Author Topic: 167xx Logic Analyzer boards repair (Read 4723 times)

charlyd · « **on:** March 02, 2019, 11:06:30 am »

Hello i just have a question in trouble shooting LA boards i think the best way is to create an extension cable or.... to put between the LA & board so you can power on the board on the workbench.

i am wandering if anyone already created a sort of extension cable or something else to test boards while powered on.

in the unit you don t have the room to get to the boards easy..

or is there someone maybe who has a schematic of the backplane voltages pins. so i can just remotely add power to the board. like +5/ +15 or ….

all experiences is welcome also for the repair on 1674x / 1675x boards with the known corrosion problems …. corrode tracks trouble shooting ..

what are suspious point to look for after sliders are removed..

i know from what i heared it is not an easy job, so why not try...

charlyd · « **Reply #1 on:** March 02, 2019, 11:57:25 pm »

@MarkL you wrote a while ago somewhere:

If you have a 16700 series, sometimes the program "pv" can provide some useful troubleshooting information, but unfortunately it's completely undocumented. There is some built-in help. From pv you can run all the tests on a card by selecting a card with the "s" command, then you can run "x modtests", or a specific test with "x <testname>". This will essentially give you the same information you can get from the GUI.

can you tell me where to find or start this?

MarkL · « **Reply #2 on:** March 03, 2019, 02:45:29 am »

It's easiest to do it by telnetting into the system, like so:
$ telnet 1.2.3.4 Trying 1.2.3.4... Connected to 1.2.3.4. Escape character is '^]'. Logic Analysis System Please Log in as: logic [displayname:0] login: root Password: uh,uhuh Value of TERM has been set to "xterm". WARNING: YOU ARE SUPERUSER !! logic:/> pv Shared Memory Segment 5 attached at Virtual Address: 0xc0ff7000 DMA Shared Memory is mapped to Physical Address: 0x4971000 Second half of DMA Shared Memory is mapped to Physical Address: 0x4997000 RPC system Initialized (Program=700016505, Version=1) pv> l # Model 16702B # System Memory=256 Mbytes, PCI device=1, PCI driver=1.01 Mod CPU: System CPU Board Mod PCI: 16702B System PCI Board Mod A : (0x38) 16756A Logic Analyzer (Master) Mod B : (0x0e) 16534A 2 GSa/s Oscilloscope Mod C : (0x0e) 16534A 2 GSa/s Oscilloscope Mod D : (0x35) 16752A Logic Analyzer (Expander) Exp (master=E) Mod E : (0x34) 16752A Logic Analyzer (Master) Mod 1 : Empty slot Mod 2 : Empty slot pv>

MarkL · « **Reply #3 on:** March 03, 2019, 03:24:43 am »

On the extender board question, I don't know of any that exist although there's little doubt Agilent had them for internal development. You'll probably have to build your own. And if you do, you'll have to contend with signal integrity for the high speed connections. And since there are no schematics, I couldn't say which ones those would be.

The bottom of the Slot E card is accessible if you turn the unit on its side and take the bottom off and relocate the card that has the mouse and keyboard connectors. If you take all the other cards out you can also get to the top somewhat, but it's cramped. And taking the cover-plate off the card can improve accessibility a little, and also speed up taking the card out and putting it back in, which you'll end up doing a lot (with the power off).

The backplane has 0.1" spacing and I was going to build a right-angle adapter out of standard headers so the card would stick straight out of the bottom. I have the parts but never got to building it, but that was my solution for getting access to both sides of a card. The connections would be very short, so I don't think it would degrade any high-speed signals significantly.

Another solution would be to buy a 16701B expansion chassis with the 3ft connector. No one wants them and they can be had very cheaply. You can probably disassemble it enough so that the backplane sits by itself and then you can plug a card into it free-standing. There is a controller card in it that might get in your way a little.

But even if you get access to a card on both sides, there's still no schematics to guide your troubleshooting.

You might look here for some additional information on how I approached cards with corrosion issues:

https://www.eevblog.com/forum/repair/series-defect-on-agilent-167xx-boards/msg1352747/#msg1352747

My best advice is not to buy 167xx cards "As-is". Always buy them "Used" (in ebay terms), or from an outfit that offers a return option, so you can test them and get your money back if they are faulty.

charlyd · « **Reply #4 on:** March 03, 2019, 07:52:43 pm »

today i tried again for some hours to get my X11 remote session working again with the bits and pieces i got from the sides: http://www.perdrix.co.uk/ & http://tech.mattmillman.com/16702b/ from what i see many people are struggling to get it working.. i used tried Xming/MobaXterm/VNC and with telnet i only can esthablish a connection like you wrote above, but it doesn t give the satisfaction.

i think there is a big need of a document with the exact steps to get the LA screen with X11 on a remote windows based desktop with all features accessible.so we can open Xterm on the laptop.
who can help with a walkthrough from
- step 1 fresh install of the LA with the IGNITE -CD.....to
- step 10 the shortcut on the destop to start the LA session.

what are all steps in between to get it work...

charlyd · « **Reply #5 on:** March 03, 2019, 07:58:08 pm »

as for the PV part and testing some modules i discovered something strange.
my test module 16557D was in position E.

after i hit [CTRL-ALT-SHIFT+F8] and opened Xterm
PV <enter>..
PV> s e ( e = module E)
PV> x modtests <enter>
starting test....and ran for minutes...

i the only errors i saw were 2 vRam errors…

and when i later did the test in the GUI i got ALL PASSED...this is strange not?

MarkL · « **Reply #6 on:** March 03, 2019, 08:24:43 pm »

There's a couple of key items to get remote X to work from the analyzer.

One, on the analyzer in your telnet session you have to set:

export DISPLAY=<IP address of your computer>:0

And two, you have to permit the logic analyzer to connect to your computer. There may be multiple layers to allow an inbound connection. You may have to permit the X protocol to connect, which is TCP socket 6000. And the X server itself may have an additional check on what IP addresses are allowed to connect to it.

An easy test to see if everything is set up is to run xclock on the analyzer from your telnet session. If everything is ok, an analog clock will appear on your computer screen. It's faster and easier than waiting for vp to start.

(If you haven't used X before, note that the nomenclature is a little unusual. Your laptop is the server and the analyzer is the client.)

Beyond this, I can't help with any details. I don't have windows. Hopefully someone else can jump in who's set up X on windows before.

MarkL · « **Reply #7 on:** March 03, 2019, 08:27:57 pm »

Quote from: charlyd on March 03, 2019, 07:58:08 pm

as for the PV part and testing some modules i discovered something strange.
my test module 16557D was in position E.

after i hit [CTRL-ALT-SHIFT+F8] and opened Xterm
PV <enter>..
PV> s e ( e = module E)
PV> x modtests <enter>
starting test....and ran for minutes...

i the only errors i saw were 2 vRam errors…

and when i later did the test in the GUI i got ALL PASSED...this is strange not?

Yes, that is strange. I don't have a 16557D but I wouldn't expect that.

What happens if you run modtests twice from pv? Or rerun the one test that failed?

Does the card work or do you know it's bad?

charlyd · « **Reply #8 on:** March 04, 2019, 05:41:37 pm »

sofar the way to go....

Step -1-
open my LA in het browser and choose -> "X-window Web Control" on the left side -> then "Start A New Session" <enter>
step -2-
create a remote session with vnc on port on 5901
VNC setting -> [LA16702B_IP_ADRESS]:5901

then the session opens but why port 5901? on port 5900 is still "connection refused".

what file can i edit inside my 16702B so i only have to fire up VNC after booting my laptop.

and i think, this is still not the right way to do it but it is a start..
Why not: i can not open Xterm in that sessie and work in it.

so help wanted.

charlyd · « **Reply #9 on:** March 04, 2019, 05:43:23 pm »

for debugging the boards.. because that s for what i started the topic.
here is the start

MarkL · « **Reply #10 on:** March 04, 2019, 09:56:38 pm »

On the analyzer, pull up the "Session Manager" window and select "Shared Session" instead of "Exclusive Session". This will enable the VNC server. It seems to remember this setting, so on boot the VNC server will be immediately available.

The VNC server port number is set in the file /usr/sprockets/bin/startVNC. It works for remote access if you change it to 5900, but it appears to break shared access for the front panel VNC client. You could view that as an advantage since it eliminates the overhead to keep the front panel updated, but there may be other unknown side-effects.

charlyd · « **Reply #11 on:** March 04, 2019, 11:25:24 pm »

the installation of the LA is out of the box from the reignited-cd. so i should say i have to use the default ports...most people here use a Linux server and not a laptop i think.. so..i come back on this one

charlyd · « **Reply #12 on:** March 08, 2019, 09:47:22 am »

hi i decided to go for the extension cable i will have a look how this works out.. keep you informed

charlyd · « **Reply #13 on:** March 11, 2019, 09:40:59 pm »

hi everybody it took some time.. but i am up and running after a total new FULL re-install and now all is working..like remote session and MobeXterm. so up and running now.
[if someone need help, in connecting and Xremote.. PM me]

so i did some investigation on broken boards i bought.
i got 3x 16751A boards. which i spended hours on to get them going.. from what i have read in other topic, you need a bit of luck to get them back on track..
yesterday i fixed #1 16751A board. i really found an issue on it after taking of all runners and cleanup the board inch by inch under the microscope.

today i started to work on board #2 16751A took all runners of and clean-up the tracks and inspected every inch under the microscope again but, i got this error every time ..

[ see logfile - test 1.txt ]

and after i quited diagnosing tonight finished running the tests a few times more and the errors are leaving.

my question does anybody experience this also after the card is for hours in the LA and powered on.

the last test i ran is:

i got this error

Slot E, SDRAM Fpga 0 done. Num trys = 50.
Slot E, SDRAM Fpga 2 done. Num trys = 50.
Slot E, SDRAM Fpga 1 done. Num trys = 74.
Slot E, SDRAM Fpga 3 done. Num trys = 74.
Slot E, fpga U25, port 1, Failed: Top Bank, Low Word, Mem U24.
Slot E, fpga U25, port 1, Failed: Top Bank, Hi Word, Mem U31.
Slot E, fpga U25, port 1, Failed: Bottom Bank, Low Word, Mem U71.
Slot E, fpga U25, port 1, Failed: Bottom Bank, Hi Word, Mem U74.

where can i find port 1 on the Xilinx XCV300-BG352 ?

sometimes it is there and later the board is error free.. what can i do about this?

MarkL · « **Reply #14 on:** March 11, 2019, 10:35:11 pm »

Quote from: charlyd on March 11, 2019, 09:40:59 pm

...
got this error

Slot E, SDRAM Fpga 0 done. Num trys = 50.
Slot E, SDRAM Fpga 2 done. Num trys = 50.
Slot E, SDRAM Fpga 1 done. Num trys = 74.
Slot E, SDRAM Fpga 3 done. Num trys = 74.
Slot E, fpga U25, port 1, Failed: Top Bank, Low Word, Mem U24.
Slot E, fpga U25, port 1, Failed: Top Bank, Hi Word, Mem U31.
Slot E, fpga U25, port 1, Failed: Bottom Bank, Low Word, Mem U71.
Slot E, fpga U25, port 1, Failed: Bottom Bank, Hi Word, Mem U74.

where can i find port 1 on the Xilinx XCV300-BG352 ?

sometimes it is there and later the board is error free.. what can i do about this?

Maybe start with U24, U31, U71, and U74? There's probably some signal in common to them all that's gone bad. Concentrate your microscope search in those areas, and see if you can trace the paths back to U25.

You could try turning on more debugging in pv with:
d r=9 d d=9
Sometimes it will tell you if it thinks specific data bits or address lines are stuck.

Intermittents can be hard. You can try freeze spray or gently pushing on different areas of the board with a stick to see if it changes anything. It could be mechanical.

charlyd · « **Reply #15 on:** March 14, 2019, 07:57:07 pm »

Hi Mark, yes i found all options about pv already.
yes i did some searching arround U25, two ram chips are located on the top side and two ram on the bottom which failed in the log.
but i was wandering if it could be the soldering of the FPGA chip or soldering of those memory chips. but really i don t think it is the solder but then the track/traces are lost under the mem chip and end up at the 4* 33ohm arrays. All mem. chips are traceble to those but they are also on both side located.
so did anybody ever reverse engineer them? because these ( i think first failing areas ) are build up on the board on 4 duplicated region on both side from the board.

i also see lots of CHIP8 and CHIP9 refered to on the log, where are they exactly located?

btw i tried all d (debug) values and all r (result ) values. do you know what the m ( mode) makes for a difference?

charlyd · « **Reply #16 on:** March 28, 2019, 09:22:51 pm »

...bump…. chip 8 and chip 9 any ideaa which chips could that be?

a different question: i see many times that there is a batterij and some blue wired tracks custom installed?

Why can that be..

i found some pictures on the web from such a board but see this many more times??

anybody why this modicication is or does someone have a better close up picture from this customization..?

charlyd · « **Reply #17 on:** March 30, 2019, 08:27:36 am »

can anybody help me out with this Brown board edge connector [AMP 0044] type model where to get it?

MarkL · « **Reply #18 on:** March 30, 2019, 02:41:35 pm »

Quote from: charlyd on March 14, 2019, 07:57:07 pm

...
i also see lots of CHIP8 and CHIP9 refered to on the log, where are they exactly located?

btw i tried all d (debug) values and all r (result ) values. do you know what the m ( mode) makes for a difference?

In some of the debug output I've seen, Chips 8 and 9 appear to associated with comparator messages. So, I think they are the two ASICs closest to the probe connectors, called "Acquisition ICs" in the service guide.

Chips 6 and 7 appear to be associated with messages pertaining to the memory, so I think those are the "Memory Controller" ICs, by the name in the service guide.

I don't know what the "m" option does, but fooling around with it on some cards caused them to lock.

Quote from: charlyd on March 28, 2019, 09:22:51 pm

...

a different question: i see many times that there is a batterij and some blue wired tracks custom installed?

Why can that be.. i found some pictures on the web from such a board but see this many more times??

anybody why this modicication is or does someone have a better close up picture from this customization..?

I have a board somewhere with that mod on it. If I recall correctly, the black cylinder is a capacitor. It's not clear to me what the mod does; perhaps it fixes some kind of power glitching issue.

MarkL · « **Reply #19 on:** March 30, 2019, 02:51:18 pm »

Quote from: charlyd on March 30, 2019, 08:27:36 am

can anybody help me out with this Brown board edge connector [AMP 0044] type model where to get it?

I think it is AMP 1-534204-4:

https://www.te.com/usa-en/product-1-534204-4.html

The big distributors don't seem to stock it, but search around for it. If you need one, worst case is you can remove one from a hopeless dead board.

darkroom · « **Reply #20 on:** May 31, 2019, 11:29:38 pm »

WRT getting X-Windows remote session working. I'm not sure a solution was clear here that worked...so I tried what people posted and ended up doing this:

1) Make sure you name the LA in the network setup. Mine is named Agilent_16702B. I found I could not name it using DHCP, as it kept changing the name, so I setup my network manually.
2) On LA open session manager and select multiple sessions.
3) Download VNC Viewer from https://www.realvnc.com/en/connect/download/viewer/ and install.
4) Bring VNC Viewer up, and in the top enter Agilent_16702B:5901 (or what ever you named your LA in networking, and port 5901 is the default I believe).

That brought up the LA screen in a window, and it allowed me to control the LA from my W10 PC as expected. I tried many different "suggestions" and didn't find anything I could get working, but this was just simple (and the way it should be IMO).

Austin


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: 167xx Logic Analyzer boards repair (Read 4723 times)

Share me