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

[Jwalling]

And then I find it.
https://community.keysight.com/message/69764

I'll put it here in case it moves again. Hope I don't mess this up...

Keywords: 54835A 54845A 54846A Understanding trigger specifications.

QUESTION:

Understanding trigger specifications
Question asked by deninnh on Jan 22, 2015
Latest reply on Jan 23, 2015 by deninnh

    Like •
    Show 0 Likes
    0
    Comment • 4

I've been using scopes for over 30 years up to the latest and thought I knew them fairly well.  I've purchased several high end scopes in the past and have to admit that trigger specifications were not high on my list of concerns.  They just worked !

After acquiring a personal 54846A which will not trigger above 1GHz on a clean sine wave (Agilent source), I decided I should read the spec.

It was surprising to find that trigger sensitivity for this 2.25GHz scope is only spec'd up to 1GHz.

So then what happens ?

Funny thing is that it *will* trigger on very fast events like an avalanche pulser or airborne ESD event (from a distance of course) with risetimes below 200ps.  But it just can't handle a sine wave.  I've looked at the source output and it is very clean spectrally.  On another 3GHz scope, the sine wave is perfect.

Is the scope broken?  All channels are identical in behavior.  Deleted cal files and ran full cal with no change.  Unit is very clean and has the last software version offered and also the last, fastest motherboard.

I realize that current  models have much improved triggering.  On another unmentioned brand of scopes I've always been able to trigger right up to the specified scope bandwidth before aliasing and other things kick in.   

REPLIES:

Generally, the HW Trigger BW of scopes, in the past, has been approximately half the scope BW. There are also sensitivity specifications, so depending on the signal magnitude, you can trigger on higher BW signals. Newer Agilent/Keysight scopes have a "Trigger Sensitivity" button, that can give higher BW triggering.

That said, there are also digital scopes that 'fake' a higher BW. They use a hidden 'Auto' trigger mode (which forces a trigger if there's no trigger event within a short period of time), then look for an edge near the center of the trace in SW, and move that to the middle of the display. If it happens fast enough, it is indistinguishable from a 'real' trigger. It's very hard to determine if this is being done.

You don't indicate the brand on the 3 GHz scope, or the frequency of the sinusoidal signal you used to test it, so I can't comment on that part .

Al

Like •
Show 0 Likes
0
Actions

    deninnh
    deninnh @ algoss on Jan 23, 2015 12:38 PM
    Hi Al,

    Thanks for the quick reply.

    I was applying a clean sine wave between 1GHz and 2.5GHz, but suspended the BW test of my "new" scope when it wouldn't get past 1GHz.  (it will actually trigger at all up to around 1.2GHz but with much jitter on the trace).

    Last night I ran the trigger sensitivity test in the service manual and it passes, but tests are only at 100MHz,  500MHz and 1GHz.

    The "other" was a TDS694C and it delivers a cleanly triggered waveform almost up to 3GHz (its BW).

    SInce there is no spec beyond 1GHz for this model, I may not have a leg to stand on for returning this scope to the dealer.

    We have a 3000X (1GHz) and 4000X (2.5GHz) at work and I noticed that the trigger spec no longer mentions at what frequency (as far as I recall).  Likewise for the other brand.  It is Y range sensitive but not frequency dependent.  These models are converting me to Agilent but my home lab budget says "old Agilent".

    So, if this scope triggers "ok" at 1GHz, would you say that was normal for the model?

    Thanks for your help.

    Den
    Like •
    Show 0 Likes
    0
    Actions
        algoss
        algoss Employee @ deninnh on Jan 23, 2015 12:56 PM
        Yes, I would say that since it triggers at 1GHz, it is working as specified. The fact that it's "jittery" at 1.2 GHz is proof, at least to me, that it's a real trigger, not one of the "fake" triggers I mentioned.

        The 3000X and 4000X only talk about trigger BW in terms of trigger sensitivity at different bandwidths.

        For many measurements that people are making with higher speed scopes, Edge trigger BW becomes less important. When looking at jitter, for instance, the trigger location is not part of the equation at all. The only thing that matters is the timebase, and the location of the edges.

        Al
        Like •
        Show 0 Likes
        0
        Actions
            deninnh
            deninnh @ algoss on Jan 23, 2015 3:27 PM
            Thanks again Al. 

            I decided to be practical about this and try it in a real world application, USB HS verification.  The trigger was good enough but of less importance since the test is done on data from a single sweep, triggered by a pulse width condition.

            It worked flawlessly and blows away the old TDS in terms of ease of use and fast storage (good old 1GB flash drive under Win98).

            Den

[nctnico]

I'm aware of that limitation. It still is a good buy IMHO. I bought the 54835A specifically for looking at digital signals and it does that OK. It already has done some work to pay for itself.

[eurofox]

Here's the 54846A. Just two resistors need to be de-populated, so my drawing is at least correct for the 54846A.

I have a fully working 54845A that pass all test and calibrate without problem and I did the change to make it a 54846A.
It detect that it is a 54846A but testing give a lot of error and finally crash!

This mean that software is different and maybe attenuators and other parts …

It could be as well that during installation of the software that is encapsulated in the installation software different versions are installed depending of the resistor selecting the model.

[Jwalling]

Here's the 54846A. Just two resistors need to be de-populated, so my drawing is at least correct for the 54846A.

I have a fully working 54845A that pass all test and calibrate without problem and I did the change to make it a 54846A.
It detect that it is a 54846A but testing give a lot of error and finally crash!

This mean that software is different and maybe attenuators and other parts …

It could be as well that during installation of the software that is encapsulated in the installation software different versions are installed depending of the resistor selecting the model.

It's not the software. I have one Symantec Ghost image that works in all the scopes in this series with the FIC motherboard.
The scopes are 54810A, 54815A, 54820A 54825A, 54835A, 54845A, and 54846A.

[eurofox]

Here's the 54846A. Just two resistors need to be de-populated, so my drawing is at least correct for the 54846A.

I have a fully working 54845A that pass all test and calibrate without problem and I did the change to make it a 54846A.
It detect that it is a 54846A but testing give a lot of error and finally crash!

This mean that software is different and maybe attenuators and other parts …

It could be as well that during installation of the software that is encapsulated in the installation software different versions are installed depending of the resistor selecting the model.

It's not the software. I have one Symantec Ghost image that works in all the scopes in this series with the FIC motherboard.
The scopes are 54810A, 54815A, 54820A 54825A, 54835A, 54845A, and 54846A.

It is working in the 548446A mode not calibrated, I did not check the performance to check if the modification work, the software ask to do a calibration what I try and fail on all channels, I start again with a test and got a lot of errors and finally the software crash.
It cross my mind to replace the 0 ohm resistor with a switch wired to the back of the scope because it is not a good idea to "play" with the board due to EMC and connector issues but just put the resistor back, 1.5Ghz is not that bad 

[Converter]

In some miraculous way, my Agilent 54845A turned into 54845B (at least so it is now shown).



I made a motherboard update and a software update v.4.5 for 54845B. I tried different versions of software, but only with this I got success.

Previously, there was an Atlas PCI-III Series 757 motherboard based on Socket 7 and AMD K6-2 300 MHz processor and 64 MB RAM.

Now the most advanced motherboard for AT-form factor with Intel 440BX chipset and Socket 370, with 133MHz bus, where the Coppermine core Pentium III processor can be installed, up to 1133MHz (or with a small upgrade of Tualatin 1400MHz) and 3x256 MB of RAM max.

The new motherboard is designated Tomato (ZIDA) 60-cbxict1
Next to her on the photo is an old motherboard Atlas Series 757


I also had to replace the Ethernet controller for the PCI slot








At the moment I used a P III - 733MHz processor.
Subjectively, now I noticed a significantly smaller loss of the frequency of updates to the oscillogram in the mode of activation of measuring functions, than it was before.







p.s.
Now I found information that the manufacturer installs in the 54845B motherboard "VP22 Motorola" (Socket 370) and processor Celeron/Pentium III - 866 (Coppermine). Thus, we can assume that I received the same without changing the form factor AT. The only difference is that I still do not have a built-in CD-ROM, but it's not required too often.

[Jwalling]

In some miraculous way, my Agilent 54845A turned into 54845B (at least so it is now shown).

I made a motherboard update and a software update v.4.5 for 54845B. I tried different versions of software, but only with this I got success.

Previously, there was an Atlas PCI-III Series 757 motherboard based on Socket 7 and AMD K6-2 300 MHz processor and 64 MB RAM.

Now the most advanced motherboard for AT-form factor with Intel 440BX chipset and Socket 370, with 133MHz bus, where the Coppermine core Pentium III processor can be installed, up to 1133MHz (or with a small upgrade of Tualatin 1400MHz) and 3x256 MB of RAM max.

The new motherboard is designated Tomato (ZIDA) BXv98-CU/v693

Are you sure about the new motherboard part #? I ask because the images for it all show a Via chipset: https://www.google.com/search?q=BXv98-CU/v693&tbm=isch&tbo=u&source=univ&sa=X&ved=0ahUKEwit6P26rbbXAhWKy4MKHSNOBswQsAQILg&biw=1518&bih=934

One on eBay has a Via chipset as well: https://www.ebay.com/itm/Tomato-Board-ZIDA-BXv98-CU-V693-Socket-370-Motherboard-4-PCI-2-ISA-AGP-slots/182114467065?hash=item2a66de3cf9:g:JG4AAOSwIjNXJcE3

[Converter]

Hi Jwalling,
It looks like there really is a ZIDA motherboard with exactly the same external design, but with VIA chipsets. But on my board a set of Intel chips is actually installed. On my board there is only an inscription: 60-cbxict1 30-00 (I edited my previous post).
The manual was not found, but the display shows: -440BX-
BIOS allows to overclock the processor bus up to 150 MHz, which corresponds to the specification 440BX.

[Jwalling]

Hi Jwalling,
It looks like there really is a ZIDA motherboard with exactly the same external design, but with VIA chipsets. But on my board a set of Intel chips is actually installed. On my board there is only an inscription: 60-cbxict1 30-00 (I edited my previous post).
The manual was not found, but the display shows: -440BX-
BIOS allows to overclock the processor bus up to 150 MHz, which corresponds to the specification 440BX.

Thanks for the clarification. Have you thought about upgrading it to a 2.25GHz 54846A/B? 

[Converter]

Thanks for the clarification. Have you thought about upgrading it to a 2.25GHz 54846A/B? 

Yes, I want to follow your example. In any case, I will try. Has someone succeeded in the calibration procedure after changing 54845 to 54846?

I also have a couple of questions: how to get access to the Windows desktop in this device? What drivers are needed to use usb flash memory to save screenshots in it?

[eurofox]

Thanks for the clarification. Have you thought about upgrading it to a 2.25GHz 54846A/B? 

Yes, I want to follow your example. In any case, I will try. Has someone succeeded in the calibration procedure after changing 54845 to 54846?

I also have a couple of questions: how to get access to the Windows desktop in this device? What drivers are needed to use usb flash memory to save screenshots in it?

I did the removal of the resistor to change my 54845 to 54846 but many errors in test and crash finally the test program.

Calibration attempt fail immediately.

Windows function is easy just by "Ctl key" and "esc key" on the keyboard.

I got my usb working but don't think I used a special driver, I suppose it is available in Windows 98.

I collected a usb driver that I find just by google somewhere on the net but did not use it.

[Jwalling]

Thanks for the clarification. Have you thought about upgrading it to a 2.25GHz 54846A/B? 

Yes, I want to follow your example. In any case, I will try. Has someone succeeded in the calibration procedure after changing 54845 to 54846?

I also have a couple of questions: how to get access to the Windows desktop in this device? What drivers are needed to use usb flash memory to save screenshots in it?

I think you'll be the first to upgrade to a 54846 successfully if it works!

Is this the right manual? http://www.elhvb.com/mboards/zida/manuals/CreatN10.zip
(Manual covers a slot processor version as well)

There is an "upgrade" for Win98 USB thumb drive support out there. It does work on these scopes.
It's been quite awhile since I did it, but IIRC it was this link that worked for me: http://www.technical-assistance.co.uk/kb/usbmsd98.php

The desktop on these is weird, I've never been able to customize it, but I didn't spend much time on it either.
I don't think you can minimize the scope app, only kill it with CTRL ALT DEL or possibly with CTRL ESC and end it from the taskbar.

[Converter]

Jay, your manual seems to fit (page 14).

I tried to reset the jumper, as you described. It really does now show 54846B. However, the calibration was not successful for each of the 4 channels. I do not know what is the reason, I think there are hardware differences that do not allow to make an upgrade 54845 to 54846 (or fine-tune the input amplifiers).

eurofox, thanks, I'll try to do it. I correctly understand that now you have the opportunity to use a usb flash to save the screenshot files?

[eurofox]

Jay, your manual seems to fit (page 14).

I tried to reset the jumper, as you described. It really does now show 54846B. However, the calibration was not successful for each of the 4 channels. I do not know what is the reason, I think there are hardware differences that do not allow to make an upgrade 54845 to 54846 (or fine-tune the input amplifiers).

eurofox, thanks, I'll try to do it. I correctly understand that now you have the opportunity to use a usb flash to save the screenshot files?

Yes I have replaced the floppy disk by 2 usb connectors and I can copy files.

[Converter]

Agilent program is more convenient to close "CTRL" + "F4".
In the original set, Windows did not have a suitable driver for flash memory. I installed this driver: http://www.technical-assistance.co.uk/kb/win98fe-usb-mass-storage-drivers.php
Now there's a removable disk (:E) and it works great.

Maybe someone managed to upgrade to Windows 2K?

[Jwalling]

Maybe someone managed to upgrade to Windows 2K?

That's the one thing I think is impossible. The problem is that there's no 2K drivers for the acquisition controller.

It's surprising that the 54845A can't be upgraded to a working 54846A. All the magic in these scopes are in the attenuator assemblies and A/D converters. Both models use the same parts - the A/D converters are part #1NB7-8353 and the Attenuator modules are 54512-63402 for both models. There must be some other minor modification needed...

I still have this 54846A in my possession, and I have some scrap 54845A ACQ boards laying around. Maybe I'll do a in-depth comparison between the two today if I get a chance.

[Jwalling]

Curiosity got the better of me and I couldn't resist waiting.

Here's another difference I found on the solder side under PGA device U91

Two resistors are not installed on the 54846A!

[Converter]

That's the one thing I think is impossible. The problem is that there's no 2K drivers for the acquisition controller.

There is one guy on this forum who was able to modify the WinXP drivers for the oscilloscope LeCroy WavePro7k, that they now work in Win7. I do not exclude anyone could do or have already done similar and the drivers Agilent Win98.

It's surprising that the 54845A can't be upgraded to a working 54846A. All the magic in these scopes are in the attenuator assemblies and A/D converters. Both models use the same parts - the A/D converters are part #1NB7-8353 and the Attenuator modules are 54512-63402 for both models.

Yes, I'm also surprised at this. I tried to restart the calibration 3 times and always got a negative result. In this case, the test of the input channels gave an error almost instantly, as if he had seen forgery in advance.

Here's another difference I found on the solder side under PGA device U91
Two resistors are not installed on the 54846A!

Are you sure that the presence of these resistors is essential? It would be good to check this: remove these resistors and try to do the calibration procedure without them. If you get the same negative result as I do, it will be a breakthrough (by the way, what is the value of these resistors?).

[Jwalling]

Are you sure that the presence of these resistors is essential? It would be good to check this: remove these resistors and try to do the calibration procedure without them. If you get the same negative result as I do, it will be a breakthrough (by the way, what is the value of these resistors?).

What I'm saying is that the 54846A does not have the resistors installed so there's nothing to remove. Your 54845A (if I'm right) has them installed. They are 261 Ohms in the 54845A.

Looking a bit closer now; there may be a little more to this:
The other two resistors right next to those that are installed in both boards have different values as well.
On the 54845A they are 90.9 Ohms (marked in blue). In the 54846A (also marked in blue) they are 68.1 Ohms.

[Converter]

Are you sure that the presence of these resistors is essential? It would be good to check this: remove these resistors and try to do the calibration procedure without them. If you get the same negative result as I do, it will be a breakthrough (by the way, what is the value of these resistors?).

What I'm saying is that the 54846A does not have the resistors installed so there's nothing to remove. Your 54845A (if I'm right) has them installed. They are 261 Ohms in the 54845A.

Looking a bit closer now; there may be a little more to this:
The other two resistors right next to those that are installed in both boards have different values as well.
On the 54845A they are 90.9 Ohms (marked in blue). In the 54846A (also marked in blue) they are 68.1 Ohms.

I made changes according to your description. Removed 2 resistors 261 Ohms and replaced 2 resistors 91 Ohm by 68 Ohm. Unfortunately, after this absolutely nothing has changed. The calibration process is still without success. So I returned everything to its original state.

[nctnico]

To me it seems these resistors are some kind of termination network for signals so it doesn't surprise me that nothing changed.

I think it is time to take hires pictures of boards from the 845 & 846 and subtract them in photo editing software to highlight the differences.

[Converter]

I investigated the bandwidth of my HP54845 in the factory settings. It seems that it has more than 2 GHz without any upgrade. As the source of the broadband signal, I used an industrial noise generator on avalanche-passing diodes (0.4-4 GHz). This is a very primitive device and there is a noticeable frequency unevenness in the spectrum of the generated signal. Indeed, in its generated spectrum there is a rise in the region of 100-300 MHz and a dip at 1.8 GHz. Nevertheless, this allows us to approximately estimate the bandwidth of the oscilloscope.

[nctnico]

The measurements I did showed my uphacked 54845 is just right by looking at the risetime. Unfortunately these scopes won't trigger on signals over 1GHz so maybe your noise measurement shows a more complete picture.

[Converter]

I added screenshots.
Unfortunately at the moment I do not have a high-quality microwave generator to make measurements more accurately (yes, I know about the limitations of the trigger). But a rough estimate can be made using this noise generator. There were used 2 scales: 10 dB/div and 5 dB/div. I have roughly noted positions with cursors corresponding to a -3 dB change in the level. This corresponds to a bandwidth of 2.127 GHz.

[Converter]

This oscilloscope HP 54845 is not inferior to the bandwidth of my LeCroy WavePro7200.
For comparison, I attach several measurements made in the same way with other of my oscilloscopes (FFT 10dB/div):
LeCroy WavePro7200;
LeCroy WavePro7300A;
LeCroy DDA-120;
LeCroy LC564DL;
Tektronix TDS5054B.

I also want to note that the HP 54845 demonstrated a restart speed of the trigger, which reaches 92 us (in the case of the easiest acquisition settings). This corresponds to processing 11K oscillograms per second, which is 30-50 times faster than all oscilloscopes from my list above (3-5 ms minimum). Faster than this, only the FastACQ-mode in Tek 5054 (7 us) - but this mode is not fully functional, and my analog oscilloscope LA354 (which reaches 1 us).