Kirkby calibration kit alternatives?

[rfspezi]

I'm in search for an affordable VNA calibration kit (6 GHz, SMA female) for my 8753E.
The Kirkbymicrowave kit seems to be a good choice - however price has risen from 380 to over 600 USD recently? 

That's why i hope somebody can recommend an alternative.

[Mechatrommer]

you are not paying the device or the material. you are paying the calibration data and service support. without it, you better off with DIY kit. which i had and compared with Kirkby Cal Kit. i'm not aware of cheaper option that can provide custom calibration data just as Kirkby did, sorry. as he put it himself, it means like... "you may believe whatever you want to believe". so its either Kirkby or DIY, imho. ymmv.

[Safar]

Some sellers from China on ebay sell kits with PNA trademark.

They wrote in description:
"Accurate electrical models are needed of the opens and short circuits. A number of eBay sellers sell low-cost calibration kits, but these are effectively useless, as they don't have the mathematical models of the opens and shorts necessary to load the constants into a VNA, and we doubt they have optimal phase difference between open and short. This kit is different, in that it is designed, rather than just assembled. With this calibration kit, there is a support page with information on what coefficients to use for Agilent/HP,R/S,ADVANTEST,ANRITSU...".

Price is just 218 USD for SMA kit and looks good generally BUT I'm don't know quality and have no any information about this kits and where to get this calibration data.

I suppose that would be clever to ask him first for characterization equipment and what data they provide for each kit and look for some info if it possible.

[Pitrsek]

SDR kits is offering one
https://www.sdr-kits.net/index.php?route=product/product&path=66_68_69&product_id=50
Depending on your frequency range, transmission line based calibration might be a viable option.

[hendorog]

Some sellers from China on ebay sell kits with PNA trademark.

They wrote in description:
"Accurate electrical models are needed of the opens and short circuits. A number of eBay sellers sell low-cost calibration kits, but these are effectively useless, as they don't have the mathematical models of the opens and shorts necessary to load the constants into a VNA, and we doubt they have optimal phase difference between open and short. This kit is different, in that it is designed, rather than just assembled. With this calibration kit, there is a support page with information on what coefficients to use for Agilent/HP,R/S,ADVANTEST,ANRITSU...".

Price is just 218 USD for SMA kit and looks good generally BUT I'm don't know quality and have no any information about this kits and where to get this calibration data.

I suppose that would be clever to ask him first for characterization equipment and what data they provide for each kit and look for some info if it possible.

Judging by the pictures of the VNA they show in the auction, these are complete rubbish.

[Tony_G]

Some sellers from China on ebay sell kits with PNA trademark.

They wrote in description:
"Accurate electrical models are needed of the opens and short circuits. A number of eBay sellers sell low-cost calibration kits, but these are effectively useless, as they don't have the mathematical models of the opens and shorts necessary to load the constants into a VNA, and we doubt they have optimal phase difference between open and short. This kit is different, in that it is designed, rather than just assembled. With this calibration kit, there is a support page with information on what coefficients to use for Agilent/HP,R/S,ADVANTEST,ANRITSU...".

Price is just 218 USD for SMA kit and looks good generally BUT I'm don't know quality and have no any information about this kits and where to get this calibration data.

I suppose that would be clever to ask him first for characterization equipment and what data they provide for each kit and look for some info if it possible.

Judging by the pictures of the VNA they show in the auction, these are complete rubbish.

Interesting. That text is copied directly from Dr Kirkby's auction. I wouldn't trust a cal kit from someone willing to copy text like that.

Full disclosure, i have 2 Kirkby kits and have been very haopy with them.

[Mechatrommer]

Some sellers from China on ebay sell kits with PNA trademark.

They wrote in description:
"Accurate electrical models are needed of the opens and short circuits. A number of eBay sellers sell low-cost calibration kits, but these are effectively useless, as they don't have the mathematical models of the opens and shorts necessary to load the constants into a VNA, and we doubt they have optimal phase difference between open and short. This kit is different, in that it is designed, rather than just assembled. With this calibration kit, there is a support page with information on what coefficients to use for Agilent/HP,R/S,ADVANTEST,ANRITSU...".

Price is just 218 USD for SMA kit and looks good generally BUT I'm don't know quality and have no any information about this kits and where to get this calibration data.

I suppose that would be clever to ask him first for characterization equipment and what data they provide for each kit and look for some info if it possible.

Judging by the pictures of the VNA they show in the auction, these are complete rubbish.

anything without calibration data are... including those heavily used brand name cal kit many sold in ebay, from accuracy stand point imho. i was to reply long post to the OP describing the pro to Kirkby cal kit, but since he's looking after cheaper option, i decided a short version of a reply. those china ebay just copy pasted Kirkby ads, less custom calibration data and less support service. and remember the china version only provides female or male version not both. Kirkby provides both gender, including throughs, a pen drive containing hand made calibration data with unique serial number to the cal kit and a -3dB verification attenuator, so we can compare the respond plot from calibrated HP8720D compared to our VNA so we know where our VNA stands. as soon we want to purchase a complete set from cheap china matching Kirkby's, the cost will be about the same if not slightly cheaper than. as i simply concluded in earlier post, anything without calibration data will be not much better or worse than a DIY kit. ymmv.

[hendorog]

Some sellers from China on ebay sell kits with PNA trademark.

They wrote in description:
"Accurate electrical models are needed of the opens and short circuits. A number of eBay sellers sell low-cost calibration kits, but these are effectively useless, as they don't have the mathematical models of the opens and shorts necessary to load the constants into a VNA, and we doubt they have optimal phase difference between open and short. This kit is different, in that it is designed, rather than just assembled. With this calibration kit, there is a support page with information on what coefficients to use for Agilent/HP,R/S,ADVANTEST,ANRITSU...".

Price is just 218 USD for SMA kit and looks good generally BUT I'm don't know quality and have no any information about this kits and where to get this calibration data.

I suppose that would be clever to ask him first for characterization equipment and what data they provide for each kit and look for some info if it possible.

Judging by the pictures of the VNA they show in the auction, these are complete rubbish.

anything without calibration data are... including those heavily used brand name cal kit many sold in ebay, from accuracy stand point imho. i was to reply long post to the OP describing the pro to Kirkby cal kit, but since he's looking after cheaper option, i decided a short version of a reply. those china ebay just copy pasted Kirkby ads, less custom calibration data and less support service. and remember the china version only provides female or male version not both. Kirkby provides both gender, including throughs, a pen drive containing hand made calibration data with unique serial number to the cal kit and a -3dB verification attenuator, so we can compare the respond plot from calibrated HP8720D compared to our VNA so we know where our VNA stands. as soon we want to purchase a complete set from cheap china matching Kirkby's, the cost will be about the same if not slightly cheaper than. as i simply concluded in earlier post, anything without calibration data will be not much better or worse than a DIY kit. ymmv.

That is not quite the full story. The cheap eBay kits clearly use poor components whereas the SDR kits use decent quality components. I'd expect that the SDR kits product will be quite a bit better than a DIY attempt.

Here is another low cost option :- If you know someone who has a cal kit then you can purchase good quality parts and characterise them. There is a Matlab script which will work out the coefficients.

[G0HZU]

I think it's possible to make a cheap but decent DIY SMA cal kit for use up to maybe 3GHz. I did this many years ago and I still use this cal kit today for many applications. The short and load and open were all DIY using SMA end launchers and I worked out the correction factors myself for the user cal file.

I get very good results with this cal kit and I typically use it to help me model SMD components up to a few GHz.

Once above 3GHz I'd prefer to use a used but decent commercial cal kit, eg a healthy 3.5mm 85033E mechanical cal kit or a 3.5mm Ecal kit. I don't think I'd want to pay $600 for the Kirkby kit. Old/used commercial cal kits can suffer damage and wear so some caution would be needed but I'd still prefer to go down this route compared to a cheapo SMA based alternative no matter how much time and effort went into measuring the cal coefficients.

[Safar]

Interesting. That text is copied directly from Dr Kirkby's auction. I wouldn't trust a cal kit from someone willing to copy text like that.

It is really interesting. I have not seen Kirkby description before. Agree.

[hendorog]

I think it's possible to make a cheap but decent DIY SMA cal kit for use up to maybe 3GHz. I did this many years ago and I still use this cal kit today for many applications. The short and load and open were all DIY using SMA end launchers and I worked out the correction factors myself for the user cal file.

I get very good results with this cal kit and I typically use it to help me model SMD components up to a few GHz.

Good point, I had forgotten that you had done that. It is very well done and is light years ahead of the eBay offering.
https://www.eevblog.com/forum/testgear/agilent-8753es-and-other-vnas-this-kind/msg1063141/#msg1063141

In comparison, the cheap eBay kit is shown in the pictures at about 10dB return loss on the load at 3GHz. It looks to be only usable below 500MHz perhaps.

Here is a useful page as well for DIY cal kits:
http://www.qsl.net/in3otd/electronics/VNA_calkit/calkit.html

Once above 3GHz I'd prefer to use a used but decent commercial cal kit, eg a healthy 3.5mm 85033E mechanical cal kit or a 3.5mm Ecal kit. I don't think I'd want to pay $600 for the Kirkby kit. Old/used commercial cal kits can suffer damage and wear so some caution would be needed but I'd still prefer to go down this route compared to a cheapo SMA based alternative no matter how much time and effort went into measuring the cal coefficients.

The issue with that is the availability of those used kits. They are very rare at reasonable prices, the sellers seem to think that even the boxes are worth a fortune.

Overall, I think that the simplest approach is to purchase some quality parts, get them tested by someone with a real cal kit, and then determine the coefficients using the script in the link above.

[G0HZU]

Thanks. I had a quick look over the Kirkby website and there are a few things that give me slight concern.

•Difference between the phase predicted by the mathematical model of the opens, and the actual phase of the opens is <4° from DC to 7 GHz
•Difference between the phase predicted by the mathematical model of the shorts, and the actual phase of the shorts is <4° from DC to 7 GHz

It would be nice to see some test  setup info and some data on this. My first reaction is to run to the hills because this spec of 4 degrees doesn't sound good... maybe I'm misinterpreting what is meant here?  .. or maybe it's a typo on the website and the spec is much better than this?

Also, the kit appears to use regular SMA F-F bullets with the standard screwed at the end. It looks cheap... The closeup images of the short and open don't inspire me. I wouldn't expect a regular SMA bullet to qualify as being part of a VNA cal kit that runs to 7GHz. Maybe these are special versions of the SMA F-F. I'm used to the Suhner ones at work. Also, what happens if the cap rotates? Do you tighten it again with a torque wrench and hope the cal is still good? Maybe it works a lot better than it looks?

An ex colleague of mine bought one of these kits recently and I could maybe try and borrow it and test it. At work, I've got access to new and UKAS calibrated PNA VNAs and various calibrated mechanical kits and several Ecal modules and some £££ Rosenberger test cables. I think the cables cost about £1500 each!

[hendorog]

An ex colleague of mine bought one of these kits recently and I could maybe try and borrow it and test it. At work, I've got access to new and UKAS calibrated PNA VNAs and various calibrated mechanical kits and several Ecal modules and some £££ Rosenberger test cables. I think the cables cost about £1500 each!

I vote hell yes!! Please do if you can swing it.

[G0HZU]

Sadly, he's unlikely to lend it to me unless he goes on holiday because he uses it every day for his business. So it might take a while.

You posted up a link to that (fringe) cal coefficient tool using GNU Octave but it looks like it is still being developed.

For my homebrew cal kit here I embed keysight's polynomial for the fringe capacitance (for the OPEN) into an RF simulator and make all the coefficients variable. Then I use a simple tline model plus the polynomial result to predict the phase result of an ideal model. I then compare against the homebrew open after a mechanical or Ecal calibration.  On the real homebrew standard I used a fine rotary tool to optimise the end of the open and I got very good agreement out to well beyond 3GHz. But I only made this kit for use with my first VNA and so it only had to be good to 3GHz. It's lasted many years and I still use it a lot. If I made another I'd expect to get better results for the short and open and maybe get it to work up at higher frequencies too. I'm not sure how valid my approach for the fringe capacitance was but it seemed to work OK on the simulator. Back then I would have measured and optimised my open and short at work on an old 6GHz HP8753ES with an 85033E cal kit.

[orin]

Thanks. I had a quick look over the Kirkby website and there are a few things that give me slight concern.

•Difference between the phase predicted by the mathematical model of the opens, and the actual phase of the opens is <4° from DC to 7 GHz
•Difference between the phase predicted by the mathematical model of the shorts, and the actual phase of the shorts is <4° from DC to 7 GHz

It would be nice to see some test  setup info and some data on this. My first reaction is to run to the hills because this spec of 4 degrees doesn't sound good... maybe I'm misinterpreting what is meant here?  .. or maybe it's a typo on the website and the spec is much better than this?

A fair proportion of that 4 degrees will be the uncertainties in measuring the actual phase of the standards.  According to the datasheet, the 8720D's uncertainty is close to 2 degrees in the 2-8GHz range for a 1.0 reflection coefficient.

Also, the kit appears to use regular SMA F-F bullets with the standard screwed at the end. It looks cheap... The closeup images of the short and open don't inspire me. I wouldn't expect a regular SMA bullet to qualify as being part of a VNA cal kit that runs to 7GHz. Maybe these are special versions of the SMA F-F. I'm used to the Suhner ones at work. Also, what happens if the cap rotates? Do you tighten it again with a torque wrench and hope the cal is still good? Maybe it works a lot better than it looks?

In the Kirkby SMA kit I have, the female standards do indeed use an F-F adapter (and the male standards seem to use an M-F adapter).  The F-F adapters have flats so they can be held by a wrench when being connected, so you shouldn't be holding them by the cap.  Given the variation in SMA connector design, I'd guess the results using the male open on its own with varying female SMA connectors were unacceptable.  Adding the adapters ensures a consistent interface to the actual male open and must give better overall results or Dr Kirkby wouldn't have done it that way.

[G0HZU]

FWIW here's a phase comparison plot that may be of interest. This is a relative phase measurement of my homebrew OPEN.

The two measurements were take many years apart and one was with an old HP8753ES with 85033 cal kit at work and the other is with my modern 13.5GHz Ecal module on a fairly modern ENA VNA. This was taken today.

There is a difference and this could be wear and tear and it could be the torque. I used different torque tools.
But the result looks sensible to me. The vertical blip in the data is a 0/360 degree wraparound in the simulator so ignore this.

Next to this is an old plot of my simulation model of the 85033 OPEN that is based on the data in the cal kit as below. I used the delay and the C0, C1, C2 and C3 data in the model and ran it on the simulator. It is compared to an 8753ES doing a self measurement of its OPEN.

60  ! OPEN:  $ HP 85033 Open
70  !    Z0  50.0 $ Ohms
80  !    DELAY  29.243E-12 $ Sec
90  !    LOSS  2.2E+9 $ Ohms/Sec
100 !    C0  49.433E-15 $ Farads
110 !    C1 -310.131E-27 $ Farads/Hz
120 !    C2  23.168E-36 $ Farads/Hz^2
130 !    C3 -0.16E-45 $ Farads/Hz^3

When I was messing with my HP8714B many years ago I calibrated a company HP8753ES with an 85033 cal kit and then used it to remeasure its own OPEN standard. I then exported the data in an S1p file. If I compare the S1p file to the mathematical model for the C0, C1, C2 and C3 polynomial and the delay and compare relative phase I get the plot on the right below. It looks like a flat line.

The VNA is measuring it's own cal kit 'OPEN' so it will display a response of a very tiny length of transmission line that has an open with some fringing capacitance at the end. It does agree with the model. Obviously, the VNA is measuring its own reference here so the result looks good but the point is the simulator model I created does seem to agree very closely.

[G0HZU]

A fair proportion of that 4 degrees will be the uncertainties in measuring the actual phase of the standards.  According to the datasheet, the 8720D's uncertainty is close to 2 degrees in the 2-8GHz range for a 1.0 reflection coefficient.

My suspicion is that due to the fact that the kit uses regular SMA F-F bullets and screw on caps the SHORT will have stray inductance between the interface between the end of the SMA bullet and the metal face/pin in the end cap. There won't be a perfect interface here so there could be a tiny air gap that introduces inductance. This on its own would introduce phase error and it may be that this could be as bad as a few degrees by 6GHz. I don't see any correction polynomial for this inductance on the website and this is probably because not many VNAs support this form of correction. Most only correct for the fringe capacitance in the open.

So the limitations of the kit could be influenced by this inductance for one thing. A regular Keysight 85033 SHORT will perform very well here because it uses a 3.5mm connector with an air dielectric. So there is no dielectric 'gap' at the metal plate at the end of the short as you would get with an SMA connector.

[hendorog]

You posted up a link to that (fringe) cal coefficient tool using GNU Octave but it looks like it is still being developed.

Yes it's fairly experimental, but I think it should do the job.

I haven't tried it in anger, but I did get it working at least - a while ago. I seem to remember I had to modify it to use a different S-param library than the one they recommend.

One of these days I'll have another crack at it - I have a new contact who has a cal kit who can do the reference measurement for me.

[G0HZU]

For my homebrew SMA cal kit here at home I used an RF simulator (that permits equations/scripts) to model the delay and the fringe capacitance and this allowed me to tweak the delay and C0, C1, C2 and C3 using a manual/iterative approach to flatten the phase error on my OPEN. It takes a couple of minutes using the UP/DWN keys on the keyboard to tweak C0, C1 etc to get it flat.

The plot below shows my OPEN being compared with the model after I had optimised the coefficients in the simulator. The OPEN was measured on my ENA analyser after an Ecal and the data compare against the model. It agrees very well

[technogeeky]

I just wanted to point out these male SMA references available for about $20 in the USA.

They manufacturer link goes back to mRS miniVNA Tiny and Pro's website.  After more and more searching, it seems like these guys are the manufacturer (or at least they have made measurements of the 50 ohm SMA load part).

The measurements they offer are here. I doubt this is sufficient for converting into a format you could input into a VNA characterization, but I just wanted to point out that they exist.

I am going to get a pair just as an additional set of open/short/load calibrations since my current best set are homemade, and because they have them in stock in their Atlanta, GA location which is down the street from me.

[Gandalf_Sr]

G0HZU

I know you're an expert on VNAs and you may remember that I bought an 8753E a while back, I also got one of the Kirkby Cal Kits.  The truth is that, I never really got to understanding the calibration process properly - I'm a trained RF engineer but have forgotten a lot of things.

I'd like to request that you mentor me and others here to bridge the gaps in our knowledge; if you're willing; perhaps in a new thread or we could do it here?

[G0HZU]

If choosing an alternative, a lot depends on what the kit will be used for.

Option 1. If you are just going to do fairly casual 1 port VNA measurements up to 3GHz and will be typically looking at antennas or basic matching networks then you might as well make your own from modified SMA end launchers and use 2 x 100R tight tolerance (or select on test) SMD chip resistors to make the 50R load. It wouldn't be that difficult to work out a reasonable USER cal file based on guesstimates of the delays in the short and open. It would cost less than $18 even with new parts? A fairly sloppy version would work to 1GHz but something tight and tidy would be OK to 3GHz for casual measurements.

Option 2.  If you want to do the same but go up to 6GHz then it gets tougher and the DIY cal kit will not perform well here unless you can generate decent correction coefficients for it. This would mean you would need access to a decent VNA and cal kit to help you set up the USER cal file for the DIY kit. A step up from this would be to buy one of the little SMA kits for maybe $100 and try and tweak up the USER cal data to improve it. Even then it is going to be inferior to a proper 3.5mm cal kit from Agilent etc.

Option 3. If you want to do fairly accurate 2 port modelling of DUTs then it gets a bit more serious because you might want to determine the K factor (stability) for an amplifier design for example. Or maybe measure passive SMD parts to create user models of them whilst capturing package resonances. A carefully designed and tweaked DIY SMA kit with an optimised USER cal file could just about do this up to 2 or 3GHz but if you wanted to do this to 6GHz then I think the best option is a healthy/used 85033 cal kit from HP/Agilent/Keysight. As long as it is not damaged it should work well. You would ideally need to be using expensive RF test cables up at 6GHz as well. i.e. cables with minimal phase distortion when flexed. It starts getting expensive very quickly and the cables gradually degrade (in terms of phase and amplitude stability) with regular use.

Somewhere in the middle of all this is the Kirkby cal kit. You can consider it too good for option 1 and good enough for option 2. I'm not sure about option 3 if there is a 6GHz requirement. I would want a proper 3.5mm cal kit for this.

[G0HZU]

G0HZU

I know you're an expert on VNAs and you may remember that I bought an 8753E a while back, I also got one of the Kirkby Cal Kits.  The truth is that, I never really got to understanding the calibration process properly - I'm a trained RF engineer but have forgotten a lot of things.

I'd like to request that you mentor me and others here to bridge the gaps in our knowledge; if you're willing; perhaps in a new thread or we could do it here?

This would be best done with a youtube video but, sadly, my presentation skills and video making skills are dire so I'm not really the best person for this. You have a fabulous analyser there that is highly capable and there are loads of powerful things you can do with a VNA like that, provided you have a decent cal kit and it is set up properly. I've not used the Kirkby cal kit but I think it's safe to assume its accuracy up to 3GHz will be very good. Good enough for quite serious work. It's the performance of the kit in the 4-7GHz range I'd be unsure about. It might be really good here too but I'd expect it to need some kind of correction polynomial for the type of SHORT it uses.

[Gandalf_Sr]

OK, thanks for the advice.  I'm travelling right now but I will try to re-calibrate my VNA and then ask the questions along the lines of - Am I doing it right?

[Mechatrommer]

Anyway, tuning or modelling a diy or stray unknown kit will require a very well calibrated vna in the first place right? Not many people can do that. One of them is kirkby, other than g0hzu... if g0hzu can produce a kit with custom calibration, with proof of vna standard used and cert of calibration at cheap that will be a good option. Vna users, used or cheap version will only want a reliable kit to start with to do their job, not dicking around with sim program and no access to devices capable of modelling a kit with some level of unarguable certainties.