HP/Agilent 1675x logic analyzer card memory up-hack

[MarkL]

...
Edit: I just looked at the cards, and the one difference that jumps out at me is that the 'A' uses XCV300 FPGAs, and the 'B' uses XCV300E FPGAs.  The Virtex-E is a 1.8V FPGA instead of 2.5V of the original Virtex.  Additionally, while the part number sounds like they'd be basically the same FPGA, the 300E has more logic resources than the 300, so they almost certainly require different bit files, which the mainframe would need to be aware of.  So, I'd guess the 'A' to 'B' was simply a change to the Virtex-E due to the original Virtex going obsolete.

I think you're exactly right on Virtex vs. Virtex-E.

In /usr/sprockets/tools/instrument/16700/16718/, there are two different versions of bit files, yari.rbt and yari_rev2.rbt.  The header of yari.rbt:

  Xilinx ASCII Bitstream
  Created by Bitstream D.27
  Design name:    yari_g.ncd
  Architecture:   virtex
  Part:           v300bg352
  Date:           Wed Oct 03 08:50:00 2001
  Bits:           1751808

The header of yari_rev2.rbt:

  Xilinx ASCII Bitstream
  Created by Bitstream D.27
  Design name:    yari_g.ncd
  Architecture:   virtexe
  Part:           v300ebg352
  Date:           Fri Oct 05 09:19:46 2001
  Bits:           1875648

The driver is /usr/sprockets/tools/instrument/16700/16718/lib16718.sl and covers the 16718A, 16719A, 16740A, 16741A, 16742A, 16750A, 16750B, 16751A, 16751B, 16752A, and 16752B.  This is where the above two bit files are referenced.  That family of cards is called "yari" and they all have the same basic architecture.  The A and B versions probably operate identically after being loaded with their respective bit files.

Thanks for solving the mystery of A vs. B!  That's been bothering me for a long time.

[DogP]

Very cool... nice analysis.  I've got a lot to learn about this system... I've had a 16500A for ~20 years, and figured it was time to upgrade.

DogP

[DogP]

Sorta "up-hack" related... has anyone made an adapter to use older logic analyzer cables with the newer units?

I've got a 16550A card in my 16500A, with a bunch of cables, but no cables for my 16750A/Bs.  They both go to the standard 40-pin connector, and the 16550A supports up to 500 MHz timing, so it seems a simple adapter to go from the 60-pin to the 50-pin connector would at least be somewhat usable.  Is the connector type and/or pinout documented anywhere?

Though maybe with the cost of connectors, it just make sense to find the right cables.  If that's the case, any recommendations for sourcing these cables?  Is buying cards with cables and scrapping the cards the best way?

Thanks,
DogP

[gslick]

Though maybe with the cost of connectors, it just make sense to find the right cables.  If that's the case, any recommendations for sourcing these cables?  Is buying cards with cables and scrapping the cards the best way?

Might be easiest to just buy "for parts" modules that include the cables.

For example, maybe try making offers on something like these. Maybe they would consider lower offers for just the pod cables, and make shipping cheaper too.

16716A    https://www.ebay.com/itm/223464547231
16715A    https://www.ebay.com/itm/202701121599
16715A    https://www.ebay.com/itm/173876217896

Could try an eBay search like this for modules with compatible pod cables:
(hp, hewlett, agilent, keysight) (16715a, 16716a, 16717a, 16718a, 16719a, 16740a, 16741a, 16742a, 16750a, 16751a, 16752a, 16750b, 16751b, 16752b, 16910a, 16911a)

Or maybe other members here have modules which have failed and they might consider selling the pod cables.

[DogP]

Thanks... yeah, I'll probably just end up making an offer on a set.

DogP

[DogP]

I ended up grabbing that 16716A for the cables... though of course "upgraded" the card as well.  So thanks for that info in this thread!

Also, while sifting through this thread, I noticed the discussion about the 1674x -> 1675x, and noticed the subtle difference in what DocBen wrote.  He says he converted the 16740A to 16752B.

So I have finally gotten around to try to get a 16740a to be a 16752b.

I decided to move the resistors on my 16742A to that of a 16752A. The main frame reported an error trying to write to the FPGA.

I saw absolutely, as I can be, no physical differences between the one 142A and three 152A PCBs. The only difference are subtle marking differences (suffixes) on the larg Xlinx chips.

So... given that it seems the difference between the 1675x A vs. B seems to be the XCV300 vs. XCV300E, and the above posts discuss unsuccessful upgrades from 1674xA to 1675xA, and note the FPGA differences, maybe the 1674xA is actually the same hardware as the 1675xB, not the 1675xA, and therefore needs to be jumpered for B to work correctly.  I didn't find any high quality pics of a 1674xA board to check the markings on the Xilinx part, but probably worth checking if you're attempting this upgrade.


Edit: Might also apply to the 16718A and 16719A (not sure which FPGA is used on those boards).

DogP

[gslick]

I ended up grabbing that 16716A for the cables... though of course "upgraded" the card as well.  So thanks for that info in this thread!

Also, while sifting through this thread, I noticed the discussion about the 1674x -> 1675x, and noticed the subtle difference in what DocBen wrote.  He says he converted the 16740A to 16752B.

So I have finally gotten around to try to get a 16740a to be a 16752b.

I decided to move the resistors on my 16742A to that of a 16752A. The main frame reported an error trying to write to the FPGA.

I saw absolutely, as I can be, no physical differences between the one 142A and three 152A PCBs. The only difference are subtle marking differences (suffixes) on the larg Xlinx chips.

So... given that it seems the difference between the 1675x A vs. B seems to be the XCV300 vs. XCV300E, and the above posts discuss unsuccessful upgrades from 1674xA to 1675xA, and note the FPGA differences, maybe the 1674xA is actually the same hardware as the 1675xB, not the 1675xA, and therefore needs to be jumpered for B to work correctly.  I didn't find any high quality pics of a 1674xA board to check the markings on the Xilinx part, but probably worth checking if you're attempting this upgrade.

DogP

Did the 16716A you picked up for parts for the cables turn out to be fully functional?

Regarding the 1674xA FPGA, my 16741A has XCV300E parts.



The ID resistors for that 16741A are as shown here (Reply #17 on: November 29, 2017, 12:53:27 am):
https://www.eevblog.com/forum/testgear/hpagilent-1675x-logic-analyzer-card-memory-up-hack/msg1361817/#msg1361817

[DogP]

Did the 16716A you picked up for parts for the cables turn out to be fully functional?

Yes, with the exception of the missing interconnect cable.

Regarding the 1674xA FPGA, my 16741A has XCV300E parts.

Interesting... so yes, with the XCV300E, the 16741A would likely need to identify as a 16752B, not A.

DogP

[MarkL]

There was a thread that went into troubleshooting DogP's 16533A and 16750B cards.  Just dropping the link here in case anyone is interested in the details (Glen...):

  https://www.eevblog.com/forum/repair/series-defect-on-agilent-167xx-boards/msg3989777/#msg3989777

[DogP]

Kinda "memory up-hack" related... I swapped out the 16702B's OEM SCSI hard drive with an ACARD AEC-7722 (firmware reflashed to AEC-7726, plus a 50-pin SCSI adapter - if not familiar, there's some discussion in this thread: https://www.eevblog.com/forum/microcontrollers/add-a-single-byte-in-a-bin-file-every-one-byte-(all-file)/ )... plus a 32GB CF card in a CF to IDE adapter.

Overall, it dropped almost 30 seconds off the boot time (down to 2:46 from 3:13 with the OEM drive, with a pair of 16754A/Bs, a 16534A, and a 16720A), so quite a bit faster.  An IDE SSD I tested was 8 seconds faster than that, but the CF card was more convenient to use.  I know that there's also the SCSI2SD adapter, though from the benchmarks I've read, they're actually slower than an OEM drive.

Anyway, in case anyone's interested, here's the bracket I used for mounting it in the 16702B: https://www.thingiverse.com/thing:5373974 .  And note that not all 68-pin to 50-pin SCSI adapters work with the ACARD adapter... here's a design I made based on the HP 5182-4551 adapter which did work: https://github.com/pdaderko/16702B/tree/main/SCSI_50_to_68_female (soldered directly to the ACARD adapter).

And I'm happy to say that I FINALLY have my 16702B back together and ready to use, for the first time since I bought it.  I guess now I can retire my 16500A(s) and get back to work on the couple of RE projects that I wanted these faster/deeper memory cards for.

DogP

[Hamster]

I used a SCA80 18gb SCSI drive when i had the 167xx frames and it greatly sped up the

[artag]

The older 16555A and 16555D cards may also have memory size as a configuration option. Both appear to have the same size memory array (34 x M5416283-60).

However, there are no individual resistors around the Actel gate array on the top side of the board and few on the input side of the board. Comparing my 16555A with the pictures at  https://www.dropbox.com/sh/2rw2yf4dsdkx0zf/AABbXunP1vkF7xeaC5_B7lmua?dl=0&preview=16555D-A1.jpg I can see no differences.

The reverse of the  board does, however, have more including some in the most likely location around the Actel gate array. Unfortunately I don't have a 16555D and haven't found a picture of the back.

Could anyone with a 16555D post a picture of the back please - ideally a high resolution photo of the whole thing but at minium a shot of the area at the edge connector end, under the Actel device.

[artag]

I wrote the above before I got my newly-acquired 16500C/16555A/16522A setup working. Now I have, I'm somewhat puzzled.

The LA board is clearly labelled 16555A 1MSa. But the mainframe identifies it as "2.0M Sample 110/500MHz LA" and shows 2MSa (/4MSa in half-channel mode). I guess it could have an ID fault and will turn out to have less memory thatn it appears but that seems unlikely. Or maybe it's a frankenstein card with the wrong rear panel. Or maybe it can be secretly enabled like those other cards and it's already been done. Or the start of adhesive corrosion I can see has unexpectedly done the same thing.

Any ideas ?


Further update :

I had some problems with that card that turned out to be a connector, so I was very unsure of whether it was working. However, it's now apparently working fine as a 16555D despite being labelled 16555A. So I assume it was modied either as a field upgrade or a hack. Howerver, I can't tell how. There is no evidence of different solder joints etc.

If anyone has a 16555A that reports as such, I'd be interested to compare (probably some details of resisters underneath) and you might be able to upgrade yours to a D by copying it.

[RAQU_PL]

With help from another forum member, here is how to fully enable a 16911A card.

To Enable 500Mhz Turbo: Change 0x000A to 0x01 , increment value @ 0x000C by the value added to 0x000A
To enable Max Sampling, Change 0x000B to 0x04 , increment value @ 0x000D by the value you incremented at 0x000B

For example on my two cards i did tests on: ( that i read the eeproms on )

0x00 0x02 0x16 0x20  [ stock 250mhz/16ms ]
0x01 0x04 0x17 0x24  [ fully enabled ]

Have anyone tried unlocking 16911a in 168xx frame?  Does anyone have 168xx frame with options other than 250 state 1Meg memory and is willing to read u90 eeprom?

Unfortunately Hamster's codes doesn't work. Behavior is that nothing changes but when trying to sample, the memory filling procedure hangs indefinitely.

My eeprom content 0x0A to 0x0D : 0x00 0x31 0x18 0x26  for 1Meg 250 MHz state.

[danielbriggs]

I’ve liberated a 16910A in a 16902A frame.
Why software do the 168xx frames use? Do they use Agilent LPA?

There’s info on this forum to generate your own licenses to enable whichever options you require.
This is easier than dumping the EEPROM’s, but that’s also a valid technique.

I think on the 16910A’s it was U102.
I’ll have a root around for the EEPROM dumps if they’re of any use. I had two 16910A’s with a variety of configs. But I definitely upgraded one to the max mem and speed with enough perseverance.

[RAQU_PL]

I’ve liberated a 16910A in a 16902A frame.
Why software do the 168xx frames use? Do they use Agilent LPA?

There’s info on this forum to generate your own licenses to enable whichever options you require.
This is easier than dumping the EEPROM’s, but that’s also a valid technique.

I think on the 16910A’s it was U102.
I’ll have a root around for the EEPROM dumps if they’re of any use. I had two 16910A’s with a variety of configs. But I definitely upgraded one to the max mem and speed with enough perseverance.

Yes, 168xx do use Ag LPA.

There are some differences between those families. In example 168xx MB bios is locked, so you can not replace motherboard, different eeprom content? for the same HW option and propably many others.

Did You mean "generic flexlm crypt generator v0.5 by haldir"?
Do You have one? I can not find it. All links that i could find are dead.

[Gerald]

same question here!

[gslick]

I’ve liberated a 16910A in a 16902A frame.
Why software do the 168xx frames use? Do they use Agilent LPA?

The Agilent {16804A, 16806A, 16821A, 16822A, 16823A} analyzers are essentially exactly same hardware as an Agilent 16901A mainframe which is populated with {(2x) 16911A, (2x) 16910A, 16912A + 16720A, 16911A + 16720A, 16910A + 16720A} modules. (A 34-channel 16912A module does not exist as a standalone 16900-series product). They both use the same Intel D915GUX, ADLINK M-880, or ADLINK M-890 motherboards. The only real mechanical difference is that the 16901A has a module shroud for removable modules with ejectors, while the {16804A, 16806A, 16821A, 16822A, 16823A} analyzers have a module shroud for two modules screwed in without thumbscrews or ejectors.

The Agilent {16801A, 16802A, 16803A} analyzers are essentially exactly same hardware as an Agilent 16901A mainframe which is populated with {16912A, 16911A, 16910A} modules, except that in addition to the main 600W power supply, the 16901A has a second 175W power supply while the {16801A, 16802A, 16803A} have a second 15W power supply, and the 2-slot module interface board (MIB) of the 16901A is replaced by a 1-slot module interface board in the {16801A, 16802A, 16803A}. The module shroud also has a single slot opening.

As the hardware is essentially the same, they use the same logic analyzer application.

In theory, a {16804A, 16806A, 16821A, 16822A, 16823A} analyzer could be turned into a more flexible 16901A analyzer. More flexible in that a 16900-series mainframe supports older {16740/41/42A, 16750/51/52A, 16750/51/52B, 16753/54/55/56A} modules, which may be easier and/or less expensive to acquire.

In practice, I don't know if anyone has figured out how to do that yet, or at least I haven't seen anyone post here that they have. As far as I know, swapping a standard 16910A module into a 168xxA analyzer results in an unrecognized module, and swapping a 16803A/16823A 16910A module into a 16901A also results in an unrecognized module.

My guess is that there are some bits in the {16912A, 16911A, 16910A} module EEPROMs that indicate whether they are standard 16900-series modules, or 168xxA-series specific modules, and the logic analyzer application will reject modules that are swapped between the two series.

And probably some bits somewhere in the hardware that distinguish between a 16901A mainframe and a 168xxA analyzer. Maybe a tag somewhere in the motherboard BIOS? Or maybe somewhere in the MIB? A 168xxA analyzer needs to know the difference between a 1-slot MIB and a 2-slot MIB, maybe that would be a good place to store model specific information.

[tv84]

Did You mean "generic flexlm crypt generator v0.5 by haldir"?
Do You have one? I can not find it. All links that i could find are dead.

There is a "xmas gift" in this forum.

[Gerald]

Successfully upgraded a 16804A from 250MHz sampling rate to 500MHz, but no way to upgrade the sample memory. Whatever option used (4Meg / 16Meg / 32Meg) it stays at the original 1Meg.

Any clues ?

[gslick]

Successfully upgraded a 16804A from 250MHz sampling rate to 500MHz, but no way to upgrade the sample memory. Whatever option used (4Meg / 16Meg / 32Meg) it stays at the original 1Meg.

Any clues ?

Did you use "FEATURE 16911_32Meg_Samples" for the 16804A? That should work.

[Gerald]

Did you use "FEATURE 16911_32Meg_Samples" for the 16804A? That should work.

Yes. Copy paste method and modifying the original 1Meg license file, the re-signing.

[Uup]

For all those that need to recover licenses for the Agilent 168x(x)/169x(x), attached is the "secret sauce".

It has been tested with the above licenses and should work with all others.

Thanks, but I have a really stupid question. What is to be done with the generated .exe file? The file just crashes, no matter what OS I run it on, including Win7 on my 16822A.

Thought I would do my homework, but after >3 hours of reading pages from different threads it isn't any clearer to me... Other than realising the amount of effort that went into figuring it all out and making the tool, that is.

[RAQU_PL]

For all those that need to recover licenses for the Agilent 168x(x)/169x(x), attached is the "secret sauce".

It has been tested with the above licenses and should work with all others.

Thanks, but I have a really stupid question. What is to be done with the generated .exe file? The file just crashes, no matter what OS I run it on, including Win7 on my 16822A.

Thought I would do my homework, but after >3 hours of reading pages from different threads it isn't any clearer to me... Other than realising the amount of effort that went into figuring it all out and making the tool, that is.

You need to run program using windows console. After the tool path, You need to enter the path of the license to sign. Example licenses You have in file with the tool or in thread somewhere.



Edit:

#START LICENSE INFO
FEATURE 16911_500MHz_Turbo_State agilent 1.0 permanent uncounted HOSTID=ANY SN=MYXXXXXXXX SIGN=
 
#END LICENSE INFO
#START LICENSE INFO
FEATURE 16911_32Meg_Samples agilent 1.0 permanent uncounted HOSTID=ANY SN=MYXXXXXXXX SIGN=
 
#END LICENSE INFO

[Uup]

For all those that need to recover licenses for the Agilent 168x(x)/169x(x), attached is the "secret sauce".

It has been tested with the above licenses and should work with all others.

Thanks, but I have a really stupid question. What is to be done with the generated .exe file? The file just crashes, no matter what OS I run it on, including Win7 on my 16822A.

Thought I would do my homework, but after >3 hours of reading pages from different threads it isn't any clearer to me... Other than realising the amount of effort that went into figuring it all out and making the tool, that is.

You need to run program using windows console. After the tool path, You need to enter the path of the license to sign. Example licenses You have in file with the tool or in thread somewhere.



Edit:

#START LICENSE INFO
FEATURE 16911_500MHz_Turbo_State agilent 1.0 permanent uncounted HOSTID=ANY SN=MYXXXXXXXX SIGN=
 
#END LICENSE INFO
#START LICENSE INFO
FEATURE 16911_32Meg_Samples agilent 1.0 permanent uncounted HOSTID=ANY SN=MYXXXXXXXX SIGN=
 
#END LICENSE INFO

Thank you, that makes sense. I hadn't looked at a licence file because I couldn't find any. The licence directories were empty.

I checked the current licences from the LA application and it has '! Running Factory Licence' but from the licence manager it shows no licences. However, at the bottom of the Licence Manager window it has 'Installing...'

Interestingly, it is fully enabled anyhow. I can sample to 500MHz (full channel) and can have a memory depth of up to 32MB, as well as having a lot of protocols and inverse assemblers.

I'm currently on V05.70.0002. I did a clean Win 7 install with V05.90 and had it go to 250MHz and 1MB, same as the stickers on the back of the unit.

I restored the backup that I made and am back to 500MHz and 32MB, so will leave it at that for the moment. It has a bunch of other software installed on this unit, which I don't need but can clean up.

Thanks again