Agilent 54831D modernising

[Wuerstchenhund]

Regarding Windows, that's true for the first generation 54810A, 15A, 20A, 25A, 35A, and 45A. The N versions were for the military (US Navy), but there was no difference in them hardware-wise. The N versions could not come from Malaysia, only the US or Singapore. However, on the second generation scopes (54831B etc.) The M versions for the military (US Marines) came with Windows 98, while the rest were shipped with Windows XP. Once again, the military versions could not come from Malaysia.

The B variants like the 54831B were initially also sold with Windows98 and only got XP shortly after (starting with a specific S/N). We had a few of the very early ones (some of the first units available in Europe) back then.

As you say the military versions were merely regulatory variants (limited to certain countries of origin) and not really a different product with differing hardware. Later M versions were also shipped with XP, similar to the commercial variants, although it took longer for the M variants due to the outstanding XP accreditation.

The WindowsXP driver for the C&T 65550 is pretty much a repackaged NT4 driver.

That I'm not so sure about...

Just try to get the XP driver to work correctly in a fully ACPI-compliant system (it works in Plug'n'Play compatible PCI systems, aside from ISA/EISA/VLB/MCA with their respective ressource allocation schemes), or to support power management that goes beyond DPMS (i.e. Standby/STR, or the various ACPI Sx modes) without hacks/workarounds. It's not supported because the driver pretty much only supports what was in NT4, because at the core it is the same NT4 driver.

[Howardlong]

Here's that pic of the hacked IO shield compared to a clean unit. This is not a procedure I fancy doing. I am currently have a shallow microATX on an adapter base, and use appropriate internal risers as necessary. I did try a mini ITX and a DTX board with single or dual PCIe to multi-PCI adapters but it wasn't very successful. Using three individual PCIe 1x to PCI adapters with risers has so far proven much more resilient in terms of card power distribution and signal integrity.

[Jwalling]

The B variants like the 54831B were initially also sold with Windows98 and only got XP shortly after (starting with a specific S/N). We had a few of the very early ones (some of the first units available in Europe) back then.

Interesting. I've never seen Win98 on the B versions, and I've worked on many of them. Your first-hand experience trumps mine. Thanks.

[Howardlong]

I had another go last night using a miniATX board, Gigabyte H110M-DS2, and three PCIe to PCI risers. Processor I used was already installed, an I3-6320, but I am sure any supported processor will work.

Unlike previous attempts, I used an original Agilent recovery image expanded to 60GB. This is based on XP SP2 and Infiniivision 3.70 from about 2003/2004. I've never been able to make the documented recovery process work, so to recover that image, I used an XP VMWare VM on another machine with the scope's disk connected to a PATA to USB adapter, and attached the disk to the VM as a physical device. From there I could access the .GHO image. Booting an ISO of my own copy of Ghost 9.0 from many years ago, again in a VM, I restored the .GHO to another SSD and made sure it would boot on the scope. At this point I made an Acronis True Image image of the SSD.

Backing up and restoring disk images with new hardware is about ten times faster than trying to do it on the scope with its original hardware, even in a VM.

A fiddly thing about using a newer mobo is that I found that the UEFI graphical BIOS setup and boot selection doesn't always work with the old CT65550 graphics board. I had to keep removing and reinserting the display board pair as appropriate and use the board's integrated graphics when I needed to boot off my USB drive. I could try setting various parameters in the UEFI BIOS but there are many options, and I chose to leave that for another time, and I've been down that rabbit hole already without much luck. I also wanted to see how far I could go with default CMOS settings.

All I needed to do to boot the restored image was to retrofit the Skylake SATA drivers from Win-raid http://www.win-raid.com/t11f23-Modded-Intel-AHCI-and-RAID-Drivers-digitally-signed.html ("Universal 32bit Intel RST AHCI+RAID driver v11.2.0.1006 mod+signed by Fernando"). I used Acronis True Image 2016 Universal Restore to retrofit the AHCI drivers. There was no messing about with any ACPI settings that I've needed to do before.

You may find you can achieve the same thing by slipstreaming the Skylake AHCI drivers into an XP distribution using something like nLite, and doing a repair install, but this will take quite a bit more time.

After that, I loaded the various Agilent specific drivers from a CD I'd made up with the .infs etc (see OP).

I ran the postsysprep.bat batch file in the C:\PciFilter directory from a command prompt as Administrator with execute protection disabled, and allowed a reboot.

At this stage, the scope app works, as do ACPI power functions. There are about five motherboard drivers missing, most notably no USB, Ethernet or HD audio plus a couple of other bus related drivers. I'll deal with those later once the OS is upgraded.

One further note, the OS is somewhat unsurprisingly requesting activation now, for which I have three days' grace. Hopefully it will allow me to to upgrade to Vista first before needing to do that.

I recommend using a modern SATA attached DVD drive at this stage as I have found the one in the scope doesn't read all disks, such as DVD+R for example, one of those irritations of the dim and distant past!

[Howardlong]

I did some more mostly dead end work on this today except for the following...

O As speculated, the existence of a CT65550 board pair does indeed directly determine whether it's detected as an 5483x or 8000 series when running the 5.71 software on a recovery XP install. The scope will run on the VP22 onboard graphics at XGA, showing as an MSO8104A, but at about 5-10 screen updates/s. Reinstalling the CT65550 makes it run rather more fluidly on the CT65550 board, but it only measure s about 220 wfm/s even so.

O On the stock VP22 motherboard running the recovery XP but with the 5.71 scope, and without the CT65550 board, despite it only running at 5-10 screen updates/s with the onboard 815 graphics, and recognised as an MSO8104A, it is rock solid in XGA, and my previous comments about the flakiness on a Skylake build when running equivalent time sampling could not be reproduced.

[Howardlong]

I revisited this yesterday, trying three motherboards from three generations of chipsets with CPUs biased towards single threading speed with reasonably low TDPs, each using integrated/chipset graphics rather than the CT65550 PCI graphics & overlay combo cards, and with an integrated PCI slot.

The software install was using the recovered XP SP2 (plus Infiniium 5.71 software*) with appropriate AHCI, ACPI and HAL adjustments, and, apart from the Skylake board, all the XP drivers loaded.

Once installed with the GPIB/scope PCI interface board, the eight or so drivers were installed, and the scope app started up in all cases without any further adjustments. The scope app runs at either 640x480 or 1024x768. the scope is recognised as an MSO8104A, I assume because the CT65550 isn't present.

If this works I'll look at getting a conversion from VGA or DVI to LCD. Already I have a couple of Mitsubishi XGA 8.4" screens with appropriate VGA & DVI/HDMI interface adapters, but I haven't done a fit check yet.

In all three cases, previous problems that I'd encountered earlier such as with equivalent time or trigger jitter were not present. This was actually part of the test, I was suspecting that there was some contention in timing between the three PCI boards and the external PCIe-PCI bridges I'd been using.

The test signals I used were a 1MHz sine wave 100% AM modulated at 1kHz, and a <33ps TDR pulse from an HP 54121A.

E8600 Core 2 Duo (3.33GHz) LGA 775 Intel DG41AN mini-ITX

This seemed to work the best out of all three options, but also is the slowest performing, although compared to the 1GHz Pentium III on the original motherboard it's like night and day. There were no faults as far as I could ascertain with the limited testing I did, but I'd prefer to do more extensive testing. This board had the only "native" PCI slot in the test, directly supported by the chipset and not requiring a PCIe-PCI bridge.

i3-3250 Ivy Bridge (3.5GHz) LGA 1155 Gigabyte GA-H61N-D2V mini-ITX

The screen update rate is good enough although at first it seemed to have a few short pauses which disappeared later in the test. I'm not sure why this would be, the unit wasn't connected to the LAN, perhaps it was the indexing service, but this was using an SSD. Some screen corruption was present in the measurements section, with the area frequently not being updated: minimising and re-maximising fixed it, but the problem was frequent enough to make it pretty unusable in its current state. The PCI slot is via a PCIe-PCI bridge on the motherboard as the H61 chipset does not support PCI directly.

i3-6320 Skylake (3.9GHz) LGA 1151 on an MSI H110M Pro VHL micro-ATX

This works but the screen update rate is extraordinarily slow, about once every couple of seconds, so unusable in this state. I am not sure if this is because there are a large number of motherboard drivers, including integrated graphics, that aren't available in XP. The underlying waveform update rate is much more reasonable. The PCI slot is via an on-board PCIE-PCI bridge as Skylake chipsets don't support native PCI.

I couldn't get the right HAL/ACPI to properly control the soft power settings, although if the OS is upgraded this would be resolved.

Summary

The next steps are to see if some the performance problems on the Skylake are resolved by upgrading to Windows 7 and using proper drivers. Bearing in mind that so much on the motherboard just doesn't work on Skylake running XP, such as USB, upgrading the OS on the platform is pretty much a necessity anyway.

I also have concerns as to whether the option keys will work any more bearing in mind the scope now thinks it's an MSO8104A and not a 54832B.

*Edit to specify I was using builds based on the recovery build plus Infiniium 5.71 scope software.

[Berni]

Interesting how widely different it behaves on different hardware.

The license keys are purely based on the serial numbers on most scopes. Id assume its the same on these old infiniiums. From digging around in the new infiinium software it seams like the serial number might be stored on the acquisition board, but i have not tried a clean install of the software to know for sure. The software holds the license keys in plain text file somewhere in the win 7 "User Data" directory, so its easy to pull them off if you like, since this is considered legacy licenses id assume the old infiniiums do this too (They are moving to FlexLM floating licenses).

[free_electron]

the virtual memory trick may have to do with these machines still creating a ramdrive ....

[Howardlong]

the virtual memory trick may have to do with these machines still creating a ramdrive ....

Hmm, intriguing, I may be living under a rock, but what virtual memory trick are you referring to?

[Jwalling]

the virtual memory trick may have to do with these machines still creating a ramdrive ....

Hmm, intriguing, I may be living under a rock, but what virtual memory trick are you referring to?

I think he may be referring to the previous generation Infiniium (54825A, etc.) that ran Windows 98 and came with 64MB of memory. The scope application would offload to a 16MB ram drive so the hard drive could be put into standbye.

[free_electron]

by te wy : latest frimware for the 5483x machines is here : http://www.keysight.com/main/software.jspx?ckey=566904&lc=eng&cc=US&nid=-32412.536882503&id=566904

( a bit hard to find on the new site ... )
they also have older revs on that page ...

[Howardlong]

Some more older stuff here: ftp://ftp.keysight.com/pub/callpub3/ddt/scope/548xx/Upg_Rec/

By the way I neglected to mention that I upgraded the recovered partition to 5.71, I've since added that to the steps.

[Howardlong]

Here's a further update regarding the use of integrated graphics rather than the 65550 PCI card pair.

It is now clear that the slow update rate is caused by using the OS's generic display driver. This means that for Skylake, which doesn't have a full suite of XP (or Vista) drivers, you need to upgrade to Windows 7.

Note that the only supported in-place upgrade from XP Pro to Windows 7 is XP Pro -> Vista Ultimate -> Windows 7 Ulimate. I've been unable to to get a fresh install OS to work, but I may try again at some point. I have also now been able to take my current working XP VP22 image and get it working on a Skylake board, upgrading it to Windows 7.

Furthermore, there seems to be some incompatibility between running the 548xx PCI GPIB/acquisition cards with the agscope app in 1024x768 mode. This first manifested itself a lot earlier where I'd noticed that ETS was showing nonsense, but I hadn't figured out why. To make matters worse, sometimes it does work at 1024x768, so I spent quite a bit of time fiddling about with fixing PCI allocations etc but I now believe that when it worked it was by luck, and memory is being incorrectly accessed at the higher resolution. The same applies to all operating systems, from XP to Windows 7.

Using 640x480 resolution while using integrated graphics works flawlessly. Owing to the application's font resizing between XGA and CGA, somewhat surprisingly, subjectively I couldn't identify a huge amount of benefit going to XGA anyway.

The waveform update rate increases from a maximum of about 300Wfm/s in the factory config (PIII 1GHz with PCI 65550 graphics) to 7kWfm/s with a Skylake i3-6520 at 3.9GHz on integrated graphics.

The benefit of using the integrated graphics is that unlike the 65550 display, it supports intensity grading.

With some external PCIe to PCI riser bridges, I've seen alot of trigger jitter, but have never seen it on motherboards with an integrated PCI slot, either through an on-board PCIe-PCI bridge or a PCI connection direct from the southbridge.

The built-in 8.4" 640x480 display, which is parallel, not LVDS, is Agilent part number 2090-0396, aka Mitsubishi AA084VB02. Getting VGA, DVI or HDMI controller interfaces for this doesn't seem too easy, although it would make a nice FPGA project which I haven't ruled out, but I'm quite time poor at the moment.

The next step is to see if I can slip in an off the shelf VGA/DVI/HDMI 8.4" LVDS display/controller board combo at 640x480 resolution.

[Berni]

That looks quite promising by now.

Parallel RGB bus display panels are the easiest to convert from other formats. There are single chip solutions for making RGB out of HDMI/DVI, LVDS, VGA etc. We did that a lot since a lot of modern embedded CPUs don't have RGB buses anymore.

Industrial motherboard with LVDS and a chip to convert to RGB is probably the most elegant. I been looking into those for potentially upgrading my MSO9000 (LVDS panel in it). I found that in most of my cases a adapter cable is still necessary to adjust the pin out and connector type since there appears to be very little standardization around these connectors. Even if the connector seams to fit its best to double check as they tend to have +12V on them(For backlight and stuff). Those pins getting wired wrong would certainly give you a bad day.

[bigeblis]

I revisited this yesterday, trying three motherboards from three generations of chipsets with CPUs biased towards single threading speed with reasonably low TDPs, each using integrated/chipset graphics rather than the CT65550 PCI graphics & overlay combo cards, and with an integrated PCI slot.

The software install was using the recovered XP SP2 (plus Infiniium 5.71 software*) with appropriate AHCI, ACPI and HAL adjustments, and, apart from the Skylake board, all the XP drivers loaded.

Once installed with the GPIB/scope PCI interface board, the eight or so drivers were installed, and the scope app started up in all cases without any further adjustments. The scope app runs at either 640x480 or 1024x768. the scope is recognised as an MSO8104A, I assume because the CT65550 isn't present.

If this works I'll look at getting a conversion from VGA or DVI to LCD. Already I have a couple of Mitsubishi XGA 8.4" screens with appropriate VGA & DVI/HDMI interface adapters, but I haven't done a fit check yet.

In all three cases, previous problems that I'd encountered earlier such as with equivalent time or trigger jitter were not present. This was actually part of the test, I was suspecting that there was some contention in timing between the three PCI boards and the external PCIe-PCI bridges I'd been using.

The test signals I used were a 1MHz sine wave 100% AM modulated at 1kHz, and a <33ps TDR pulse from an HP 54121A.

E8600 Core 2 Duo (3.33GHz) LGA 775 Intel DG41AN mini-ITX

This seemed to work the best out of all three options, but also is the slowest performing, although compared to the 1GHz Pentium III on the original motherboard it's like night and day. There were no faults as far as I could ascertain with the limited testing I did, but I'd prefer to do more extensive testing. This board had the only "native" PCI slot in the test, directly supported by the chipset and not requiring a PCIe-PCI bridge.

i3-3250 Ivy Bridge (3.5GHz) LGA 1155 Gigabyte GA-H61N-D2V mini-ITX

The screen update rate is good enough although at first it seemed to have a few short pauses which disappeared later in the test. I'm not sure why this would be, the unit wasn't connected to the LAN, perhaps it was the indexing service, but this was using an SSD. Some screen corruption was present in the measurements section, with the area frequently not being updated: minimising and re-maximising fixed it, but the problem was frequent enough to make it pretty unusable in its current state. The PCI slot is via a PCIe-PCI bridge on the motherboard as the H61 chipset does not support PCI directly.

i3-6320 Skylake (3.9GHz) LGA 1151 on an MSI H110M Pro VHL micro-ATX

This works but the screen update rate is extraordinarily slow, about once every couple of seconds, so unusable in this state. I am not sure if this is because there are a large number of motherboard drivers, including integrated graphics, that aren't available in XP. The underlying waveform update rate is much more reasonable. The PCI slot is via an on-board PCIE-PCI bridge as Skylake chipsets don't support native PCI.

I couldn't get the right HAL/ACPI to properly control the soft power settings, although if the OS is upgraded this would be resolved.

Summary

The next steps are to see if some the performance problems on the Skylake are resolved by upgrading to Windows 7 and using proper drivers. Bearing in mind that so much on the motherboard just doesn't work on Skylake running XP, such as USB, upgrading the OS on the platform is pretty much a necessity anyway.

I also have concerns as to whether the option keys will work any more bearing in mind the scope now thinks it's an MSO8104A and not a 54832B.

*Edit to specify I was using builds based on the recovery build plus Infiniium 5.71 scope software.

Hello there
I also tried the 54830D PC upgrade, I currently use the E8500 + GA P43-ES3G configuration, it has a native of five PCI slots. However, I restore the original WindowsXP system backup to the new motherboard, it prompts me to activate, and enter the machine comes with the CD-KEY is invalid, this causes the machine can not run, so I try to completely reinstall a system.
I use the original system backup folder inf and system32 to get the hardware driver, and have all the equipment to get the driver, but I installed the application software, has not been completed correctly, and eventually lead to agscop operation error.
Do you know what exactly need to install the software Caixing it?

[Howardlong]

I had a similar problem after doing an XP repair install to get the OS to work on a new motherboard (resetting AHCI and ACPI settings) using retail edition media from my MSDN subscription. I think may be because the key on the scope is an OEM or volume licence key which won't work with the retail edition, so you need to use the retail edition's key instead.

I also found that sometimes it gets stuck in a rebooting catch 22 loop where I couldn't activate. To correct this, I restarted in safe mode and re-installed IE 8 in safe mode, which then allowed me to activate. Sounds bizarre, but that's how I fixed it, apparently it's a known feature.

[free_electron]

the agilent provided win XP key looks for a marker in the VP22 bios. if that marker is not there it will not take the key.
Older VP22 boards (on machines that did not come with XP) sometimes didn't have the marker. You need to update the bios in that case

[bigeblis]

I had a similar problem after doing an XP repair install to get the OS to work on a new motherboard (resetting AHCI and ACPI settings) using retail edition media from my MSDN subscription. I think may be because the key on the scope is an OEM or volume licence key which won't work with the retail edition, so you need to use the retail edition's key instead.

I also found that sometimes it gets stuck in a rebooting catch 22 loop where I couldn't activate. To correct this, I restarted in safe mode and re-installed IE 8 in safe mode, which then allowed me to activate. Sounds bizarre, but that's how I fixed it, apparently it's a known feature.

Use only the retail version of the system? Let me try.

[bigeblis]

the agilent provided win XP key looks for a marker in the VP22 bios. if that marker is not there it will not take the key.
Older VP22 boards (on machines that did not come with XP) sometimes didn't have the marker. You need to update the bios in that case

But I have already seen the use of other examples of successful operation of the main board, there should be ways to bypass the inspection mechanism that you say?

[plesa]

Maybe try following utility for key transfer ( I did not tested it)
http://www.majorgeeks.com/files/details/rweverything.html

Or use XP for volume licensing, which does not require activation.
Look into setupp.ini
Retail: 76487335
Volume License: 76487270
OEM: 76487OEM

[bigeblis]

Maybe try following utility for key transfer ( I did not tested it)
http://www.majorgeeks.com/files/details/rweverything.html

Or use XP for volume licensing, which does not require activation.
Look into setupp.ini
Retail: 76487335
Volume License: 76487270
OEM: 76487OEM

This software never used, try!

[bigeblis]

the agilent provided win XP key looks for a marker in the VP22 bios. if that marker is not there it will not take the key.
Older VP22 boards (on machines that did not come with XP) sometimes didn't have the marker. You need to update the bios in that case

I have a 54830B and 54830D each one, 54830B is the VP22 motherboard, but the 54830D uses the M-815 motherboard, they have Agilent's boot flag, obviously not mandatory to specify the VP22 motherboard is the premise of XP mode of work. In addition, Agilent's upgrade instructions do not seem to specify a range of XP serial numbers that can be escalated and can only be specified to upgrade to a 128M deep storage serial number range. So I think the replacement of the motherboard should be feasible.

[andy2000]

OEM windows XP looks for a marker in the BIOS.  At least that's how it works with name brand computers that came with XP (eg. Dell).  There are ways to add the code to any BIOS.  If you look around, you'll find utilities that can automatically add it to almost any BIOS image.  Then it's just a matter of flashing it.  You may have to import a new OEM key into XP if you can't find an Agilent BIOS code, or extract it from the old BIOS. 

Assuming the scope has an XP key sticker on it, have you tried doing phone activation?  MS might be willing to transfer it considering how old XP is.

[Howardlong]

I've done at least half a dozen XP phone activations in the past couple of months on this project, I can't remember if any were with just the original installation though. I definitely did several after doing a repair with a retail version to fix AHCI and ACPI.

[bigeblis]

OEM windows XP looks for a marker in the BIOS.  At least that's how it works with name brand computers that came with XP (eg. Dell).  There are ways to add the code to any BIOS.  If you look around, you'll find utilities that can automatically add it to almost any BIOS image.  Then it's just a matter of flashing it.  You may have to import a new OEM key into XP if you can't find an Agilent BIOS code, or extract it from the old BIOS. 

Assuming the scope has an XP key sticker on it, have you tried doing phone activation?  MS might be willing to transfer it considering how old XP is.

So, should extract the original motherboard BIOS, analysis of the internal SN?