Author Topic: Kirkby calibration kit alternatives?  (Read 45374 times)

0 Members and 1 Guest are viewing this topic.

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Kirkby calibration kit alternatives?
« Reply #75 on: February 25, 2018, 02:45:40 pm »
Quote
I've attached the output pdfs for a couple of new graphs I added to the script. These show the difference between the model and the s1p file and I think these are comparable to what G0HZU shows in post #47.
Yes, your data looks much better now.

That METAS VNA tool used by suj looks to be very powerful. With my simulator I had a go at plotting the fringing capacitance of the Agilent/Keysight 85033 OPEN vs frequency and I think it looks like the plot below. I hope I've got this right. It only changes a small amount and these corrections only really affect the flatness above about 5GHz.

With a cheapo SMA cal kit like the Kirkby kit the SMA based standards won't have the same design integrity as a proper 3.5mm kit so the corrections are much more important because they have to correct for the flaws over most of their 7GHz range. The male SHORT in the Kirkby kit does not look good at all with its limited corrections. I couldn't live with it unless it was corrected better in the cal kit file.
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #76 on: February 25, 2018, 04:26:38 pm »
Also, I just realised that METAS VNA Tools II tool you are using is free.  :-+
but i need to email someone to get the download link. if click the download link, setupping outlook will popup  |O i just tried copy paste the email address into yahoo mail.

Yes, I had to do the same for gmail. That is a bit of a pain and is a very basic, manual approach they have taken to registration.
 
No reply yet. Cutting them some slack though because its free after all and it looks really handy.

They might have their hands full if all the VNA nuts on here start registering en masse.
 

Offline suj

  • Regular Contributor
  • *
  • Posts: 92
  • Country: pl
Re: Kirkby calibration kit alternatives?
« Reply #77 on: February 25, 2018, 04:38:03 pm »
...No reply yet. Cutting them some slack though because its free after all and it looks really handy.

They might have their hands full if all the VNA nuts on here start registering en masse.

I think you don't have a reply because today is Sunday and the institute is not working today  8). I didn't wait for the link too long.
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #78 on: February 25, 2018, 04:39:41 pm »
Quote
I've attached the output pdfs for a couple of new graphs I added to the script. These show the difference between the model and the s1p file and I think these are comparable to what G0HZU shows in post #47.
Yes, your data looks much better now.

That METAS VNA tool used by suj looks to be very powerful. With my simulator I had a go at plotting the fringing capacitance of the Agilent/Keysight 85033 OPEN vs frequency and I think it looks like the plot below. I hope I've got this right. It only changes a small amount and these corrections only really affect the flatness above about 5GHz.

With a cheapo SMA cal kit like the Kirkby kit the SMA based standards won't have the same design integrity as a proper 3.5mm kit so the corrections are much more important because they have to correct for the flaws over most of their 7GHz range. The male SHORT in the Kirkby kit does not look good at all with its limited corrections. I couldn't live with it unless it was corrected better in the cal kit file.

Interestingly we now have a way of fixing the shortcomings in the correction data. Although it won't help if the VNA doesn't support L corrections of course, which goes back to your point of needing a better constructed kit which doesn't rely on the corrections as much.

Although there is always the option of dragging the data out and doing the calibration on the PC with modern software.

It would be interesting to compare the Dr Kirkby coefficents with ours on a tighter vertical scale on the same graph to see if there is any significant advantage one way or the other.

 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #79 on: February 25, 2018, 04:42:41 pm »
...No reply yet. Cutting them some slack though because its free after all and it looks really handy.

They might have their hands full if all the VNA nuts on here start registering en masse.

I think you don't have a reply because today is Sunday and the institute is not working today  8). I didn't wait for the link too long.

Yes - it is pretty clear based on the email method and the delay over the weekend. The registrations are processed manually.
AI hasn't taken over yet :)
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Kirkby calibration kit alternatives?
« Reply #80 on: February 25, 2018, 05:34:31 pm »
If you haven't downloaded it already here is the bundled s1p custom data on the Kirkby male SHORT in a text file and I've listed the custom corrections below.

Quote
Male short. (Use SHORT(F) or SHORT -M- depending on model of VNA - see notes above) **
Minimum frequency:  0 GHz
Maximum frequency:  7.0 GHz
Offset Zo: 50.0 ohms
Offset loss: 3.0 G ohm/s
Offset delay: 57.997  ps

There is only a delay correction and I get the phase response below. This does not look good but I guess the male part of the kit will be used less often. The older 8753 VNAs kind of default to 'insertable device' calibrations so they would often use both male and female kits for a full two port calibration. In this case it's debateable if the Kirkby kit is 'fit for purpose' for a decent 6GHz lab VNA like an 8753?
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Kirkby calibration kit alternatives?
« Reply #81 on: February 25, 2018, 05:51:58 pm »
With my manual optimiser for the Lx corrections I can flatten the response and it's easier if I drop the Zo slightly. But I get a slight fixed phase offset with the flat response. Not sure what has happened there. Maybe it needs to be measured again on a VNA?
How does it look if you try and optimise it with METAS?

It looks better with just a Zo tweak but maybe also try it with Lx corrections. It's a shame this Lx correction can't be added to the 8753 firmware in some way. It might be possible to bodge it in to the firmware with a crude patch?
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #82 on: February 25, 2018, 05:53:30 pm »
If you haven't downloaded it already here is the bundled s1p custom data on the Kirkby male SHORT in a text file and I've listed the custom corrections below.

Quote
Male short. (Use SHORT(F) or SHORT -M- depending on model of VNA - see notes above) **
Minimum frequency:  0 GHz
Maximum frequency:  7.0 GHz
Offset Zo: 50.0 ohms
Offset loss: 3.0 G ohm/s
Offset delay: 57.997  ps

There is only a delay correction and I get the phase response below. This does not look good but I guess the male part of the kit will be used less often. The older 8753 VNAs kind of default to 'insertable device' calibrations so they would often use both male and female kits for a full two port calibration. In this case it's debateable if the Kirkby kit is 'fit for purpose' for a decent 6GHz lab VNA like an 8753?

Yes I think you have a point. Even with corrections it is not great.

This is what the optimiser spits out when I remove the data above 7GHz. L1 has hit the limit.

Code: [Select]
offs_delay = 54.877614 ps
offs_loss = 2.058719 Gohm/s
offs_Z0 = 51.050105 ohm
L0 = -3687.287669 * 1e-12 H
L1 = 10000.000000 * 1e-24 H/Hz
L2 = -978.814153 * 1e-33 H/Hz^2
L3 = 2713.293761 * 1e-42 H/Hz^3

Residual RMS error : -48.553224 dB


 

Offline suj

  • Regular Contributor
  • *
  • Posts: 92
  • Country: pl
Re: Kirkby calibration kit alternatives?
« Reply #83 on: February 25, 2018, 06:28:51 pm »
How does it look if you try and optimise it with METAS? It looks better with just a Zo tweak but maybe also try it with Lx corrections.

Results from Metas. A better result if the optimizer can change "Offset Z0" - column "Agilent definition". Anritsu and R&S probably don't support Z0 changes and the effects can be seen for Z0 = 50 Ohms.





Normalized to S1P.
Black - data from S1P
Red/brown - HP (with optimized Z0)
Orange - fixed Z0



« Last Edit: August 01, 2019, 09:57:01 am by suj »
 

Offline suj

  • Regular Contributor
  • *
  • Posts: 92
  • Country: pl
Re: Kirkby calibration kit alternatives?
« Reply #84 on: February 25, 2018, 06:56:22 pm »
...It's a shame this Lx correction can't be added to the 8753 firmware in some way.

By the way. It should be possible to do it off-line with Metas. It has a built-in driver for the 8753 and 8720 series (as well as 8510C, ENA, PNA). You can make measurements and off-line error correction using this software. Definitions of the "short" element that you can prepare have the option of defining L.
The procedure is shown in the tutorial on the program's website.
It looks like a Swiss army knife  ::)
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #85 on: February 25, 2018, 07:03:28 pm »
...It's a shame this Lx correction can't be added to the 8753 firmware in some way.

By the way. It should be possible to do it off-line with Metas. It has a built-in driver for the 8753 and 8720 series (as well as 8510C, ENA, PNA). You can make measurements and off-line error correction using this software. Definitions of the "short" element that you can prepare have the option of defining L.
The procedure is shown in the tutorial on the program's website.
It looks like a Swiss army knife  ::)

Bingo. I was thinking of using scikit-rf, but Metas would no doubt be more straight forward.

Really looking forward to getting hold of it.
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Kirkby calibration kit alternatives?
« Reply #86 on: February 25, 2018, 07:22:10 pm »
Quote
By the way. It should be possible to do it off-line with Metas.
Wow, that sounds great. I hope to get a copy of METAS soon :)

See below for a quick and dirty OPEN and SHORT made from a pair of SMA F-F bullets. One has a commercial SC end cap for the short. Sadly, the two bullets are from different manufacturers so they are not going to be the same but the end cap for the short is a known part and it isn't expensive. Maybe $9?

So maybe if I buy some Amphenol SMA F-F bullets from Farnell ($8 ea?) we could have a go at making a cheap cal kit for the masses?

But for now, have a go with METAS to see how flat you can get the phase data for these two bad boys below?


 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #87 on: February 25, 2018, 07:23:09 pm »
I was having a heck of time with this one. The fit wasn't lining up with the Metas one by suj

Then I found a bug in the model calculation for the Lx params. Fixed that and tada, now it lines up :)
Although it does tend to find another result at 53 ohms Z0. This one took a bit of nudging.

Code: [Select]
Optimization results:
offs_delay = 32.331122 ps
offs_loss = 6.333923 Gohm/s
offs_Z0 = 60.858025 ohm
L0 = 872.429450 * 1e-12 H
L1 = 5981.654164 * 1e-24 H/Hz
L2 = 37028.127991 * 1e-33 H/Hz^2
L3 = 3160.750516 * 1e-42 H/Hz^3

Residual RMS error : -53.679006 dB


 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #88 on: February 25, 2018, 07:54:24 pm »
Quote
By the way. It should be possible to do it off-line with Metas.
Wow, that sounds great. I hope to get a copy of METAS soon :)

See below for a quick and dirty OPEN and SHORT made from a pair of SMA F-F bullets. One has a commercial SC end cap for the short. Sadly, the two bullets are from different manufacturers so they are not going to be the same but the end cap for the short is a known part and it isn't expensive. Maybe $9?

So maybe if I buy some Amphenol SMA F-F bullets from Farnell ($8 ea?) we could have a go at making a cheap cal kit for the masses?

But for now, have a go with METAS to see how flat you can get the phase data for these two bad boys below?


I did a bit of blue sky thinking:

There is nothing preventing us from having different coefficients for different frequency ranges.
The fit will be better by limiting the frequency range it is over.

Also, if the cal is done offline in software, then changing coefficients (i.e. recalibrating using a different set of coefficents) is purely software and doesn't require another sweep. So it would be technically possible to do a full sweep and change cal definitions on the fly. The definitions would match the real cal standards very closely.

In fact I expect this could be as good or better than a high quality cal kit - assuming it only has one set of coefficents for the entire range of the kit. Perhaps this is how the eCal units work?
 

Offline suj

  • Regular Contributor
  • *
  • Posts: 92
  • Country: pl
Re: Kirkby calibration kit alternatives?
« Reply #89 on: February 25, 2018, 08:07:44 pm »
But for now, have a go with METAS to see how flat you can get the phase data for these two bad boys below?

Two cases for the "open" element and two for the "short" element. For each element once Z0 opened and once set to 50 \$\Omega\$

Code: [Select]
Agilent Definition
Standard Type: Open
Offset Z0 (Ohm): 60.711892
Offset Delay (ps): 27.289762
Offset Loss (GOhm/s): 3.507799
C0 (E-15 F): 414.232256
C1 (E-27 F/Hz): -6871.633507
C2 (E-36 F/Hz^2): -4959.099088
C3 (E-45 F/Hz^3): 470.623016

Agilent Fit Error
RMS Error
0.001900
Max Error
0.003349
_______________________________________________
Agilent Definition
Standard Type: Open
Offset Z0 (Ohm): 50.000000
Offset Delay (ps): 58.437897
Offset Loss (GOhm/s): 2.711473
C0 (E-15 F): -304.566047
C1 (E-27 F/Hz): -8873.121112
C2 (E-36 F/Hz^2): 2622.781977
C3 (E-45 F/Hz^3): -343.691735
Agilent Fit Error
RMS Error
0.002197
Max Error
0.003648
_______________________________________________
Agilent Definition
Standard Type: Short
Offset Z0 (Ohm): 56.059063
Offset Delay (ps): 41.557996
Offset Loss (GOhm/s): 3.717653
L0 (E-12 H): -222.842514
L1 (E-24 H/Hz): -19759.249841
L2 (E-33 H/Hz^2): 11091.105982
L3 (E-42 H/Hz^3): -651.849387
Agilent Fit Error
RMS Error
0.001378
Max Error
0.003844
_______________________________________________
Agilent Definition
Standard Type: Short
Offset Z0 (Ohm): 50.000000
Offset Delay (ps): 32.074189
Offset Loss (GOhm/s): 4.973737
L0 (E-12 H): 492.787861
L1 (E-24 H/Hz): -6736.857074
L2 (E-33 H/Hz^2): -3868.927304
L3 (E-42 H/Hz^3): 506.666193
Agilent Fit Error
RMS Error
0.001556
Max Error
0.004609

Open element, Z0 open:





Open element, Z0=50 \$\Omega\$:





Short element, Z0 open:





Short element, Z0=50 \$\Omega\$:



« Last Edit: August 01, 2019, 09:59:47 am by suj »
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #90 on: February 25, 2018, 08:18:50 pm »
See if you can manually scale the normalised phase charts - the glitch is killing all of the detail due to the auto scaling.

From what I am seeing it is hard to get a good match 0-6GHz. Much better dividing into 0-3GHz and 3-6GHz.
 

Offline suj

  • Regular Contributor
  • *
  • Posts: 92
  • Country: pl
Re: Kirkby calibration kit alternatives?
« Reply #91 on: February 25, 2018, 08:20:33 pm »
I think I need to change the way the phase is displayed. Error during changing the phase from -180 to +180 obscures the situation.
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Kirkby calibration kit alternatives?
« Reply #92 on: February 25, 2018, 08:26:46 pm »
Hi suj, thanks for putting the data into METAS.

I noticed that the loss response of that SMA F-F (OPEN) on S11 looks a bit poor up to 3GHz and I tried another SMA F-F bullet and this one looks better. Both are clean and look similar but one has lower loss. I assume there must a difference in materials used and/or subtle differences in Zo along the connector.

Quote
Also, if the cal is done offline in software, then changing coefficients (i.e. recalibrating using a different set of coefficents) is purely software and doesn't require another sweep. So it would be technically possible to do a full sweep and change cal definitions on the fly.
That would be neat, I'd definitely be interested in being able to do this. My original DIY SMA kit must be about 12 years old now and it would be nice to retweak the corrections for any wear. I spent ages optimising the cal kit for it when I first got my HP8714B VNA and this involved lots of repetitive tests and calibrations. I don't really want to do this the hard way again.

I've dug out the HP8714B from under the stairs and powered it up for a bit of nostalgia. I do like the big old CRT display on it even though it is just mono.
« Last Edit: February 25, 2018, 08:28:50 pm by G0HZU »
 

Offline in3otd

  • Contributor
  • Posts: 19
  • Country: it
Re: Kirkby calibration kit alternatives?
« Reply #93 on: February 25, 2018, 08:34:37 pm »
Then I found a bug in the model calculation for the Lx params. Fixed that and tada, now it lines up :)

 :-[ uh, oh, you mean the indices of the Lx coefficients were off by 1in the std_model_s_params() function, right? I'll upload a corrected version shortly, thanks!
 

Offline suj

  • Regular Contributor
  • *
  • Posts: 92
  • Country: pl
Re: Kirkby calibration kit alternatives?
« Reply #94 on: February 25, 2018, 08:34:47 pm »
Hi,
I changed the files on my server. After refreshing, the normalized data should look OK.
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #95 on: February 25, 2018, 08:37:44 pm »
Then I found a bug in the model calculation for the Lx params. Fixed that and tada, now it lines up :)

 :-[ uh, oh, you mean the indices of the Lx coefficients were off by 1in the std_model_s_params() function, right? I'll upload a corrected version shortly, thanks!

Haha, welcome! I was considering sending you an email to let you know we were discussing your code here!

Yes that was the issue. I just figured you hadn't used L params much due to the 8753 et al not supporting them.
 

Offline in3otd

  • Contributor
  • Posts: 19
  • Country: it
Re: Kirkby calibration kit alternatives?
« Reply #96 on: February 25, 2018, 08:51:39 pm »
Also, if the cal is done offline in software, then changing coefficients (i.e. recalibrating using a different set of coefficents) is purely software and doesn't require another sweep. So it would be technically possible to do a full sweep and change cal definitions on the fly. The definitions would match the real cal standards very closely.

If you have a VNA where you can access to the calibration coefficients, like the 8753 and, I think, most of the ones not targeted at hobbyists, you can simply use something like scikit-rf to do the calibration using the calkit S parameters directly (no need of building any model of them) and then write the computed calibration coefficients back to the VNA. I don't  know if this is what the METAS VNA Tools can actually do or if it allows only a post processing of raw data acquired from the VNA.
Then another issue is the calkit repeatability, with cheap stuff you never know how it will behave tomorrow  :)
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #97 on: February 25, 2018, 09:09:39 pm »
Also, if the cal is done offline in software, then changing coefficients (i.e. recalibrating using a different set of coefficents) is purely software and doesn't require another sweep. So it would be technically possible to do a full sweep and change cal definitions on the fly. The definitions would match the real cal standards very closely.

If you have a VNA where you can access to the calibration coefficients, like the 8753 and, I think, most of the ones not targeted at hobbyists, you can simply use something like scikit-rf to do the calibration using the calkit S parameters directly (no need of building any model of them) and then write the computed calibration coefficients back to the VNA. I don't  know if this is what the METAS VNA Tools can actually do or if it allows only a post processing of raw data acquired from the VNA.
Then another issue is the calkit repeatability, with cheap stuff you never know how it will behave tomorrow  :)

Yes, very true. In this way we can remove the model entirely.

With the cal kit repeatability issue, I think that is where buying high quality open/shorts and characterising them should produce a better result than DIY.

Also, the proposal I mentioned a while back, was to not use an open standard at all when the VNA has a female SMA port.
As long as it is characterised then is there any point in connecting a standard to it, aside from a slight noise reduction? This is one less connection to make when calibrating.

I need an s1p file from a calibrated VNA which has a female SMA port left open. Since we have optimisation software, I will try to create a model for it.
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Kirkby calibration kit alternatives?
« Reply #98 on: February 25, 2018, 10:07:48 pm »
Quote
Also, the proposal I mentioned a while back, was to not use an open standard at all when the VNA has a female SMA port.

I've done that with a known SMA barrel at the end of an SMA cable and I have a very 'short' male SHORT made from a cut down Suhner (good quality) SMA connector. See the image below.

It is possible to just use an SMA end cap for the short instead but the end cap will obviously spin as it is being undone so this can cause wear.  The short below has a nut that can still spin (so no wear issues) but this means it has some delay in it and the performance isn't great and I don't use this very often and I try to only use it below 1GHz.
« Last Edit: February 25, 2018, 10:10:58 pm by G0HZU »
 

Offline hendorog

  • Super Contributor
  • ***
  • Posts: 1617
  • Country: nz
Re: Kirkby calibration kit alternatives?
« Reply #99 on: February 25, 2018, 10:32:55 pm »
Quote
Also, the proposal I mentioned a while back, was to not use an open standard at all when the VNA has a female SMA port.

I've done that with a known SMA barrel at the end of an SMA cable and I have a very 'short' male SHORT made from a cut down Suhner (good quality) SMA connector. See the image below.

It is possible to just use an SMA end cap for the short instead but the end cap will obviously spin as it is being undone so this can cause wear.  The short below has a nut that can still spin (so no wear issues) but this means it has some delay in it and the performance isn't great and I don't use this very often and I try to only use it below 1GHz.

For the short I was intending to use these parts. These are the ones used in the SDR kits. I have one of their Male kits so I already have the male short below.

Female short:
http://sdr-kits.net/documents/32Z114-000L5.pdf

Male short:
http://rosenberger.de/ok/images/documents/db/32Z111-000L5.pdf

The wear issue is a bit tricky. I believe the Male below can spin internally (am not near it right now to verify), but the Female obviously can't and so would need to be held still while the nut is tightened. The specs for both say 500 mating cycles. However I am sure they will change their characteristics over their lifetime - as in3otd alluded to.


 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf