Agilent 8753ES and other VNAs alike

[Gandalf_Sr]

OK, I've ordered one of these floppy disk drives https://www.amazon.com/gp/product/B01GHF5VS2/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1 , mainly because it works with Windows 10 which is standard across my PCs now.

Been messing with the 8753E and it all seems to work but I have clearly forgotten a lot about this subject, any suggestions for a simple test that I can walk through that will allow me to understand what I'm doing?

[trukresom]

Hi Gandalf_Sr,

One important thing to be procured when beginning with network analyzers is a calibration kit compatible
with the connectors of the components you want to measure. The modules which I am measuring have all
SMA (fem.) connectors. Thus I needed two APC7-SMA adapters, 2 (short) SMA cables and a matching
SMA calibration kit. Normally the frequency range which interests me is below 1 GHz. So I made myself
a calibration kit based on cheap SMA connectors. Further to verify my calibration kit I purchased one of these:
http://www.hhft.de/files/products/network_analysis/SMA/datasheet_eco1.pdf

But before entering the calibration menu of your 8753 don't forget to set the Cal Kit to one which is similar to
SMA by pressing:
[Cal  ] --> <Cal Kit> --> <Select Cal Kit> --> <3.5mm D>
This works surprisingly good, at least for the frequency range which I am using.

To verify my calibration I often re-measure the cal kit itself. This should resemble at what you can see on
page 6-8 in this document:
http://exodus.poly.edu/~kurt/manuals/manuals/HP%20Agilent/HP%208753ES%20Documentation/8753ES%20Option%20011%20User's%20Guide%20(Jun02).pdf

[Gandalf_Sr]

trukresom, Thanks for the extra information. 

I already own this calibration kit from eBay http://www.ebay.com/itm/RF-Network-Analyzer-Calibration-Kit-SMA-2-Kits-included-1-Plug-and-1-Jack-NEW-/371738580436?hash=item568d5895d4 which I've used successfully on my N1996A, for now I'll try to use it on the 8753E too.  I have also bought two used 7 mm to SMA (female) adapters that I believe to be Amphenol.

When calibrating, is it necessary to allow for the 7mm - SMA adapters or is this automatically taken into account when the Cal measurements are done?

[trukresom]

Hi Gandalf_Sr,

When I look at the picture of your cal kit, it seems to me that the version at the right is contains exactly the same components
as mine from Heuermann. I think that these components are manufactured by Rosenberger.

... or is this automatically taken into account when the Cal measurements are done?

After the calibration procedure the reference plane for the measurements is situated somewhere in the
calibration kit usually a few mm ahead of the mating plane. And this is (hopefully) independent of the
adapters and cables used to connect the DUT. So if you change a measuring cable to connect your
DUT you should get the same results after a recalibration. At least this is what we expect.

[Gandalf_Sr]

Merci mon ami

[trukresom]

Merci mon ami

De rien !
et un Bonjour de la Provence

[Gandalf_Sr]

So I took a stab at running through the calibration menu on my 8753E but it's not very clear about where things are supposed to be connected, I think the left connector is the forward port and the right is the reverse (mine's a 2 port with the S bridge inside).

KE5FX linked to the VNA Help web site which I will read through tonight.  It looks like I might also need a calibration disk for the 8753E which I don't have - do I need one? Or the ability to connect to it via GPIB so I can use the program KE5FX linked to but that says I need a NI GPIB - USB interface, I have a cheap eBay Agilent 82357B which I strongly suspect is a fake and I've never tried out out and I'm not sure if that will work, I looked for "NI GPIB-USB-HS" and got a load of hits on eBay and Amazon but again, many people are saying that these are fakes.

I also need a simple step-by-step calibration guide that includes pictures of what to connect where. 

Any advice is much appreciated.

Thanks

[Gandalf_Sr]

OK, I found this Youtube video on doing a calibration and it's pretty much what I understood.  I have a few questions though:

1.  Shouldn't he have selected a Cal kit before he did the calibration?
2.  Once the full two-port calibration was done, wouldn't it make sense to save it somewhere?

[trukresom]

Hi Gandalf_Sr,

I think the left connector is the forward port and the right is the reverse

The forward parameters S11 and S21 are measured with the source signal coming out of port 1 and when measuring
reverse parameters S22 and S12 the analyzer switches the source generator to port 2.

It looks like I might also need a calibration disk for the 8753E which I don't have - do I need one?

If you had exactly one of the calibration kits which are listed on your analyzer in the CalKit selection menu then you don't
need a calibration disk because the coefficients of these are already stored in the analyzer. But with a non-HP calibration
kit to be exact, one should enter user-defined coefficients for this cal kit. But doing this seems to be very complicated.
First, because it is not clear how to determine the correct coefficients and secondly because it is complicated to enter these
coefficients into the machine. This is the reason why I suggested to select the HP 85035C CalKit on the 8753 together with
your SMA calibration kit as a starting point because they are very similar. At least up to 1 or 2 GHz this should not give too much error.

I also need a simple step-by-step calibration guide that includes pictures of what to connect where

I also had about the same difficulties when I started using a VNA 3years ago. I suggest to start with a simple introductory exercise.
Measuring the inductivity of a small air core inductor with some turns of enameled copper wire, something like this:
http://www.cettechnology.com/products/air-coil-inductors/
This can be done by measuring the S11 Parameter. We use only Port 1. First you have to decide which frequency range:
You select Start and Stop frequency. I suggest Start: 30 MHz Stop: 300 MHz. Later you should also eventually modify the IF-BW,
source power, averaging and the like. It is important to set up these parameters before calibrating. This because a modification
of some of these parameters makes the calibration obsolete.

Select the Cal Kit:
[Cal  ] --> <Cal Kit> --> <Select Cal Kit> --> <3.5mm D>
Then you can calibrate:
[Cal] --> <Calibrate Menu> --> <S11 1-port>
Connect your OPEN reference. Press <OPEN>. Wait for beep.
Connect your SHORT reference. Press <SHORT>. Wait for beep.
Connect your LOAD reference. Press <LOAD>. Wait for beep.
Press <DONE>

The machine is calibrated now.  To measure our coil we solder it onto one of these SMA connectors
http://uk.rs-online.com/web/p/sma-connectors/6163422/
But before soldering the coil we connect this connector on the VNA without the coil and we enter the following:
[Meas] --> <s11>
[Format] --> <Smith Chart>
Now we should see a small vertical trace at the right side of the chart. This is essentially caused by the extension
of our measuring cables by the electrical length of this SMA connector. We try now to compensate this by pressing
[Cal] --> <More> --> <Port Extensions> --> <Extension ON> --> <Extension PORT1>
Now we turn the knob clockwise until we see a single point, the ideal open.
Now we solder our coil onto the terminals of this SMA connector. (Please remove the connector to do this to avoid
static discharge of the soldering iron)
Reconnect the connector with the coil and press
[Marker]
Turn the knob to a frequency of interest and read the value of the inductivity on the top of the screen.

Shouldn't he have selected a Cal kit before he did the calibration?

Yes, absolutely. This is not clear. At some point he is pressing the Preset Button. On my machine this selects the 7mm Cal Kit as default.
There are 2 other things in the video which I find not very good:
1. He should avoid the mechanical stress on the small connector on port 2 with the filter pcb suspended on it by using a second cable.
2. When he is removing the calibration standards from the measuring cable he is turning the standard ccw against the cable connector.
   This also should be avoided. Only the nut of the cable connector should be rotated.

Once the full two-port calibration was done, wouldn't it make sense to save it somewhere?

Yes, this could be done but personally I don't have much confidence in the long time stability of a calibration.

[Gandalf_Sr]

Awesome information, thank you

I do have some calibration parameters for my cheap SMA calibration kit (it's on the eBay page that I linked to), I don't know if these are sufficient for the 8753E but, if they are, presumably they can be entered somewhere - I'll have a look.

I appreciate the example test you listed, I'll walk through it later.

Thanks,

Ted

[hendorog]

Hi Gandalf_Sr,

When I look at the picture of your cal kit, it seems to me that the version at the right is contains exactly the same components
as mine from Heuermann. I think that these components are manufactured by Rosenberger.

There are also low cost cal kits from these guys:
www.sdr-kits.net
http://www.sdr-kits.net/Webshop/products.php?96&cPath=5

Which also use Rosenberger parts (possibly the same parts). They have also measured them against a genuine HP cal kit and determined at least some of the parameters which are needed to setup a User Cal Kit in an 8753. They also come with an individually measured load.

[KE5FX]

So I took a stab at running through the calibration menu on my 8753E but it's not very clear about where things are supposed to be connected, I think the left connector is the forward port and the right is the reverse (mine's a 2 port with the S bridge inside).

You have a built-in S parameter test set, so it can go both ways.  By convention the left connector is port 1 and the right connector is port 2.  When the analyzer is measuring Sxy, 'y' is the transmitting port (which can also receive reflected power) while 'x' designates the receiving port, which will have a 50-ohm termination.  When measuring S11 or S22, both forward and reflected power are monitored at the same port, and the other one isn't used at all.

KE5FX linked to the VNA Help web site which I will read through tonight.  It looks like I might also need a calibration disk for the 8753E which I don't have - do I need one? Or the ability to connect to it via GPIB so I can use the program KE5FX linked to but that says I need a NI GPIB - USB interface, I have a cheap eBay Agilent 82357B which I strongly suspect is a fake and I've never tried out out and I'm not sure if that will work, I looked for "NI GPIB-USB-HS" and got a load of hits on eBay and Amazon but again, many people are saying that these are fakes.

It's worth trying what you have.  The worst it can do is not work.  (Well, it could also lock up your VNA after a lengthy calibration process that you don't feel like repeating, but that particular pain is further down the road.)

So far, I haven't seen any confirmed fake NI GPIB-USB-HS adapters; all of the ones I've bought from China have looked good and worked well.

I also need a simple step-by-step calibration guide that includes pictures of what to connect where. 

The manuals are really very good -- go surf around on the Keysight page, searching for not only "8753ES" but also the earlier models like "8753D."

Also look for tutorials written for various EE courses that have found their way online.  Anything by Dunsmore is worth reading, such as this.  He designed a lot of the original hardware in the 8753A and also does some work with UC-Berkeley's EE program.

[KE5FX]

1.  Shouldn't he have selected a Cal kit before he did the calibration?

Yep.

2.  Once the full two-port calibration was done, wouldn't it make sense to save it somewhere?

Yes, for two reasons: one is that calibration is sort of a pain in the neck, given all the necessary swapping of standards and pressing of buttons.  The other is that all of the standards, cables, and fittings you're working are expensive and have a finite lifetime.  A happy calibration kit is one that's nestled safely in a closed box.

As trukresom suggests, calibrations don't last forever.  But this, more than anything else, is how you tell the difference between a crappy VNA and a good one.  The one you have is one of the good ones.  As long as the temperature is reasonably stable and your cables and connectors are cleaned, maintained, and used properly, you can expect most saved calibrations to be usable for weeks at a time, at minimum.

[G0HZU]

I do have some calibration parameters for my cheap SMA calibration kit (it's on the eBay page that I linked to), I don't know if these are sufficient for the 8753E but, if they are, presumably they can be entered somewhere - I'll have a look.

Setting up a USER cal kit on the 8753ES is not a lot of fun if you want to get the best out of the VNA and the cal kit.
If you look back at the start of this thread (two years ago) I listed the USER cal kit menu as a confusing mess on the 8753ES.

You can try playing with the cal kit manager program that Agilent (kind of?) adopted to see if that makes things easier but I did it the hard way using manual entry for all the cal kit data.

It can take a lot of work proving/tweaking a user cal kit to get the most from it. This is what makes the USER CAL menu system in the 8753ES such a pain to use because it isn't intuitive to use and it's very easy to select the wrong soft menu key or mess up the data entry.

The VNA cal kit manager program will probably be your best bet

http://www.vnahelp.com/ckman_new2.html


I found the 8753ES to be particularly frustrating to use when using multiple USER cal kits. I have several 'user' cal kits here at home and also at work depending on which test fixture system I use and the 8753ES isn't your friend when trying to juggle numerous cal kits. Otherwise, it is a fabulous VNA and the prices for them seem to be in freefall these days. Amazing value for money for the hobbyist

[Gandalf_Sr]

Thanks guys, I got the vnahelp program installed on my Windows 10 machine and it looks like it's running, problem is that, when I click on any help item, I get taken to this Microsoft website https://support.microsoft.com/en-us/kb/917607 that tells me "Error opening Help in Windows-based programs: "Feature not included" or "Help not supported"

I tried installing the Windows 8.1 help version but it tells me that it's not the right help for my OS

Sigh...

I can try on a different (Windows Vista) machine tomorrow.

[EDIT] If you go to this link

and follow the instructions in the post at "EricN1719 replied on  August 26, 2015" ....

you will have a fully working system that correctly opens .hlp files right from within the VNA help application - at least it worked for me on Windows 10 64 bit.

[G0HZU]

A happy calibration kit is one that's nestled safely in a closed box.

That's partly why (for use up to 3GHz) I've always made my own SMA cal kits for home use. They cost virtually nothing and can be thrown away (or the USER cal file edited) when they wear or misbehave and I don't care if I lose or misplace them. I've had very good results from my various homebrew cal kits and I still use my original one (must be 10 years old?) for casual stuff. It's not going to be as good as an 85033E cal kit but I have found that for sub 3GHz work there is very little difference. I only see a difference when taking very critical s2p models of components and the difference isn't really significant.
However, the performance quickly degrades above 3GHz and my homebrew cal kits would not be suitable for a 6GHz version of the HP8753E(S).

[Gandalf_Sr]

Just a heads up.  I haven't lost interest but my Mother in Law passed away yesterday and I won't get back to looking at this for a week or two; she was a lovely 83 year old who lost her struggle with cancer.  Thanks to all here for their ideas and support.

[Mosaic]

"That's partly why (for use up to 3GHz) I've always made my own SMA cal kits for home use"
 can u advise on how this is done please?
thx
Ancel

[hendorog]

As another option, instead of building your own you can just purchase the cal kit parts for $20:

http://www.sdr-kits.net/Webshop/products.php?49&cPath=5

These connectors are spec'd to 18GHz and (IMHO) should perform better than DIY but not as good as a kit from Dr Kirkby - because they don't come with a full set of cal data.

SDRKits also do cal kits in a box which come with a 4 wire measurement of the actual load supplied - which of course you can do yourself with a (calibrated) multimeter.

[G0HZU]

Yes, most people will be better off just buying a basic kit like the one above.

I made my homebrew SMA cal kit using surplus SMA PCB end launchers. It isn't as good as a commercial cal kit but see below for the SMA load. You can see how accurate it is right up to 3GHz. I optimised the short and load using a dremel to get the reference plane for both of them right at the end of the end launcher. Over the years I did a lot of useful design work with this cal kit here at home and it cost me nothing unless you include the £0.01 for the resistors in the load. The connectors were free. Salvaged from a scrap PCB.

The homemade load below was measured using a fairly modern Agilent VNA and a 13.5GHz Agilent Ecal module.

[hendorog]

Yes, most people will be better off just buying a basic kit like the one above.

I made my homebrew SMA cal kit using surplus SMA PCB end launchers. It isn't as good as a commercial cal kit but see below for the SMA load. You can see how accurate it is right up to 3GHz. I optimised the short and load using a dremel to get the reference plane for both of them right at the end of the end launcher. Over the years I did a lot of useful design work with this cal kit here at home and it cost me nothing unless you include the £0.01 for the resistors in the load. The connectors were free. Salvaged from a scrap PCB.

The homemade load below was measured using a fairly modern Agilent VNA and a 13.5GHz Agilent Ecal module.

Impressive 
What is limiting the performance above 3GHz do you think? i.e. would these Vishay RF resistors help?
http://nz.mouser.com/Vishay/Passive-Components/Resistors/High-Frequency-RF-Resistors/_/N-7jdaj?P=1z0zls5

[G0HZU]

I'm not sure what limits the performance above 3GHz but when I made it I tried testing SMD resistors from various manufacturers on a modern VNA at work (at work we have reels and reels of 100R SMD resistors from various makers) and I selected the 'best' 0805 or 0603 resistor package I could find amongst them all (when tested on a VNA) and then I bought a strip of them from Farnell and did a select on test with a 4 wire DMM to find two 100R resistors from that manufacturer/range that gave almost exactly 50R when in parallel. I think it took me a few tweaks to position them to get that return loss performance and it's stayed that good ever since. But it doesn't work well above 3GHz. Probably due to several reasons including the SMD package and my soldering. At the time I only had an old 3GHz HP8714B here at home and this homemade cal kit was good enough for the stuff I was doing. I just couldn't justify buying a proper cal kit. The VNA only cost me £250 back then (over 10 years ago) as it had a known fault and I bought it for scrap money, ordered the repair kit (for this fault) from Agilent for about £80 and the VNA sprang back to life.

[KJDS]

The key to high frequency precision loads is that there is length along the resistor. The ground needs to get closer. Travel along the line to the resistance and it's 50ohms. At the grounded end of the resistor then it's 0 ohms, but halfway along the resistor it is 25 ohms but that isn't reflected in the transmission line round it. Genuine cal standard loads use a rod resistor with the outer of the coax tapering in towards the grounded end.

[G0HZU]

To show what my old homebrew SMA SOL cal kit can do, here's a screenshot comparing ATC's own s parameter file of a 3.3pF ATC 800B capacitor against my measurement of the same capacitor using my homebrew SMA cal kit and a homebrew test fixture and my VNA here at home. The test range is 500MHz to 3GHz because the ATC data only starts at 500MHz for this 3p3 cap.

But you can see how well my measurement agrees for capacitance, quality factor and Zreal/imaginary up to the capacitor's own self resonance near 3GHz. My fixture is a very tight 1 port test fixture and measures the cap on its own and I suspect that the ATC measurement may have a subtle shift in the self resonance due to differences in the test fixture. My fixture is a 1 port fixture and ATC used a 2 port fixture. But my results are very good for a cal kit and fixture made from scrap parts that cost me nothing

The blue (wiggly) trace data is the official ATC data (port 2) and the red traces (port 1) are made using my homemade cal kit and fixture. To be able to make measurements like this (especially the Q measurement) takes great care with the whole test procedure and much attention to detail. But it shows what can be achieved.

[G0HZU]

See below for the same cap measured using my snazzy Agilent N4431-60006B 13.5GHz 4 port Ecal module. Interestingly, this manages to capture some of the subtle dips in the Q measurement that are present in the official ATC data file

http://www.keysight.com/en/pd-491111-pn-N4431B/rf-electronic-calibration-module-9-khz-to-135-ghz-4-port?cc=US&lc=eng

This is better than my homebrew cal kit but the differences are so tiny that they are insignificant. It would be interesting to see how well the commercial cal kits would fare here. I don't 'need' a low cost commercial cal kit here because I have the Ecal unit but I'd be happy to review a commercial SMA cal kit. I think I can do it VERY critically, with much stricter tests that the one here.