.

Before buying any LCD, I'm seriously thinking about buying a cheap logic analyzer and try to rebuild the picture by exporting captures from the logic analyzer. Sounds like a bad idea to someone?

I've been trying to do some sort of the same by trying to acquire clocks and data from the scope with an Arduino Uno.
But due to my lack of experience with Arduinos and coding, I failed acquiring the lowest frequency clock so I won't waste more time on this solution.

Stable clock, frequency dividers to simulate the clock, CP1, CP2 and Scan Start, with arduino sync with the CP pulses to output some form of test screen.
"BUT" I would not recommend this kind of lengthy experiment. Fastest, properly 1 month, slow 6 months just to test the screen.

You can just trigger on Scan Start and watch the Data lines that all those are in sync, no jumping. You can adjust the delay trigger to watch other part of the data. The concept is when the scope is outputting same screen information, then the scope on the Data lines will be sync steady. This showed that the ASIC ADG322B is outputting correct information to the LCD, just like a LCD test Generator.

Edit: Or use a sync separator IC, but still not recommended to do it.

Edit: "by exporting captures from the logic analyser" - do you have a brand of cheap logic analyser that has this feature?

Edit: Furthermore why would you capture suspected data and export these data to the LCD???

[tacpaf69]

Thanks again Armadillo for your support.

I'm not trying to capture data to send them after to the LCD module.
For now, I'm assuming the LCD module is failed. Before spending (maybe wasting) money into a spare LCD module, I'd like to know if the scope has any failure to report.
To do so, I need to see the "welcome" screen, which is displayed after scope power up sequence, where it reveals the result of its self tests.


Being unable to see anything from the scope, I need to find a way to emulate what the ASIC is attempting to display.

What I'm trying to do with the logic analyzer is capture on my computer the data sent by the ASIC to the LCD Module.
Once acquired into my computer, because the protocol between the ASIC and the LCD module doesn't seem that complicated, I'll isolate pictures here and there thanks to the S clock signal.
I'm intending to process these information with an "advanced" text processor. Actually, even Excel should be able to do something out of these information (clocks and data) with a bit of VBA programming.
I should then end up with a snaphot of what the ASIC was trying to display.

Of course it won't be real time, but the idea is to isolate and render the snapshot:
- from the startup sequence, where the scope usually provides firmware version and result of the POSTs
- from the self diagnostic, assuming the keyboard properly works.
- and if the self diagnostic seems OK, I can even attempt to acquire the signal from the probe comp pin. If the scope is fine, the snapshot rendered on my computer should depict a nice 5V 1kHz square wave.



Cheap logic analyzers (I mean ones such as Saleae, DSlogic, XBee, ... and all their clones) are unable to export captures.
They simply acquire data and send them over USB to your computer. Actually, to use these logic analyzers, you need a software to use them.
You may use Sigrok or Saleae s/w for instance.
These s/w have the ability to export clocks and data into a CSV file typically.

[Armadillo]

Incredible feat. 
If the French can built the exocet, who dare say it cannot be done!. 

[tacpaf69]

Well well well ... Look what I've found on J202: at least 3 pins out of 12 are badly damaged!
Since the first day I opened this device, I've always been surprised by the fact the solder joints from J202 never seemed genuine, someone attempted something onto these joints: either soldering wires, or attempting to replace the connector. Anyway, here's the story of how I got there.

Tonight, I was trying to understand why the contrast bias would drop when LCD module is connected.
First, disconnect the LCD module, then power the scope up.
Bias contrast is -16VDC unloaded (pin1 from J202) when measured from the solder joint below the PCB.
Add a dummy load (62k) between bias contrast pin and ground. Still -16VDC.
On the disconnected LCD module, measure resistance between pin1 and a ground reference from the LCD module. 285k.
So the load resistance from the LCD module is higher than 62k but still, when connected, the bias contrast "drops" to 0.7VDC, instead of a steady -16VDC. So something is happening when the ribbon cable from the LCD module is connected to the main board.

For the record, pin1 from J202 is bias contrast, and pin2 is contrast reference.
I measured pin2 from J202 at -24VDC (ie. within acceptable range), when measured from the solder joint below the PCB.
But if one attempts to measure pin2 directly from within the J202 connector (ie. where the ribbon cable from the LCD module usually sits), with a needle type probe, you get ... nothing!
That may explain why the pin1 measures 0.7VDC: with no contrast reference going to the LCD module, the IC in charge of managing the contrast (LA5315) might get a bit lost, and could force a bit of anything onto the bias contrast pin (0.7VDC for instance, probably due some kind of diode within the LA5315).

Pin 9 and 11 also look pretty bad within J202, meaning the LCD module may not properly get its 5VDC, as well as CP1 (the clock that pulses each row of pixels).

Next step: try to bypass the J202 connector by soldering directly the ribbon cable onto the main board. I hate when it comes to soldering onto tiny tracks of a flex circuit. But do I have any other choice? 
Next next step: if previous operation proves to be successful, I'll have to replace the connector from J202 (which also means that the logic analyzers I ordered yesterday won't be of any help).

Does anybody have a clue of the connector reference? (good luck with the poor quality pictures I'm posting!  This is what you get when working with a old and simple binocular and a mid range smartphone)

[Armadillo]

lionelkjh is going to be very unhappy about it. 

[tacpaf69]

Next step: try to bypass the J202 connector by soldering directly the ribbon cable onto the main board. I hate when it comes to soldering onto tiny tracks of a flex circuit. But do I have any other choice? 

Well, I do have another choice.
Have you ever seen a quick fix uglier than this? 

What's the result? Wait for it 

[tacpaf69]

 

[Kilo Tango]

Congratulations, well done 

now to try and source a replacement LCD ribbon connector 

Ken

[JFJ]

Does anybody have a clue of the connector reference?

Tektronix switched over to using using the Zero Insertion Force (ZIF) version of your connector. Presumably, because it grips the ribbon cable better and doesn't cause damage to it. The connectors are available from RS Components (RS Stock Number: 542-7084):
https://uk.rs-online.com/web/p/fpc-connectors/5427084/?searchTerm=542-7084...

If you wish to buy from elsewhere, try Googling "FFC/FPC connectors".

RS also do compatible ribbon cables (e.g. RS Stock Number: 468-2656), but you would need to check the length of the Tektronix LCD cable.

[tacpaf69]

Thanks for your help JFJ.
A ZIF would definitely prevent further problems with the connector.

I found a shop in a city nearby which should be able to source the part for me.
I'm waiting for a quotation.
Should be cheaper than ordering directly to RS, plus I won't get 10 connectors, but just one or two, which will do the job.

Stay tuned!

[tacpaf69]

Still working on the thing.

I called the shop a bit too late to order a ZIF connector: a non ZIF connector had already been ordered.
It was supposed to be delivered last week but ... hopefully it will arrive this week!

In the meantime, I tried to get rid of the existing connector from the mainboard.
The shop where I ordered the spare connector told me to pay a quick visit to another shop nearby where they repair cellphones and stuff like that.
The guy in the other shop was unsure about he would be able to do the job without damaging rest of the board, so he declined.

So here I am, trying to get this connector out from the board on my own.
First option: I do have a hot air gun, but a big one to clean paint jobs and stuff like that. I can adjust air temperature from 80 to 400°C but ... Knowing that a pro wouldn't do the job with appropriate equipment, how would I succeed with non adapted equipment?

So I had  to think about another solution.
And as usual, I came up with a crude, dirty solution as I like them!  But of course, blended with lots of carefulness to not damage the tracks from the PCB.
So I started grinding most of the plastic part of the connector with a Dremel tool.
Then I melted the remaining plastic parts of the connector with my good old soldering iron, gently pulling the go-through pins one by one.
And after a while and a bit of cleaning, I'm pretty happy with the result. Not perfect but acceptable.

Hopefully there's no hidden damage to the surrounding tracks, vias, components nearby, ...

Now let's wait for the new connector to be there!

[tacpaf69]

Christmas came a bit earlier than usual this year! 

All seems to be OK apart from one pixel in the lower right corner: neither always ON or OFF, just ... grey.
Can you spot it? Near the "CH1" from the trigger settings. No big deal of course!