Agilent 54845A Failing all trigger self tests

[ARF]

So... I started studying the acquisition board design and ran into a rather stupid problem. The board contains a lot of logic manufactured by Motorola in PLCC packages. However, when I downloaded datasheets for some ICs, they did not have a description of the pinout for the PLCC housing. Quite strange.

https://www.mouser.com/datasheet/2/308/MC10H105-D-21730.pdf

I assumed that the pinout would be the same for the DIP and the PLCC, but it appears that this is not the case. There are no data for the PLCC. Maybe I'm too tired and need to sleep...

[MarkL]

Look back in this thread to Reply #20 on May 09, 2018, 12:48:59 pm.  You will find the pin mapping.

[ARF]

The fact that the guys have not been able to solve this problem for many years is a completely disappointing. I'm not smart enough.

However, I found an interesting thing that I did not see on the forum (or I missed it). Like many others, I have an error with all triggers, including Logic Trigger.
According to the internal help, Logic Trigger is easy to test; it is sufficiently isolated from other blocks  of the ACQ board.

I did the Logic Trigger self-test separately and the first error I received was: "Logic trigger memory test is failing".

The help says:

"The logic trigger memory is tested by writing out test patterns to all setable bits accessible over the serial IO port, reading data back in and comparing all setable bits. The two test patterns 0xAA and 0x55 are used.  Not every bit of every word is writable. Fifty bytes of the sixty-four byte memory are tested."

This means that the error is completely digital! Between the PCI board and the ACQ board. Serial data is not written(?) to the memory registers of the Logical Trigger.

The help says:

"The logic trigger self test includes tests to check the integrity of the serial data port, the startable oscillator which includes parts external to the logic trigger IC and a group of tests which are contained entirely inside the logic trigger IC except for circuitry to read and write from the serial port. The logic trigger self test consists of 11 separate tests listed and described in order of execution:"

It shows how to check these dependencies:

"
The logic trigger test is somewhat self contained but does depend upon the serial input port and the 100 Mhz startable oscillator. The startable oscillator is used for pattern duration and range. 

To check the 100 Mhz startable oscillator:
1. Check the enable line toggles during the test.   U84 pin 15
2. Varify that the oscillator is running during the test.   U84 pins 18,19

The logic trigger also provides holdoff for the trigger system. A 50 Mhz free running oscillator is used as the holdoff clock.
1. Verify that the 50 Mhz osc. is running.   U81B pin19
"
Everything that is described is present on my board: turn-on signals, 100 MHz, 50 MHz. This brings us back to the digital problem.

Now comes the fun part. Where is this chip with internal memory located on the board?

Considering that I see turn-on signals and clocks during the test, the problem is most likely with reading from this digital interface. It looks like the registers are being written correctly.

[ARF]

Presumably this is so, but there are probably errors there.

[ARF]

So.. go to the next...
Where is the Logic Trigger IC?

I found this thread:
https://community.keysight.com/forums/s/question/0D52L00005Idye3SAB/54835a-self-calibration-error

Q: I have a 54835A 4 channel scope ( VIN #033 Rev. A.04.30.)  The scope works fine , performance tests and all self tests  are  fine . If i start the self calibration from the UTILITY menu , vertical for all channels is pass but trigger for all channels is fail .
The error details are : Logic Trigger Duration, Logic Trigger Delay, Main Trigger Treshold Delay, Main Trigger Runt Delay, Main Trig Time Qual Runt Delay

A: This indicates that the Logic Trigger IC on the Acquisition board has a problem. Normally I would recommend replacing the Acquisition board and this would solve the problem. The problem with that suggestion is there are no replacement boards. These boards had to be discontinued early because a lower level part ran out of supply.
My gut feeling is the Start able Oscillator has stopped running
1FJ9-0002 is the Logic trigger part, and you might see I the crystal is running that is next to that IC. I am sure you could replace the crystal, but if it the Logic Trigger IC then that is big trouble.

This support person looks like he knows what he's talking about. He has internal docs. The only problem is that I don’t remember the 1FJ9-0002 component on my board.
https://www.ebay.com/itm/224709823013
I googled the component and found a link on ebay. This is a ceramic custom chip.

On my 54845A board there are 2 small square boards installed, on which some kind of logic is located on custom “JAPAN 1821-4786” chips. It looks like a regular board with a regular mounting density. It seemed strange to me that this was done on a separate adapter board, and not on the main board right away. Now I have a theory. This adapter is a replacement for the 1FJ9-0002 ceramic chip. Perhaps at the very first revisions there was a ceramic chip, then a problem arose with production and it was replaced with a cheaper option. Pay attention to the pin pattern - it looks like a pattern for a chip. I haven’t yet found a photo of the 54810 board from the back side to compare.

Notice how similar the components around the Logic Trigger are on the 54810 and 54845 boards.

UPD: I was right. This must be a direct replacement. Look at the last photo.
https://www.ebay.com/itm/224709823013
https://www.ebay.com/itm/126311386857

UPD2: The further, the more interesting. I found this thread and a mention that the 54520/54540 uses the same Logic Trigger chip. Moreover, I checked that the Time Base chip is the same!
https://www.eevblog.com/forum/repair/hp54520-logic-trigger-problem/
This means:
Logic Trigger ASIC 1FJ9-0002 = NO NAME??? = 54520/40 = 54810 /20 = 54835/45
Time Base ASIC 1SE6-0056 = 54810-89002 REV A = 54520/40 = 54810 /20 = 54835/45

The photo of the 54520 board has a rather familiar picture.
https://damien.douxchamps.net/elec/equipment/hp54542a/battery/
https://damien.douxchamps.net/elec/equipment/hp54542a/onboarding/
And now we have a block diagram that can be applied to the 54810/54835, which is nice.
https://www.keysight.com/br/pt/assets/9018-05775/service-manuals/9018-05775.pdf

[ARF]

Let's continue...

I was right, it's a digital problem.

After it became clear where to look for the Logic Trigger IC, I looked for and checked the digital interface. In this case, I was just lucky and didn’t have to search for long. On the LT substrate there is a TTL - ECL translator U1 100324 (https://www.mouser.com/datasheet/2/308/100324-1190082.pdf). This is the data input for the LT ASIC. On one of the bits, I noticed strange signals in some parts of the data block. Sometimes (not always!) the logical 0 of the input signal was about 1.57V and had a slight ringing. 1.57V is an undefined state for the TTL logic. Because of this, sometimes a logic 0 was converted by the 100324 chip into a false signal sequence. This always happened on a specific input data pattern.

If someone is looking for this problem, then I advise you to stop the scope, select “trigger group” or "logic trigger" in the self-test menu and run the self-test every time to catch the digital signals for the LT.

Then I looked where logical 0 was lost. This bit is connected via U166 SN74ALS273 (https://www.ti.com/lit/gpn/SN74ALS273) to the common data bus. U166 is used for data bus arbitration. You can see that the data bus itself does not have a very nice signal, but this is not a defect. I'll write about this a little later. U166 must set the output by the edge of the clock and we should have a clean signal on the output. Instead, I see copying artifacts from the data bus, slow edges, and sometimes not full swing on the output.

The interesting thing is that due to the fact that U166 is located on the back side of the ACQ board, I soldered a regular 2.54" 0.5m flat cable so that it should be convenient to connect the oscilloscope probes. After soldering the cable, the oscilloscope started working, the errors completely disappeared and I was able to calibrate all channels The ringing disappeared. I connected this with the fact that the cable is an unmatched load on the data line and because of this, the level of logical 0 was shifted slightly. Later I soldered this cable to other ICs for other measurements and when I returned it back, I already received LT other errors. Perhaps this is due to the heating and cooling of the U166 during soldering of the cable. One way or another, my U166 was degraded.

I ordered the components, but it won't be fast. I'll post the results when I receive the new chip.

[MarkL]

I've been following your progress on this thread, but I can't really add much since I don't have this scope.

However, I will point out that the Agilent 16533A and 16534A scope cards for their logic analyzer line (which I do have) uses the same Logic Trigger 1FJ9-0002 ASIC or 1821-4786 carrier board, and the same Time Base 1SE6-0056 ASIC.  Other scopes from that era also use this chip set.  I'm sure HPAK re-used this design as much as they could.

I think you're right that the Logic Trigger carrier card replaced the ASIC at some point for unknown reasons.  The 16533A/16534A cards also exhibit this change in their lineage as well.  I have some of both types and the behavior and specifications are identical.  However, I don't think the 1821-4786 carrier card is a complete drop-in for the 1FJ9-0002 ASIC because there are other changes in that area on the 16533A/16534A cards between the two versions.

I'm just pointing this out since these cards might be a possible source for ASIC replacements, if you should determine you need one.

I don't want to send you on a wild goose chase since you are working on a completely different device, but I can also say the scope cards have a history of bad resistors, either completely open or wrongly high value.  I've had a couple of bad termination resistors, one of which was on a clock line which caused multiple errors during self-test.

I've also bad resistors on the output setting pins of local voltage regulators that caused the wrong voltage to be output.  In one instance, it was on a regulator used to supply ECL termination networks on the card, again causing multiple and seemingly unrelated self-tests to fail.