ADVANTEST R9211B

[m k]

First one other thing,
I/O map picture has *5E04xx FPUSEL and *5E08xx FP[V/Y]POL.
Maybe latter is polarity something.

Since clock is clearly working, no overly short pulses, keep it.

Trigger is different, I don't really understand how this machine operates behind the scenes.
Getting Started Manual section 5 is for Timing Analyzer.
There page 5-6 has a picture of 2 defined labels, page 5-8 has defined 1 edge.
After that it does Run and after that it shows a Timing Waveforms display and all waveforms are the same.
There label is named RAS but display is RAS 00 for all rows.
After that it changes the 2nd label to CAS and display is CAS 00, can't say what that 00 is.

So you should be able to do the same and change all rows to their individual pins that you have assigned earlier.
After that your picture should be so that all labels are different, triggered to 1st label and rest are what they are.
If you have not assigned any pins to other labels they all should be the same, up or down.
Your picture should also be so that triggered situation is center of the screen.
If you now want different waveforms you do new Run, or scroll around to see what other things have happened.

One difference between manual pictures and your earlier waveform pictures.
Manual picture has 'to trig' fields on the upper section, so maybe your trigger has never actually been there.

Above Then Find Edge field is Present For fields.
There must be something, > 30ns is fine.
It means that triggering level must stay longer than 30ns before it is accepted as a trigger.

External trigger input is for Armed by field of Timing Trace window and BNC Input selection there.
Don't know how it operates, maybe it's for general start of data collection since its other selection is Run.

I also remember from somewhere that some analyzers are tricky, maybe this is one of them.

[stfjohn]

I managed to get the dollar sign on all the labels, but the result doesn't seem to change...

[m k]

I don't really understand what I'm seeing.
POD 06 pin 9 CPU can be rational, it can be the first time R/_W goes down after reset.
Seems that you must add that other POD and make it a State Analyzer.

But anyway, only one trigger should be active at the time.
Turn Markers on to see what you have.

Overall timing is also something that it shouldn't.
There are some close to 50MHz pulses and the CPU is 10MHz thing.
Change Time/Div to 1000ns and do POD 06 pin 9 CPU picture again.
Then just press Run and take another picture of what you got.

First picture is from the very beginning and second from random position later.
Maybe we can learn something from their differences.

[stfjohn]

It's impossible: now the analyzer reads only 8 channels! The self test says that everything is fine, but evidently it is not so! I don't understand anything anymore! In any case the readings are always different every time I turn on the Advantest!

[m k]

Then we go back.

Earlier pictures have POD 1 and numbers after that.
That latter part must mean connected POD pins.
When you named those other labels you didn't add that "all" text, so it is internally created.
Means that if you have more than a single bit assigned to the label the machine add rows for all bits automatically.

So I think your 1st picture of reply #92 is right and so is reply #95.
Means that all bits to POD 1 and all other labels to trash.

After "POD 1" is "00" text, the latter must be a connected bit.
Next 'B' label has "all" text, no idea why the ending is not "01" that "POD 1 00" line indicates.

Go to System Configuration menu, maybe you can put the machine back to its initial state.
You have Analyzer 1 and 2, if both are on turn other one off.
Then change the Type of the active one to State.
Then select some other menu, stuff there are what they are, that's not important.
Then go back to System Configuration menu and change the active analyzer to Timing.

If you then go to Timing Format you should have a default settings.
Maybe that is already like you had in reply #92, but without those extra labels.

Next you can go to Timing Trace menu, POD pin 00 is right and 15 left.
So for triggering POD pin 06 you select 7th from right.

Next you should be in Timing waveforms display with original POD 1 lines.
Turn markers on, you should have only one trigger available.
Markers mean that you can check times between O and X, so no use for us.

Now you can tap Run and check what happens.
If waveforms are fine then fine, if not then go back to Timing Trace menu and take the trigger off.
Tap Run again and now the picture should be like it originally was.

Seems that the trigger must be delayed, so maybe that state analyzer side of the machine is needed.
One other possibility is to have a counter/timer, feed it with a clock pulse and flip external trigger when certain amount of pulses have happened.
Why's that.
Since all signals seem to pulse very early we must find means to delay the trigger so much that this first phase is gone.
For that we can use external trigger input, but I don't know where to connect it.
With state analyzer we can arm the timing analyzer part starting when high address lines (A23 to A20) have a state 5 (0101).
Assumption here is that when I/O is accessed first time the system has reached its normal operating phase.

[stfjohn]

Ok. The fact remains that, by enabling labels, the plotted channels are always only 8, while before all 16 were plotted!

[m k]

Those extra labels were clearly a wrong idea.

What you get if you change Time/Div to 1000ns, 1us or what ever it is?

Without any triggers the machine collects data from the Run moment as long as it has memory.
With luck you can see some _CS pins going down.
It goes so that you turn Advantest on and after a short moment tap Run from HP.
Then you scroll left and right, and find something or not.

[stfjohn]

last scan...

[m k]

None of those pictures make any sense.
I can't even guess what there could be, other than HP doing its own.

Timing Trace menu has a field Find Pattern.
Put there xxx1.

Now you should have a trigger delay, but I'm not sure.
What you get if trigger is POD 06 going down?

It should be so that trigger enable start waiting address 5E0xxx.
(5E0xxx, 5E4xxx, 5E8xxx, 5ECxxx)
if you change Find Pattern to xxx3 the enabling address is 5F0xxx.

Both cases then trigger first write after that address.
Most likely there are more than one write, so you should see a sane address and possibly one _CS.
There can be more, so scroll around.

E,
wrong address space.

[stfjohn]

With the configuration you suggested I don't get any scans.

[m k]

Pretty clean.

Change Pattern back to xxxx and remove all triggers.
Put Advantest on and wait text on screen.
Then tap Run.

If picture is still empty totally different than original pictures go to System Configuration menu and change Analyzer.
Maybe POD 1/Analyzer 1 connection has a problem.
You must turn active analyzer off first if you want to use POD 1 with Analyzer 2.
If POD 1 and Analyzer 2 is not possible change to POD 2.
If POD 2 and Analyzer 2 is ok try POD 2 with Analyzer 1, if possible.

[stfjohn]

I've tried every possible combination, but I always get more or less the same image.

[m k]

If you compare this last one and first of reply #107 you'll see that generally they are the same, but details are different.
Even that both are supposed to be from the equal moment.
I'd say that HP is doing its own.

So are we now swapping the repair project?
It can be a simple contact thing.
We know it worked.

[m k]

Previous empty waveforms display informs that the machine can possibly read what is coming in.
All pins are more or less the same, so Pattern xxx1 is never happening.
No error messages from the machine, so problem is more likely close to PODs.

The service manual I found online doesn't have schematics, but power supply voltages seems to be there.
Maybe one voltage for PODs is missing, or has lost its filter capacitor.

Other "easier" possibility is a POD clock.
Seems that Timing format uses internal 100MHz clock, maybe it's bad.
Change Analyzers to State format, then POD clock can be selected without a doubt.
Connect POD clock to CPU CLK pin 15 and POD 01 to CPU E pin 20, it's 1/10 of CPU CLK, 40/60 up/down duty cycle.
Leave other POD pins as they are.
Put Search Pattern as 0000 tap Run and turn Advantest on.
If POD is working you should see an empty display, POD 00 A22, POD 06 R/_W and any _CS pin can't be all down.
If display has a rational data the lowest byte should have that 40/60 pattern first and later it's more scrambled.
If display has 0000 and FFFF with some hiccups the error is still present, but POD clock is also present.

[stfjohn]

Here last images,

[m k]

So POD clock is there, but POD pins are not rational.

Address bits are right and _CS pins left.
So result is the same as Timing display, style is a bit different since clock is 1/10 of internal Timing clock.
CPU pin E is free rolling, so it should be visible somewhere.
Means that HP PODs are not operational anymore.

POD input is registered, so that part is operational, but actual POD data is just noise.
If you want to go forward with the HP you must open it.

Since all combinations are the same there is a good possibility that a supply voltage near PODs has given up.

[stfjohn]

I opened the logic analyzer and checked the voltages as per the service manual: they are all perfect! Furthermore, the analyzer passes all the self tests correctly except for pod 2 (which, for some strange reason, is not checked...).