Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack

[Jwalling]

Thanks, I've been looking at this one on eBay but when I zoom in on (open in a new tab) the picture, I can see a 'rejected' sticker at the top and a red 'X' alongside input number 2 - I assume a bad input channel is something that's hard to fix?

Probably a bad attenuator or A/D converter. Easy to fix if you have the part(s) as they are easily removed. A/D converter is socketed, and the attenuator is held in by a screw or two. It says "see attached tag" Ask the seller if the tag is still on it, and what it says if it is.

EDIT: The attenuators were VERY problematic on those... Suitable donors can also be found in the 54540A 54542A and the 2CH variants.
The bid on that unit is from Watronics https://www.ebay.com/usr/watronics  He used to work for HP. I had an email conversation with awhile back about why they had so many problems. Apparently HP/Agilent knew about it, but were never able to figure what the exact problem/cure was. I think that the problem was the leaf contacts on the shuttle relays were making poor contact with the gold plated posts.

[Gandalf_Sr]

Thanks, it went for $212 plus shipping but I missed it, a reasonable price I think.  I'll keep looking.

[analogRF]

I've got an older HP branded 54845A which was failing several tests and all calibrations. I have fixed everything up. Replaced the hard drive with a compact flash, fresh install of win98 and the software with all the upgrades up to V4.5 and added memory (64MB now from 16MB before). Cleaned up the ADC ASICs etc... and now the scope passes all self tests (multiple times), passed all calibrations and everything is honky dory except a peculiar issue that was also there in the beginning when I got the scope (it used to be win95 and SW V3.72)
As you can see in the pictures, when the graphic display is enabled, all is fine and the scope works perfectly. As soon as I disable the graphic display (top right corner) the waveform is segmented as you can see. The issue is very visible on fast time bases but it also exists on slow time bases specially with low frequency square wave it is very visible and distorted
For the life of me I cannot figure out what can be wrong with it. I don't think it is the acquisition board. When I got the scope I thought it was a software/driver issue so I replaced the clunky noisy hard drive and upgraded the software but the problem persists. The only hardware issue I can think of is the display board on the PCI slot but what is in there to test?

[Tony_G]

Does it happen if you remove the additional memory? The only thing I can think of is that something is going wrong on writing the bits out to the video card memory - Maybe a DMA conflict or something like that.

TonyG

[analogRF]

yes it does. I have played with memory chips. As i explained, this also happened in the beginning before any upgrades with 16MB original RAM and win95 and V3.72.
I have no idea where to go from here...actually I just noticed if you look at the photo that I posted, when the graphic display is disabled it seems the graticule lines are not even anymore! two of them are spaced further than the others it seems...i have to check this when I get home...

[Tony_G]

I'm at a loss sorry. Interested though in what you find if you do end up finding the cause.

TonyG

[nctnico]

I suspect this is a fault on the PCI interface board. There is a connection between the video card and the PCI interface board. This is likely an interface to project an overlay onto the output of the videocard. When the PCI interface board is generating the image, something goed wrong. Likely an address line is stuck. Better retest with a triangle wave which filles the entire screen. This should give a clue about what information shows up where.

[analogRF]

that's my suspicion too but why everything is fine when the graphical display is enabled?
actually if you look at the pictures i posted, when the graphic display is disabled and waveform is distorted, clearly the graticule lines are not evenly spaced anymore neither the small ticks on the horizontal axis. they are all misplaced. I just looked at it on the scope too. I will post new pics with triangular wave in a few hours. i have to run now....

oh, btw, i put in 4 sticks of ECC memory exactly as stated in the manual of the AMI motherboard and enabled ECC check in BIOS with no effect on anything...everything remains the same...

[nctnico]

The memory on the motherboard has nothing to do with the graphics. I guess when the graphical display is enabled the CPU is used to draw the signal (and other graphics) while without the graphics the overlay generator on the interface is used. Did you already try with a triangle waveform? And how about different time/div settings?

[analogRF]

here is triangle waveform. this happens on all time bases. It's not just the waveform that is distorted. the whole graticule, the whole screen is actually distorted. Even with no signal fed to the scope (just the base line) still the graticules are distorted in that mode.

yes, I can see what you are saying about an address stuck in the video memory. But if the VGA memory on that board is bad, it should show up somewhere else, no? The two memory chips on the VGA board (NOT the interface board) are KM416C254DJ-5 which are 512KB (256K x 16 bit) chips and windows 98 reports 1MB video memory

Is it possible that someone moved the boards on PCI slots and I need to change the IRQ in the BIOS or something? You see, in the Agilent manual it says change the IRQ9 to ISA (from default PCI) but I cannot see what IRQ9 is doing. but I can see IRQ10 is given to the VGA board.

[analogRF]

vertically nothing changes and all is fine. only horizontally and only inside the graticule area everything is misplaced

[nctnico]

I'm pretty sure the distorted display is generated on the interface board and overlayed onto the video output:

Disconnect the flat cable connecting the interface board and the video card and see what happens. If I'm correct then the display won't show a signal with the menus (graphics) disabled.

[analogRF]

OK, I'll do this today but I don't think the scope will show anything if that cable is disconnected.

I ran the Video SRAM test, Video RAM test, and Video Control Register test individually and they all pass. I believe the SRAMs are on the interface board (2 x M5416283-60 chips by OKI) because the datasheet says they have RAM and SAM operation independently.

I have ordered 2 PCI extender cables and it will take some time to arrive and then I can probe around that board hopefully. In the meantime I will try to check the tantalum caps on that board

What is that battery doing in there by the way?

EDIT:  I think the SRam here means Serial access not Static

[analogRF]

what data is being kept alive in that 32KB RAM on the interface board by the battery?

I will also check the resistor arrays on these two boards. lately I have had 2 cases of resistors being open in SMD resistor arrays

[analogRF]

I'm pretty sure the distorted display is generated on the interface board and overlayed onto the video output:

Disconnect the flat cable connecting the interface board and the video card and see what happens. If I'm correct then the display won't show a signal with the menus (graphics) disabled.

I disconnected that small ribbon cable and the scope software gave an error "the acquisition refresh rate cannot be measured" (or something close to that I forgot the exact words) and then when it got to the software it says the scope is not calibrated and none of the channels could be even turned on. the display in both cases (with and w/o graphics enabled) was blank with no trace and no graticule

[nctnico]

Too bad the scope won't work with the flatcable disconnected.

The OKI M5416283 are DRAMs with fast serial ports for video purposes. Basically you can write data while the video data is being output at a constant rate. I'd check the address lines on these chips and see if any one of them is out of the ordinary (doesn't change at all or has levels half way the power supply).

[analogRF]

yes, I will do that but I need extender cables to arrive because I have no access to the boards to do any measurements...
for now I will check the basics such as pull up resistor arrays and also tantalum caps...

[analogRF]

took the boards out. checked the resistor arrays and they are all ok....so are the caps but it is hard to tell inside the circuit...the battery still has 3.2V after 20 years

[nctnico]

took the boards out. checked the resistor arrays and they are all ok....so are the caps but it is hard to tell inside the circuit...the battery still has 3.2V after 20 years

It doesn't hurt to get a magnifier and check the solder joints.

[analogRF]

of course. I did that...cleaned the contacts and put them back in and of course no change...
the scope is well calibrated and works so beautifully except that glitch when the graphic display is off
even measurements in that mode are all ok and spot on just the way that the points are drawn in the display is wrong and only in the graticule area

i wonder how it passes all those video memory tests...i will be waiting for the extender cables to come and and have another crack at it with a scope.
 

[INFINIUM]

You know why the 3volts battery is there ,I Have one HP 54845  similar ,with a DALLAS chip I think with a battery inside or is this one in the board you show?
the screen gave me the CMOS fault?  the battery was OK  3,2V??

[nctnico]

CMOS fault from the BIOS means you have a problem in the Dallas chip. You can probably source these somewhere because billions of motherboards where made using this Dallas chip. However try to see if you keep getting the error after the oscilloscope has been powered down for a day.

[analogRF]

You know why the 3volts battery is there ,I Have one HP 54845  similar ,with a DALLAS chip I think with a battery inside or is this one in the board you show?
the screen gave me the CMOS fault?  the battery was OK  3,2V??

No I was talking about the CR2032 battery on the PCI board that is the interface between the mobo and acq board. It is seen in the upper left side of your photo.  I still dont know what information that battery holds up in the SRAM that is on the same PCI board. Cal data? system state?
however I am sure it has nothing to do with motherboard CMOS (BIOS).

The CMOS error you are getting at boot time is most likely related to the battery inside the Dallas chip being dead but the system should be bootable I guess. But it probably will use default bios setup and you must properly setup the BIOS parameters every time you boot it up
until you replace that dallar chip. This is my speculation.

[INFINIUM]

Thanks ,
the OSD  was abandoned more than 15 years, almost now dead in 2000 it was OK.
First  the CMOS ERROR asked me the bios FLOPPY  , last month..
I change the video card battery that was  with 3 Vols.
nothing  the same error , next week started to suddenly stop and start ,intermittent ..
now does not start at all  , the service manual said , put the resistor to inhibit the Power supply and replace it if does not start.
I guess is true   , trying to replace the PS , but there are to many signals in the control connector , I am busy in a small garage with cars ,etc. and I need the schematics to prevent putting wrong voltages in sensible circuits. .

[INFINIUM]

Somebody can  measure the voltages coming from the  power supply  from the 2 big connectors in the HP54845 OSD  in one I think is +5 minus 5  , in the other 12v + minus and one pair XX  volts ? in total there are  5 pairs with different colours , I am trying to replace this power supply for one standard for PCs , the small white connector has the 15 volts inhibit signal and the extra for sincro.. test or something rare.
colour code  cables and voltage . in resume  I need,
if one day the PS fails you also will have less trouble.
Please be aware that without PCB cards or problem in the laching relay the PS does not turn on