Agilent E7495 linux root account

[technogeeky]

I wanted to bring to attention what I believe is an example of drifting or failing calibration. In a specific sense, none of this matters (because you are probably going to normalize, which makes the measurement you make relative to the problematic measurement below). It's still interesting, and I'd love to hear speculation on why this is happening. I can make further measurements (such as using a different spectrum analyzer to view the output, or a SDR).

The following will tell you the default settings after preset, then show which one or two things to change from the previous. For all tests, the only connection is a 0.5 meter Agilent N-to-N cable.


...

Any thoughts, guesses, or further tests would be appreciated. I have SDRs which I can use to monitor the output from the CW/generator. I can use another spectrum analyzer to view the output. I think I've done this, and I didn't see any sudden jumps at 1GHz exactly. Likewise, when I use another signal generator into this spectrum analyzer (E7495a), I similarly do not see any sudden jump. So this leads me to believe that this effect is from some sort of calibration table that the unit stores for the Two Port Insertion measurement mode specifically.

It's worth noting, too, that the procedure for this originally calls for using a two 10dB attenuators (back to back) on the RF out port. Using an attenuator set  does, of course, reduce this effect but it's still noticeable.

I can verify this effect with my E7495B. But this is not erroneous behavior: it is caused by the preamp not spanning the complete frequency range (see https://www.eevblog.com/forum/testgear/agilent-e7495-linux-root-account/msg621126/#msg621126).

When do the test with a short cable (RG316) and a 20dB attenuator, the insertion loss measurement is flat to within 2dB. After normalization I get a nice straight line across the whole frequency range. The insertion-loss measurement is probably designed for longer cable runs with significant loss. This would explain the use of the 20dB attenuation when normalizing. So if your cable is quite short and has negligible attenuation the preamp is saturating, thus giving you wrong results.

Thanks!

I had another problem yesterday which I'm sure is new. The display on the unit keeps turning off the backlight, crazy fast. I can actually sometimes catch it saying in the status bar something like "Turning off backlight in 3 seconds" with a countdown.

Any ideas where this could be stopped or configured?

[technogeeky]

Thanks!

I had another problem yesterday which I'm sure is new. The display on the unit keeps turning off the backlight, crazy fast. I can actually sometimes catch it saying in the status bar something like "Turning off backlight in 3 seconds" with a countdown.

Any ideas where this could be stopped or configured?

This, along with another problem (the network interface not working), just disappeared after I left the unit unplugged for a while (a few hours). So that's good and bad, but more good than bad.

I still had practical problems making two port insertion loss measurements (even with 20dB worth of pads) and Auto (+10dBm) output level. I would get sudden burst of uncertainty in the response right after 1GHz, for almost 500MHz. In any case, I managed to make the measurements by going to 0dB.

[ferdinandkeil]

Thanks!

I had another problem yesterday which I'm sure is new. The display on the unit keeps turning off the backlight, crazy fast. I can actually sometimes catch it saying in the status bar something like "Turning off backlight in 3 seconds" with a countdown.

Any ideas where this could be stopped or configured?

This, along with another problem (the network interface not working), just disappeared after I left the unit unplugged for a while (a few hours). So that's good and bad, but more good than bad.

I still had practical problems making two port insertion loss measurements (even with 20dB worth of pads) and Auto (+10dBm) output level. I would get sudden burst of uncertainty in the response right after 1GHz, for almost 500MHz. In any case, I managed to make the measurements by going to 0dB.

Glad to hear that the problem went away on its own.

Regarding the insertion loss measurement, 10dBm just might be a little much for the preamp. Did you normalize (aka calibrate) the measurement?

[technogeeky]

Thanks!

I had another problem yesterday which I'm sure is new. The display on the unit keeps turning off the backlight, crazy fast. I can actually sometimes catch it saying in the status bar something like "Turning off backlight in 3 seconds" with a countdown.

Any ideas where this could be stopped or configured?

This, along with another problem (the network interface not working), just disappeared after I left the unit unplugged for a while (a few hours). So that's good and bad, but more good than bad.

I still had practical problems making two port insertion loss measurements (even with 20dB worth of pads) and Auto (+10dBm) output level. I would get sudden burst of uncertainty in the response right after 1GHz, for almost 500MHz. In any case, I managed to make the measurements by going to 0dB.

Glad to hear that the problem went away on its own.

Regarding the insertion loss measurement, 10dBm just might be a little much for the preamp. Did you normalize (aka calibrate) the measurement?

Yep. I don't believe the 10dBm claim, but that's what it says on the source level of my instrument. I'll take some SDR screenshots (while restraining the tracking generator to a small frequency range). It normalizes fine. But once I make a measurement, and it's still in +10dBm mode, there is a bunch of noise right after 1000 MHz.

[JessieAMorris]

Just picked up a working unit off the 'Bay for $300 shipped. Excited to start playing with some RF goodness.

[walter2]

I recently got 2 E7495A, both booted and worked. Working with a fellow conspirator here in kelowna, we took his known good PCMCIA/Flash update card with 6.25, and did the flash upgrade to the best unit with 6.24 working on it.

the update never completes, it runs for a very long time, gets to anywhere from 70-90% of the flash update, and never goes further. when we transferred the flash card right to the machine (rather than via the pcmcia adaptor), it runs faster, and doesn't complete, but also throws up a strange message "inflate returns -3" repeatedly, then a blosk fo data too fast to read, then stalls again. we tried the normal on-screen update, and the hidden lower right button full refresh, but fail at the end.

we have tried all variants of the process, but cannot get to an updated machine. this appears to indicate that the internal flash on the motherboard is bad at some location, and doesn't permit a valid update. is this right?  if so, how can I fix the flash on the motherboard.?? does anybody have ANY internal documentation for this unit?  I see nothing at hp/agilent/keysite.

you can email me directly at walter2 -at- sphere.bc.ca

I would really like to get this unit working, and am incredibly sorry I tried this update.  any help appreciated.

regards,
walter

[technogeeky]

I recently got 2 E7495A, both booted and worked. Working with a fellow conspirator here in kelowna, we took his known good PCMCIA/Flash update card with 6.25, and did the flash upgrade to the best unit with 6.24 working on it.

the update never completes, it runs for a very long time, gets to anywhere from 70-90% of the flash update, and never goes further. when we transferred the flash card right to the machine (rather than via the pcmcia adaptor), it runs faster, and doesn't complete, but also throws up a strange message "inflate returns -3" repeatedly, then a blosk fo data too fast to read, then stalls again. we tried the normal on-screen update, and the hidden lower right button full refresh, but fail at the end.

we have tried all variants of the process, but cannot get to an updated machine. this appears to indicate that the internal flash on the motherboard is bad at some location, and doesn't permit a valid update. is this right?  if so, how can I fix the flash on the motherboard.?? does anybody have ANY internal documentation for this unit?  I see nothing at hp/agilent/keysite.

you can email me directly at walter2 -at- sphere.bc.ca

I would really like to get this unit working, and am incredibly sorry I tried this update.  any help appreciated.

regards,
walter

walter:

From here, inflate() returning -3 is evidently the error Z_DATA_ERROR. This error arises during extraction from a compressed file (hence inflate). Z_DATA_ERROR implies that the source data is missing, invalid, or corrupted. I would get a new or different CF card, or a different CF card to flash adapter.

I know the update process is run during boot (it checks to see if a card with an update is present, and asks you if you want to try and install it). Since this is just some command, it should be possible to run this command from the local memory inside the machine. You can telnet in, use FTP to upload the update file/images, and then run the update manually.

[ogden]

From here, inflate() returning -3 is evidently the error Z_DATA_ERROR. This error arises during extraction from a compressed file (hence inflate). Z_DATA_ERROR implies that the source data is missing, invalid, or corrupted. I would get a new or different CF card, or a different CF card to flash adapter.

Also it is worth checking MD5 checksum of the file(s). It could be so that during download file somehow became corrupted.

[walter2]

For "greater clarity" here, I was able to use that SAME update flash card to successfully update my second unit, so there is no question that it works. we also confirmed it was good on three other machines.

I would have thought the messages about inflating were just as others suggested, that it is an update source problem after decompression, but it is certain that is not the issue, the problem is at the destination, the internal flash on the motherboard. I believe this message indicates that a checksum failed after attempted writing, but am not positive.  more disturbing is that I have zero info on the mother board.  the issue always arises towards the end of the process, which points to a bad or marginal area at the end of the local flash storage.

anyone with motherboard info, and how to fix the on-board flash would be greatly appreciated. there is no local flash card or other easily replaceable memory on the card, other than what appears to be a boot ROM.

regards,
walter

[TheSteve]

Is it possible the problem unit had a bunch of user data stored on the internal memory before the upgrade was started so it is running out of space?

[walter2]

I don't think that is the case, since the "re-build" option re-does EVERYTHING, including errasing and re-formatting the flash...  I suspect the problem is a poor sector of flash on the motherboard, works a bit some times, fails others.

since you are in BC, come see us for stuff day next saturday, the 21st! we are in west kelowna.
all the details are here, assuming the posting software allows it:
http://www.sphere.bc.ca/test/stuffday.html

if it gets vaporized, google sphere research corp., and look for the stuff day link at the top of /index.html.

all the best,
walter

[RCHRDM]

Hello

I will be picking up an Agilent E7495 next week.  I am not sure of which options are included.

Does anyone have unlock codes for the options for this instrument?

Richard

[technogeeky]

Hello

I will be picking up an Agilent E7495 next week.  I am not sure of which options are included.

Does anyone have unlock codes for the options for this instrument?

Richard

There are no codes. The nature of the unlock process is:

* patch the binary so that it doesn't check if the unlock code is valid or not
* insert any dummy text instead of the code in the license file, for each option you want to enable

This will be easier to understand once you have a device in front of you. The license file will have examples of valid codes that ship with the device, so you will have an example of the structure of each option.

[RCHRDM]

I got my E7345 today.  It has a few cosmetic issues, but it boots up!  There is no power meter option.

There appears to be a gasket around the LCD that is loose on the left and bottom of the display, so I guess I will have to take the thing apart and replace the plastic screen overlay or cover.  I attached a picture.  The spectrum analyzer works OK.

There is also a missing threaded insert that takes the battery door retaining screw.  Would someone mind taking a picture of where the battery door screw threads into the case to hold the door closed?

Richard

[RCHRDM]

Does anyone have a service manual or parts list?  I tried Agilent and they said they did not have one...

Richard

[RCHRDM]

I am in the process of hopefully getting some additional parts to replace those missing from y E7495.

Could you please step me through the "hacking" process to enable the power meter?  I can follow along with the my instrument now.

BTW, I have a service manual now if anyone needs one.

Richard

[technogeeky]

I am in the process of hopefully getting some additional parts to replace those missing from y E7495.

Could you please step me through the "hacking" process to enable the power meter?  I can follow along with the my instrument now.

BTW, I have a service manual now if anyone needs one.

Richard

What do you mean about missing parts?

The power meter is not a software option. If it's physically installed, it will be working (I think).

You have the real service manual? I'd love to see it. PM me.

[RCHRDM]

After I got the E7495 home, I noticed that it was missing the connector for the power meter.  (Someone had put tape over the hole, so I could not tell from the auction photos.) 

I took it apart to repair a problem with the LCD cover and found that the power meter board was installed, but two ribbon cable connections were missing.  I assume one of them is for the power meter connector on the case.

I also found that there were a few other missing things including the main case gasket, a couple of screws, and the cover for the PCMIA area.

There is a guy who has a parts machine, so we are talking about pulling some parts from it.

When the instrument is powered up, the power meter button is grayed out, so I assume the option is not activated.

Richard

[kk6ioz]

To extend the lower range, can we build a transverter with a 400MHz LO to cover 0Hz to 400MHz for the VNA measurement? and would a transverter work for SWR measurement.

It would be great if a SW hack could be found.

[jokkydee]

The company I work for have a N1996A and I wanted to enable the stimulus/response N8995A-SR3 option so I contacted Keysight and they said the unit is no longer supported and that I can't even pay to get the option enabled. I mentioned that people have hacked the software to enable options, and they just referred me to their forums and said hopefully I can find someone who can help. So now I have a clear conscience with hacking the firmware to enable options since I tried the formal path. Can anyone explain how I can do this, considering I don't have much experience with Linux? I read over this forum but couldn't get a clear picture of what exactly I need to do. I understand I need to patch the firmware, PA0PBZ mentioned a patch for the E7495 but that uses the elgato process, I don't know what I need to do for the N1996A. Can anyone please help?

[PA0PBZ]

You've got mail.

[jokkydee]

Thanks PA0PBZ, I'll let you know how it goes. Also, reading the installation note for the TG option, https://literature.cdn.keysight.com/litweb/pdf/N1996-90009.pdf?id=1132460 it notes that several adjustments are required after this option is enabled, including TG flatness and return loss calibration. Also they mention some minimum performance checks like freq response and DANL. There's a calibration software that requires a license to use it in order to run these adjustments. I wonder how critical these are. I've contacted Keysight regarding this software to see what they say.

[PA0PBZ]

It's not only the software you need but also the instruments, which is an even larger problem. If you just normalize before your measurement flatness should not be a problem.

[jokkydee]

I've got the options enabled after the patch, thanks for the help. I take it that normalising will mean for the purposes of doing measurements with the unit itself it should be ok, but if for example, characterising a cable for use in an external system by exporting the insertion loss data, it may not be reliable without those adjustments and checks?

[PA0PBZ]

When normalizing you connect the output of the instrument to the input and the instrument builds a correction table that takes the wrinkles out of the signal generator and the receiver. That way you can do your measurement and it will not be influenced by any anomalies of the instrument.