Author Topic: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack  (Read 40125 times)

0 Members and 1 Guest are viewing this topic.

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 17392
  • Country: nl
    • NCT Developments
I have been hunting for a 1GHz+ scope for over half a year. I prefered one which has a high enough samplerate to do this in a single shot. Ofcourse it is 'easy' to spend several $k on a Lecroy Wavepro 7000 series or something similar but I didn't want to spend that much and avoid too much overlap with the oscilloscopes I already have. The older Lecroy LC584 is also interesting and I nearly got one but the Ebay seller bailed out on me (I did got a refund though). I also looked at the Tektronix TDS694C and TDS784 but the problem with these scopes is that they are older than the Agilent 54835A, have an obsolete NVRAM (SRAM + battery) inside and need special calibration software. The 54835A OTOH can calibrate itself using the auxilary output so it doesn't depend on calibrations values or options stored in a NVRAM. Big plus! My main application for the 54835A is looking at high frequency signals (digital and analog). The 54835A has a maximum depth of 65kpts on 2 channels and 32kpts on 4 channels and no peak-detect. I guess that if you don't want to spend much then deep memory and peak-detect are out of the window. OTOH the 54835A has a lot of tools for signal analysis like FFT, color grading and stacked/chained math functions.

After some haggling and waiting I got this one from an Ebay seller called express_test:


First I would like to point to this nice page with a different repair job on a similar scope: http://www.dasarodesigns.com/projects/fixing-an-hp-agilent-54845a-infinium-oscilloscope/ It has many details on the circuits.

The most appearant problem is that it doesn't boot but otherwise it is a complete mistery. The upside is that the hard drive is detected which means it spins up and likely still has the software on it.

I opened it first to check whether all the wiring is connected and there are no potential safety issues.

Some pictures of the inside. First the part where the PC lives. Inside there is a VA503-A Super Socket 7 motherboard from FIC with an AMD K6-2 400MHz processor and 64MB of SDRAM memory.


The acquisition board is at the underside. This is also where most of the airflow is directed.


Time for firing it up for the first time. With a PS/s keyboard connected I got passed the BIOS error message and after a short while the firmware started. The scope runs on Windows 98 (first edition). After a quick check it turned out (only) channel 2 is not working. All in all the noise isn't as bad as I had expected seeing the two 120mm fans. The high pitched whining noise from the hard drive is way more irritating.

Time to take try and fix it
Changing the battery (which was empty) helped to get rid of the BIOS error and made the scope boot without needing to press a key. Note the bodge capacitor!


The first thing I wanted to try is to swap the attenuators. From the service manual I learned that the probe connector panel had to come off before the BNCs of the attenuators can be unbolted. Unfortunately this means peeling the sticker off before the screws which hold the panel can be reached. That didn't work out very well:



I think it will still look reasonable when put back together again.

Some further disassembly:



A spark plug socket for a socket wrench fits nicely on the BNC's nuts BTW.

After swapping the attenuators it turned out the problem still persisted. Not happy! In the next step I swapped the ADC hybrids between channel 1 and 2:



The ADC hybrids are connected to the board using some sort of socket with pogo-pins (spring loaded pins) which make contact between the ADC hybrid and the board. Ofcourse I cleaned everything with alcohol. Still no luck though. The problem moved with the ADC hybrid. In a final attempt to rule out any other problem I cross connected the attenuator outputs using SMA cables.


Still no go so the only conclusion is that the ADC hybrid which is now on channel 1 is broken. I managed to source a module for (what I think) is a reasonable price but I have not received it yet.

After this I ran a full diagnostic which also pointed to some errors in the trigger circuit. Maybe it has been damaged because a piece of shielding metal wasn't fixed to the frame. Here is a picture which shows the shied re-attached to the frame using some M3 screws:


When browsing on the hard drive I found an interesting help file: HP4XS.HLP . This seems to contain a partial guide to do component level diagnostics/repairs. Hopefully it is helpful if it turns out the trigger circuit is really defective.

Cleaning
With nothing else to do let's give the outside a thourough cleaning to get rid of the grease marks and sticker residu. Much better if I may so so:


Hard drive image
I have not been able to find any older versions of the software for this oscilloscope so I made a backup using Ghost. For this I needed to connect an extra hard drive. The problem is that there are no power extension leads to do this so I made one out of the wires from an old AT power supply. The VA503-A motherboard has both AT and ATX style power connectors and the AT one isn't used but suitable to drain some power from.

After this creating an image was easy.

Rubber fan mountings
The rubber fan mountings where completely perished and I have a bag of leftover cable glands.


I like to think this made an improvement.

Time for some uphacking!
I already knew from the service manual that the hardware for the 54835A, 54845A (1.5GHz 8Gs/s) and 54846A (2.25GHz / 8Gs/s) is exactly the same so there had to be some option strappings to set the model. Forum member Jwalling pointed me towards some resistors on the acquisition board which set the type of oscilloscope. The bad thing is that these resistors are on the solder side of the acquisition board so I had to take the entire board out.

The board is covered with a rubber-ish RF shielding slab.

After removing the RF shielding:


When glancing over the board something odd caught my eye and I had a WTF! moment:


This chip isn't attached to the board except with one leg! Some flux and solder fixed that but oddly enough it didn't seem to have any effect on the errors the self test showed.

Back to the uphacking part. I located several resistors and decided to attach some switches after a few futile attempts to change the resistors using SMT tweezers.



The scope software can be stopped using ctrl-alt-del which opens up the task manager on Win98 and restarted from the menu so testing all the combinations could be done quickly.
Code: [Select]
R3 R2 R1 R0
-  -  X  X  54835A
X  X  -  -  54845A

X = placed, - = open

I have not found the 54846A mode but I have not included R4 in my test because I didn't spot it when looking for potential candidates. For now I'm fine with the 54845A mode.

There where also two other things on my wishlist: more memory and a faster processor. Much to my surprise FIC still has an FTP server with documentation and BIOS files ftp://ftp.fic.com.tw/motherboard/. How cool is that? From the documentation I learned the maximum memory is 768MB divided over 3 modules of 256MB each. It also should support the AMD K6-3 450MHz processor which (ofcourse) is faster and according to legend among the fastest socket 7 processors ever made. So I went on Ebay and spend $18 +shipping on 3 256MB memory modules and $15 on an AMD K6-3 450MHz. The memory was advertised as new but I got 3 different modules and anti-static bags so I'm quite sure these are pulls. But hey, they where only $6 each. The biggest SDRAM module I had in my pile is 64MB so I guess back then 256MB ones would have been insanely expensive. The AMD K6-3 450MHz however did look brand new:



Installing the memory was straightforward. The processor proved more difficult. First of all I needed to dig up a Pentium-I cooler from my scrap heap because the original heatsink is glued onto the processor. Secondly with the new processor installed and setting the right core voltage & multiplier it wouldn't boot. Drats! After some Google-fu I found a forum message from the beginning of this millenium saying FIC has a beta BIOS version which supports the K6-3 450MHz. I found this in FIC's FTP server with version number JN4116. However the current BIOS version on the motherboard said JN4204 which I couldn't find in FIC's archive so this could be a special Agilent version. I also would expect version 4204 to be newer than 4116. I decided to give it a try and save the old version twice (and check!) before attempting to flash the BIOS with version 4116. End result: BIOS version 4116 works with the AMD K6-3 450MHz and so does the oscilloscope software! The performance increase (waveforms/s) is about 15%.

Running Windows XP
The PC part is powerful enough to run XP. There are three advantages of XP: it supports USB sticks, it runs the processor cooler and it can run the latest version of the software I found. Part of the process is also to install an SSD consisting of an mSata module in an mSata to 44pin IDE converter:

I also needed a CD-ROM drive so I threw that into the mix as well:


While doing this I ran into a problem. The XP installer would get either into a blue screen or it would not recognise the mSata drive. The first problem was caused by a slightly faulty CD-ROM drive the latter because the PCB which converts the 40 pin IDE into 44 pin IDE+power has been designed without any idea about signal integrity:


Using a UDMA IDE cable+seperate 40 to 44pin convert allowed XP to install so on to the software. Installing the software took a while and after installing it, it complained about needing a license. So I also installed the licensing server. After this the software starts but crashes because it can't find a DLL. Also the driver for the acquisition board is missing. I guess a seperate software package is needed which has the driver and a low level interface layer. I have not been able to locate either of those (except from an Ebay seller which asks nearly $500 for it). According to a post from Wuerstchenhund the new software is not a big improvement over the old software so I decided to go for plan B and use Win98 instead. So I saved an image from the XP install and copied the image from the original software onto the SSD.

Windows 98 and the original software seem to work just fine with the converter PCB in the middle so this is the final setup:

BTW: this picture shows another problem I didn't spot until I made this picture!

USB sticks & Win98
I still wanted to be able to use USB sticks to save screendumps and data to. Sure I could use the network for that but USB is easier. After some Googling I found this page http://www.technical-assistance.co.uk/kb/usbmsd98.php which has a software package to install USB stick support on Win98. After installing I had to re-install the driver for the USB sticks I had inserted earlier but after that it worked just fine.

Leaky capacitors!
This picture says it all:


This meant I had to dissasemble the PC part and thus take all the boards and the motherboard out:

I had some new low ESR capacitors lying around and for good measure I added a 4.7uf 100V ceramic.
After putting everything together the scope still worked so I guess the capacitor replacement went OK.

Time for some play
Let's try a 200MHz AM modulated signal first:


Add some color grading in the mix:


Let's try FFT and stacked math functions on a 200MHz FM modulated signal (+/-10MHz FM swing)

The yellow line is the base FFT trace. The green is the max hold and blue is the averaged FFT trace. I have not tried it but it should be possible to have more than one FFT trace.

I checked the risetime from a square wave. When the samplerate is 4Gs/s the (calculated) bandwidth is 1.1GHz. At 8Gs/s the bandwidth is 1.5GHz. Not bad!

What is left is to wait for the ADC hybrid to turn up and see which errors remain after I have fitted it. To be continued...
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: Jwalling, edavid, kado, Muxr, kridri, mash107, 1Ghz, JanNousiainen, fonograph, Hydron, alm, Pitrsek

Offline alm

  • Super Contributor
  • ***
  • Posts: 1245
  • Country: 00
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #1 on: July 08, 2017, 03:44:58 pm »
Thanks for that write-up!

Weird that that loose logic IC did not give any self-test errors. Maybe it was just interfacing with the dead ADC module, and hence the failure was being hidden?

I am surprised that upgrading the CPU resulted in an increase in waveform/s. I would have expected that to be handled completely in hardware that writes directly to the screen (DMA).

So there is no difference in the calibration procedure for the various models? I could imagine the adjustments for front-end flatness etc might be different. Nice upgrade, anyway.

Now fingers crossed that replacing the hybrid will indeed fix all the errors. It seems likely, based on the tests that you have performed, but you never know until you try it. It could always be something obscure where one of the surrounding components is marginal, barely working with some hybrids and barely failing with others.
 
The following users thanked this post: Johncanfield

Offline Tony_G

  • Frequent Contributor
  • **
  • Posts: 581
  • Country: us
  • Checkout my old test gear channel (link in sig)
    • TGSoapbox
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #2 on: July 09, 2017, 12:24:13 am »
In case you need it I saved a copy of the software folders from the Keysight FTP Server (they went missing last time I checked there)

https://1drv.ms/f/s!Amqar8_XQ9Uzj6YhN1xHdM8tQdKMOA

TonyG
 
The following users thanked this post: mash107, Converter, SpencerTC

Online TheSteve

  • Supporter
  • ****
  • Posts: 2962
  • Country: ca
  • GHz
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #3 on: July 09, 2017, 02:33:00 am »
Great post so far. You know I love a good bandwidth upgrade so I'm looking forward to the updates.  :-+
VE7FM
 

Offline Jwalling

  • Supporter
  • ****
  • Posts: 1240
  • Country: us
  • This is work?
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #4 on: July 09, 2017, 09:39:41 am »
When browsing on the hard drive I found an interesting help file: HP4XS.HLP . This seems to contain a partial guide to do component level diagnostics/repairs. Hopefully it is helpful if it turns out the trigger circuit is really defective.

Very nice find! I wish I knew about that file years ago!!
What on earth made you look through all the help files?

EDIT: I just took a look at some images I've saved off from this series of scopes. Given the same firmware revision, this help hp4xs.hlp file is the same one that's on the 54810A through 54825A scopes. There is also a hp1xs.hlp file, and I believe that this one is used for the lower end scopes up to the 54825A.
« Last Edit: July 09, 2017, 10:04:24 am by Jwalling »
Jay

System error. Strike any user to continue.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 17392
  • Country: nl
    • NCT Developments
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #5 on: July 09, 2017, 10:06:46 am »
When browsing on the hard drive I found an interesting help file: HP4XS.HLP . This seems to contain a partial guide to do component level diagnostics/repairs. Hopefully it is helpful if it turns out the trigger circuit is really defective.
Very nice find! I wish I knew about that file years ago!!
What on earth made you look through all the help files?
Being bored for 5 minutes and dumb luck made me click that file ;)
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 17392
  • Country: nl
    • NCT Developments
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #6 on: July 09, 2017, 10:26:58 am »
In case you need it I saved a copy of the software folders from the Keysight FTP Server (they went missing last time I checked there)

https://1drv.ms/f/s!Amqar8_XQ9Uzj6YhN1xHdM8tQdKMOA
Thank you! I'm downloading the archive right now. I found the FTP archives too and noticed the files where missing but I assumed my FTP clients where not compatible with the server.
« Last Edit: July 09, 2017, 11:31:30 am by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 17392
  • Country: nl
    • NCT Developments
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #7 on: July 09, 2017, 02:09:57 pm »
Update: I managed to update the software to version 4.3 thanks to Tony_G's archive. I see FFT phase has been added to the math functions as well as other minor improvements.

Upgrading from a USB stick instead of the LS120 drive needed some hacking using a hex editor on the oscilloscope program. The upgrade process always loads the upgrade from drive a: and you can't set it look on a different drive  :palm: . With help of the hex editor I changed all path (not text) references to a: into e: where the USB stick sits. After that it would load the software from the USB stick and do the upgrade process. I'll have to do the same to go from 4.3 to 4.5.

According to the documentation 4.5 is an incomplete package and meant to be installed on top of 4.3. Since my scope ran version 3.72 it seemed wise to install 4.3 first and then 4.5.
« Last Edit: July 09, 2017, 02:15:20 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline stj

  • Super Contributor
  • ***
  • Posts: 2156
  • Country: gb
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #8 on: July 09, 2017, 04:47:21 pm »
you cant use USB on the FIC VA503,
it has the first implementation - v1.0
1.0 is bugged and will not work right.
(hence USB v1.1)

stick a pci card in it for USB function.

one other thing about those boards, most have the pcb pads for a proper mini-din ps2 socket under the full-size din socket and mouse headers.
« Last Edit: July 09, 2017, 04:49:47 pm by stj »
 

Offline Jwalling

  • Supporter
  • ****
  • Posts: 1240
  • Country: us
  • This is work?
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #9 on: July 09, 2017, 04:54:05 pm »
you cant use USB on the FIC VA503,
it has the first implementation - v1.0
1.0 is bugged and will not work right.
(hence USB v1.1)

stick a pci card in it for USB function.

They work fine with thumb drives with the Win98 USB hack. I know because I've done it myself.
It's Win98 first edition that has the problem with USB storage.
Jay

System error. Strike any user to continue.
 

Offline stj

  • Super Contributor
  • ***
  • Posts: 2156
  • Country: gb
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #10 on: July 09, 2017, 05:20:03 pm »
no, your just getting lucky because your data packets over usb are short.
the via oppollo MVP3 chipset has buggy usb implementation - that's official.

i have that board in a tower - it served me 24/7 for very many years and still acts as a network backup unit.
there really isnt much i dont know about it.
inc the fact it can overclock the cpu at 125MHz if your ram is up to it. (7ns or faster)
 

Offline Jwalling

  • Supporter
  • ****
  • Posts: 1240
  • Country: us
  • This is work?
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #11 on: July 09, 2017, 05:31:19 pm »
no, your just getting lucky because your data packets over usb are short.
the via oppollo MVP3 chipset has buggy usb implementation - that's official.

i have that board in a tower - it served me 24/7 for very many years and still acts as a network backup unit.
there really isnt much i dont know about it.
inc the fact it can overclock the cpu at 125MHz if your ram is up to it. (7ns or faster)

One thing I do know: VIA chipsets were always  bug-ridden POS. In short, they suck! :--
I have a FIC VA-503 motherboard right now I could test with - how would I make the bug manifest itself?
Jay

System error. Strike any user to continue.
 

Offline stj

  • Super Contributor
  • ***
  • Posts: 2156
  • Country: gb
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #12 on: July 09, 2017, 05:38:18 pm »
transfer lots of data in a single pass.

writing a shitload of stuff to a usb key will often cause the interface to hang after while.
i imagine any streaming devices like tuners will cause absolute chaos.
btw, the first intel chipsets with usb 1.0 had the bug too - it was in the design spec they where supplied with!!!
oops!  :palm:
 

Offline lukier

  • Supporter
  • ****
  • Posts: 605
  • Country: gb
    • Homepage
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #13 on: July 09, 2017, 05:43:01 pm »
nctnico: is 54835A the same architecture as 54381D (https://www.eevblog.com/forum/testgear/agilent-54831d-modernising/)? I mean does it use the ancient C&T graphic card?

If yes then there should be still some, albeit limited, room for improvement (in the PC department), instead of relying on the Socket7 motherboard that I guess is not really good anymore (performance, bad caps).

LeCroy's are better in that respect, as they only use a single PCI card to communicate with the oscilloscope board so one can upgrade as far as Windows 7 32 bit can get (even i7 Haswell with PCIe to PCI adapter). Other connectivity is pretty standard, LVDS for the LCD (I've upgraded mine to XGA), standard signals for the CCFL inverter and, on WaveRunner 6K, the front panel is just a USB HID device (a bit problematic on newest hardware, doesn't like Intel root hubs, so one must use a separate USB controller).

Maybe, because of different PC integration architecture the CPU processing power is not as important as in LeCroy's case. Still, more modern CPU and motherboard with SATA and USB would be nice I guess. I have no idea how the Agilent software would react to upgraded PC internals though.
 

Offline Jwalling

  • Supporter
  • ****
  • Posts: 1240
  • Country: us
  • This is work?
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #14 on: July 09, 2017, 06:06:17 pm »
nctnico: is 54835A the same architecture as 54381D (https://www.eevblog.com/forum/testgear/agilent-54831d-modernising/)? I mean does it use the ancient C&T graphic card?

If yes then there should be still some, albeit limited, room for improvement (in the PC department), instead of relying on the Socket7 motherboard that I guess is not really good anymore (performance, bad caps).


Before anyone starts talking WinXP on the 54810A series.  ;)
The 54830/31/32A/D used 1GHz or 1.2 GHZ Pentium III processors on a newer platform without ISA slots. They did use the same C&T graphics driver chip, though. However, the ACQ controller is newer (In fact the older ACQ controller from the 54810A series is present but only used for power on functions, front panel control, and the scope's video frame buffer) I think upgrading to XP would be a problem as the driver for the ACQ controller would not want to talk to the old ACQ board.

EDIT: And I don't think the pin-out for the ribbon cable is the same anyway...

EDIT2: which wouldn't matter anyway. Duh. You'd have to find an XP driver for the OLD controller.

The picture shows the newer ACQ controller in red, the old one in blue

If one could find a newer motherboard (just to upgrade the hardware for performance) with an ISA slot (For GPIB functionality) it just might work.
« Last Edit: July 09, 2017, 06:19:22 pm by Jwalling »
Jay

System error. Strike any user to continue.
 

Offline Jwalling

  • Supporter
  • ****
  • Posts: 1240
  • Country: us
  • This is work?
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #15 on: July 09, 2017, 06:12:14 pm »
One other thing -  IME the FIC motherboard won't work with GB Ethernet cards for some reason either. I tried a Realtek 8169 based card which I had Win98 drivers for.
Jay

System error. Strike any user to continue.
 

Offline Jwalling

  • Supporter
  • ****
  • Posts: 1240
  • Country: us
  • This is work?
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #16 on: July 09, 2017, 06:43:53 pm »
Something I've always wondered, but have never figured out. What does the battery powered SRAM on the ACQ controller hold?

I've been too chicken to disconnect the battery from a working scope to find out... :scared:

Never seen one that was dead either.
Jay

System error. Strike any user to continue.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 17392
  • Country: nl
    • NCT Developments
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #17 on: July 09, 2017, 06:59:02 pm »
nctnico: is 54835A the same architecture as 54381D (https://www.eevblog.com/forum/testgear/agilent-54831d-modernising/)? I mean does it use the ancient C&T graphic card?

If yes then there should be still some, albeit limited, room for improvement (in the PC department), instead of relying on the Socket7 motherboard that I guess is not really good anymore (performance, bad caps).
The 54835A uses a C&T CT5550 graphics card. I'm not a fan (understatement) of AMD and VIA but I have to assume HP/ Agilent did some reliability tests before deciding to use this particular motherboard for a high end oscilloscope. I actually found a service note from Agilent from 2005 which describes a recall to replace the existing motherboard with the VA503-A and put a VIN33 (latest hardware version) sticker on the back of the scope. I think upgrading the motherboard can be done but the acquisition and videocard will need a PCI slot. And there is also the limitation of the AT style keyboard connector in the casing and how much power can the power supply deliver? All in all there won't be many options to make the PC part much faster.

@stj: I'll keep an eye open for USB problems. So far USB sticks are working OK and if not then I'm quite sure I have a USB-PCI card somewhere in an old PC.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline lukier

  • Supporter
  • ****
  • Posts: 605
  • Country: gb
    • Homepage
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #18 on: July 09, 2017, 07:15:45 pm »
I think upgrading the motherboard can be done but the acquisition and videocard will need a PCI slot. And there is also the limitation of the AT style keyboard connector in the casing and how much power can the power supply deliver? All in all there won't be many options to make the PC part much faster.

There should be plenty of more modern options, there are Core2Quad motherboards featuring 2 PCI slots. You only have to be brave enough to take a metal nibbler and cut the ATX IO shield opening (I have no idea why Agilent decided to do their own cutouts, seems silly) :) Often newer processors have better performance/power consumption ratio so I wouldn't worry about the PSU that much (and you've gained some watts by going the SSD route, CD drive can be disconnected as well).

However, maybe Agilent software depends heavily on a particular motherboard and/or BIOS as is the case in some older Tek scopes (which are even worse as they use very custom form factor motherboard).
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 7211
  • Country: us
    • SiliconValleyGarage
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #19 on: July 09, 2017, 07:17:48 pm »
no need to peel the front label off to remove that probe interface board.  from inside the scope : between bnc of channel2 and channel3 you will see 2 metal clips. simply press those and the netire pcb including the metal stiffner pops off. you don't need to muck with the screw nor destroy the front panel....


as for the battery ont he interface board : that holds the attenuator click count and the self diagnostic / calibration data. if that bettery goes empty simply rerun the diagnostic.

you can launch scope.exe using the service command  " scope.exe /service" that gives extra options in the self test menu where you can set attenuator click and othe roptions
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 
The following users thanked this post: nctnico

Offline stj

  • Super Contributor
  • ***
  • Posts: 2156
  • Country: gb
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #20 on: July 09, 2017, 07:20:34 pm »
there are 3 in the family,
the 503, the 503A and the 503+
i cant remember the differences,
but one has both AT & ATX power connectors and sockets for both 66MHz 72pin sims, and 100MHz sdr-dims.
(i have that model)
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 17392
  • Country: nl
    • NCT Developments
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #21 on: July 09, 2017, 09:32:11 pm »
no need to peel the front label off to remove that probe interface board.  from inside the scope : between bnc of channel2 and channel3 you will see 2 metal clips. simply press those and the netire pcb including the metal stiffner pops off. you don't need to muck with the screw nor destroy the front panel....
I looked again but with the attenuators fitted it is impossible to get to those clips. There is a square hole as well. Maybe that is to be used to push the probe connector panel out. BTW someone before me already peeled the sticker off in the places where it is mostly damaged. Maybe I can get things to improve by melting the glue a little bit.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Jwalling

  • Supporter
  • ****
  • Posts: 1240
  • Country: us
  • This is work?
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #22 on: July 09, 2017, 10:57:34 pm »
no need to peel the front label off to remove that probe interface board.  from inside the scope : between bnc of channel2 and channel3 you will see 2 metal clips. simply press those and the netire pcb including the metal stiffner pops off. you don't need to muck with the screw nor destroy the front panel....
I looked again but with the attenuators fitted it is impossible to get to those clips. There is a square hole as well. Maybe that is to be used to push the probe connector panel out. BTW someone before me already peeled the sticker off in the places where it is mostly damaged. Maybe I can get things to improve by melting the glue a little bit.

There is a little metal "nub" that you can push out from the inside with a small flat-blade screwdriver. I'd be surprised if there's no info about it in the service guide...
I'll take a picture tomorrow and post it.

Done. The metal nub is highlighted. It's right below the CH3 BNC.
Edit: Crap - that was the 54831B. I'll try again in a bit...

Here it is.
« Last Edit: July 10, 2017, 01:26:45 pm by Jwalling »
Jay

System error. Strike any user to continue.
 

Offline karkoon

  • Contributor
  • Posts: 35
  • Country: in
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #23 on: July 10, 2017, 02:45:20 am »
Nice write up. Thank you.


Typed with thumbs.
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Banned!
  • Posts: 2766
  • Country: gb
  • Occasionally active on the forum, available via PM
Re: Agilent 54835A scope (4 channel 1GHz / 4Gs/s) repair & uphack
« Reply #24 on: July 10, 2017, 09:50:33 am »
Nice write-up and pictures!  :-+

I have been hunting for a 1GHz+ scope for over half a year. I prefered one which has a high enough samplerate to do this in a single shot. Ofcourse it is 'easy' to spend several $k on a Lecroy Wavepro 7000 series or something similar but I didn't want to spend that much and avoid too much overlap with the oscilloscopes I already have. The older Lecroy LC584 is also interesting and I nearly got one but the Ebay seller bailed out on me (I did got a refund though).

Well that's bad. But there should be plenty of similar scopes on the market.

There recently was a WavePro 950 (1Ghz 16GSa/s) for less than £1k on ebay UK (maybe it still is) but I guess it depends on what you wanted to spend.

Quote
I also looked at the Tektronix TDS694C and TDS784 but the problem with these scopes is that they are older than the Agilent 54835A, have an obsolete NVRAM (SRAM + battery) inside and need special calibration software. The 54835A OTOH can calibrate itself using the auxilary output so it doesn't depend on calibrations values or options stored in a NVRAM. Big plus!

It's a nice feature, although most newer DSOs shouldn't show a noticeable parameter shift even more than a decade, unless something is broken.

Quote
My main application for the 54835A is looking at high frequency signals (digital and analog). The 54835A has a maximum depth of 65kpts on 2 channels and 32kpts on 4 channels and no peak-detect. I guess that if you don't want to spend much then deep memory and peak-detect are out of the window. OTOH the 54835A has a lot of tools for signal analysis like FFT, color grading and stacked/chained math functions.

Don't expect too much, especially on a Windows 98 unit. You get a bit more than with an older InfiniiVision scope, but it shows that these Infiniiums were HP's first attempts at doing a Windows scope. The 54835A was designed as a general purpose bench scope, not as an analysis scope, where its capabilities are more limited. The architecture is really super-slow, but the old Win9x models with Socket7 mobo even managed to strangulate this slow architecture through its even slower PC part (slow and buggy VIA chipset). Also, not all Win9x variants can be upgraded to XP, which is recommended so you can use the later software which contains more functions and lots of bug fixes.

Also, reliability wasn't exactly great on these Infiniiums. We had lots of them and lots of problems with dead ADC hybrids, acquisition board issues, trigger issues, issues with the interface card and dying PSUs. Agilent only really got a handle of that with the successor Infiniium DSO8000/DSO80000.

As other said upgrading isn't easy because of the different interface cards but frankly going beyond the K6-3 450MHz won't offer you much performance improvement as when you go to a P2 or faster the architecture becomes the limiting factor (and the Windows part is only used as display on these scopes). I'd still consider getting a P2 board with ISA slot to get rid of the buggy VIA chipset, though.

Depending on what you want to do, a DSO8104A (which with some patience can be found <$1500) might have been a better option (although even my DSO8064A's BW exceeds 1GHz), as it has a faster architecture without the antique C&T65500 secondary graphics adapter and hardware mixing, runs XP on a late P4 mainboard with PCIe, SATA and fast onboard gfx, is much more reliable, offers up to 128Mpts memory (unlockable via hack) and 32Mpts FFT, and even has Peak Detect which seems to be important for you. Also the XGA touch screen is a notable improvement over the VGA display of the 54800 Series, as traces look cleaner and you can get more information on the screen (although you probably still want to use a mouse).

« Last Edit: July 10, 2017, 09:56:00 am by Wuerstchenhund »
Brexit n - The undefined being negotiated by the unprepared in order to get the unspecified for the uninformed.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf