Siglent SSA Calibration procedure

[Orange]

Hi Tautech,

I was one of the first users, I think it was around 2017. We had open telnet access on this machine at that time 

ps.

My guess is why Siglent cannot publish their calibration procedure is that this is an automated process, written for a specific set of calibration instruments, in combination with the ability to write the calibation files.
Documenting such process takes time and effort. I don't think that the SSA3000X has build in calib procedures. If it has, it would help to get the thing calibrated with other instruments (read non R&S equipment)

[Orange]

Why do you still promote your NAND back-up method ?
It is not needed. If you make a copy of the calib directory, and it's associated back -up folder you are covered.

Anyhow , if you have an NAND back-up, you need expert knowledge to restore it IF you screw up completely.

be aware that Siglent can easy build in counter measures if invalid ADS files are loaded. They can still call it an 1.2.9.3a version, with modified OS system files.

I don't think it is a bold statement, If I counted correctly, there are now at least 3 brickt instruments, and the users that don't want to report here, are probably more....

where scripts from tv84 can brick your instrument

That's a bold statement. The only script suspicious is the NAND dump .ADS for the SSA, since then removed.

Although not sure if it is from the script...

So, for those that want to play around with calibs in the SSA, I suggest trying a telnet and issuing the NAND backup commands there before doing any changes.

[Sighound36]

Tv84 is one person that has serious integrity and skills and in all my time in dealing with this decent chap I have never found there to be any issue what so ever.

Needing to thoroughly read the various threads and experiences that others have had here on EEV blog will help you understand any possible issues that may arise. However you do this upgrading at your own risk

[tv84]

Why do you still promote your NAND back-up method ?
It is not needed. If you make a copy of the calib directory, and it's associated back -up folder you are covered.

Anyhow , if you have an NAND back-up, you need expert knowledge to restore it IF you screw up completely.

be aware that Siglent can easy build in counter measures if invalid ADS files are loaded. They can still call it an 1.2.9.3a version, with modified OS system files.

I don't think it is a bold statement, If I counted correctly, there are now at least 3 brickt instruments, and the users that don't want to report here, are probably more....

My procedures are not mandatory. People use them if the want. This is not a kid's playground.

I'm not aware that any user had continuing investigating the matter and that it continued to be a problem. Nonetheless, I proposed a diagnostic path wher the problem appeared. Haven't heard from any of the guys. I'm here to help not to create problems!

A full NAND dump allows to restore any "modified OS system files", so there you can see the benefit of having NAND dumps when you start messing with these things.

BTW, why did you share your calib files? So that anyone can go and uncalibrate their equipments? It's much harder to solve that than to restore a NAND.

[Orange]

Why do you still promote your NAND back-up method ?
It is not needed. If you make a copy of the calib directory, and it's associated back -up folder you are covered.

Anyhow , if you have an NAND back-up, you need expert knowledge to restore it IF you screw up completely.

be aware that Siglent can easy build in counter measures if invalid ADS files are loaded. They can still call it an 1.2.9.3a version, with modified OS system files.

I don't think it is a bold statement, If I counted correctly, there are now at least 3 brickt instruments, and the users that don't want to report here, are probably more....

My procedures are not mandatory. People use them if the want. This is not a kid's playground.

I'm not aware that any user had continuing investigating the matter and that it continued to be a problem. Nonetheless, I proposed a diagnostic path wher the problem appeared. Haven't heard from any of the guys. I'm here to help not to create problems!

A full NAND dump allows to restore any "modified OS system files", so there you can see the benefit of having NAND dumps when you start messing with these things.

BTW, why did you share your calib files? So that anyone can go and uncalibrate their equipments? It's much harder to solve that than to restore a NAND.

Sharing my files, was just to give you an idea how this thing is calibrated, and what is involved to adjust it. By no means if is a reference set. But I assume that readers how are doing this are not stupid.
Like you say this is not a playground. I'm not questioning your integrity or your work, by no means. I just wanted to warn potential hackers of the substantial risks involved.

[tv84]

My guess is why Siglent cannot publish their calibration procedure is that this is an automated process, written for a specific set of calibration instruments, in combination with the ability to write the calibation files.

Getting back OT:

(IMO) I think an automated iterative process is used, controlled by a PC, where SCPI commands are sent to SSA and other calib auxiliary equipments.

I would say the SSA has the ability to (re-)insert in its calib files the values commanded via SCPI.  I've seen this in other equipments and estimate that the same approach is used here.

[avlijas.sladjan]

My guess is why Siglent cannot publish their calibration procedure is that this is an automated process, written for a specific set of calibration instruments, in combination with the ability to write the calibation files.

Getting back OT:

(IMO) I think an automated iterative process is used, controlled by a PC, where SCPI commands are sent to SSA and other calib auxiliary equipments.

I would say the SSA has the ability to (re-)insert in its calib files the values commanded via SCPI.  I've seen this in other equipments and estimate that the same approach is used here.

Maybe it is true. That is reason why is good to have old HP instrument with full service manual, schematic and ebay parts

[rf-loop]

Hi,
Attached is a calibration set for an SSA3032X. I made it as a back-up, and to adjust the TG levels. Mine was calibrated only to 2.1 GHz and the output levels from 2.1...3.2 GHz had severe dips...
 (so called calibrated from the factory).

I only made changes to the TG part, where you can figure out which values to change, make it relatively flat again. Lucky these the values are decimals. (freq, and relative level)
I managed to get them better. File in question was cali_tg

For all the other values in the many files that comprise the total calibration, it is a real challenge, since most of them are hex numbers.

And yes I have other test equipment to verify my gear. The SSA3021X I got was changed to 3.2GHZ, and it was OK for the entire range, except the TG.

Anyway, with the current state of affairs, where scripts from tv84 can brick your instrument, I would not take the risk again to start messing with it if it was a new instrument.

As I told previously I have tweaked it. Done some experimental things. As can see data there is quite simple. But warning. Do not break its integrity. Every position need be exactly in same positions and every byte just exactly in its place and not any extra bytes. It do ask dummies like are you sure...  It must not edit with any kind pof editor what can produce hidden bytes like lf or cr and so on. I dit this my experimental using  hex editor.  But all data etc is now far from me and better I do not say anything about how these values inside cal data are.  TG is simple. There is frequency bands and corection to this band. If I remember right correction resolution is 1dB.  This is bit rough but when SA is used Normalization is what do fine adjust.

Also as can see freq reference DAC can also tweak there. (but if I remember right it can also adjust in UI in some hiddeen menu -- if I remember right with my after "best before" date memory).

You factory cal have looks like done 2017-05-15 (inside one file is this date)  or 2017-08-16 so it is not very early version.

[tv84]

Attached is a calibration set for an SSA3032X. I made it as a back-up, and to adjust the TG levels.

Here is another set taken from another SSA.

[rf-loop]

Attached is a calibration set for an SSA3032X. I made it as a back-up, and to adjust the TG levels.

Here is another set taken from another SSA.

From bit older SSA.

[luudee]

Hi Guys,

so I want to perform "Normilizing" on my SSA.

Typically, I would do this by attaching two cables, which I would later attach to
my device under test, and instead of the DUT I would place a small path-throug
bridge between the cables. I guess this is the "normal way" to do it, and it works
great.

Now I would like to normalize at the N-Connectors on the SSA itself, without any
cables.

Is there a way to do that ?

Thanks,
rudi