This is fantastic as usual. I eventually *will* be in a position to apply this knowledge, but even at the lower frequencies I'm constrained to for now, it really gives some great perspective.
Thanks Larry. With your callsign, I'm wondering if your intended audience are hams? At my local (very active) ham club, I would estimate that less than 5% understand ohms law. This material would be viewed as rocket science. Sad but true.
The cal standard coefficients in the VNA use a time delay, then frequency dependent polynomial capacitance for the open, and inductance for the short. So what you can do is measure your standard, then fit a model to it, then plug those parameters back into the VNA as a custom cal kit. I tried bootstrapping an SMA cal kit by creating zero length shorts with mercury, but didn’t work out too well. Ended up finding a pristine Maury 8050J 3.5mm kit on eBay.
Louis, I too have seen the 4th order polynomial (IIRC) for the fringing capacitance but that is way beyond my capabilities to measure. The minimal change I made to the "open" was a simple but worthwhile improvement (as you can see in the plots) and not intended to be a precision solution. I am open to any suggested improvements. Larry, W0QE
Would be interesting to see how bad of an calibration is still use full, for someone without experience it is hard to figure out how good it has to be. Maybe show how big the erros are. I know it depend on what you measure, in wich area of the smith the DUT is. But it would be handy to know and see what a cheap standart compared to a expensive one. Again thanks for the amazing videos always learn something new, SimSmith is the LTSpice of RF!
Heinrich, The worst errors due to the "open" being slightly too long will occur with impedances on the right side of the Smith chart. If you are trying to use the calibration standards at 1GHz my old design gave an approximate 1 degree error in Gamma. This error is in addition to what may be inside the analyzer itself but I thought a 1 degree error was pretty reasonable for home made loads. The slight shortening for the "open" reduced this error significantly and seemed like a worthwhile modification for those who can grind off .012" or .30mm to the "open". Your question of "how good" has no definitive answer but I try to trade off difficulty and cost versus accuracy. Earlier in life electronics was a profession and now it is a hobby which definitely has an effect on what I consider "accurate enough". Larry, W0QE
I know there is no definitive Answer, but i have the feeling that sometimes the people think they need extremely high accurate standards, but what they want to know is a result of +-5 or more percent. The modification is a good idea. I think it is important to know how accurate of an measurement you need, for standard current voltage and Frequency it is easy to tell if i have to know it down to the mV but for RF i never got the feeling. most of the time even as a professional it only has to be good enough chasing the last µm is not important and will cost allot of time and money.
Heinrich, I agree with you 100% but achieving a certain level of accuracy at 10 MHz will be easier than achieving it at 100 MHz and hugely easier than achieving it at 1 GHz. For some the hobby can become an obsession but if you enjoy what you are doing all is well. I never know what level of RF understanding my viewers have which is another consideration. Thanks for the comments and good or bad they are welcome. Larry, W0QE
Thanks. Good explanations and video. Is there a way to do a good home made calibration set without having another reference calibration kit ? Let's says using a high quality 50ohms load (for up to 26GHz for instances) and DIY the open and the short because they're simpler to build by yourself ?
Oliver, What I built did not require a reference calibration kit. The reference kit was used to verify what I built. I would have no idea how to build anything to go to 26GHz. I don't think that the short and especially the open are easier than the load due to fringing capacitance of the open. Larry, W0QE
@@w0qe Thanks Larry for taking time. Comparing the result of your DiY kit to the calibration of the high quality calibration kit, your determined that the fringing capacitor of the open of your DIY kit makes it longer that is should be. That's why you've grinded the central pin to improve it and make it closer to your better calibration kit. To an extend, can we assume that building an home kit using your technique we will end up with such an open that is too long and that we should also grind it's central pin ? My intend what not to use the 26GHz load for designing a calibration kit to such a frequency but knowing that this load has a 1.2 SWR up to 26GHz , I can sure assume it's going to be very good at 3Ghz on the nanoVNA . Seing no decrepancy (both being a "spot" at the center of the smith chart) between by home built load and the 26GHz load, i think I can assume the home built is "good" compared to the reference 26GHz one at least up to 3GHz. It 's to note this is due to the fact that a "spot" rotated is still a "spot" and any phase difference (delay) on a perfect load on a smith chart can not be seen. Please correct me if anything wrong in my understanding. Regards Olivier
@@olivierschmidt2344 Some old Hewlett Packard documentation I have refers to the correction of the "open" with a fairly complex polynomial that described the fringe capacitance so just shortening the pin a little is probably not the solution even though it did help at the lower frequencies. A lot of the old HP documentation is available on line and that might help you. If nothing else it is very interesting to try and understand what these engineers struggled with as the frequencies went higher and higher. Larry, W0QE
@@w0qe Kurt Poulsen de OZ7OU did some very detailed work on the subject. See for instance www.hamcom.dk/VNWA/SOT259%20calibration%20standards.pdf. he did even some electro magnetic simulation on the subject. Have not yet taken time to analyse and review his paper. The DG8SAQ VNA seems to be able to take into account some "modeling" of the load in order to take into account maybe some capacitiv effect in the resistor loads. It would wish that nanovna-saver took into account into their SW. By the way I mounted my SMD resistor on the loads a little bit different . As the resistor film on the SMD is mounted on the upper side (just under the marking), I flipped them before soldering on the flattened back surface of the connected. Doing so, the resistor film is just at the calibration plane...maybe the matter of 1mm...
![ARGENTINA vs. PERÚ [1-0] | RESUMEN | ELIMINATORIAS SUDAMERICANAS | FECHA 12](http://i.ytimg.com/vi/xtI25ENDDSY/mqdefault.jpg)
Thanks for all the nice comments. While home made loads are not perfect they can be quite good for most users.
Larry, W0QE
Great video Larry, thanks. I have had this problem at work when our technicians lose the cal kits.
This is fantastic as usual. I eventually *will* be in a position to apply this knowledge, but even at the lower frequencies I'm constrained to for now, it really gives some great perspective.
Excellent work! It's time to fire up my hobby cnc and make some opens :D
Absolutely amazing! Thank you very much for videos like this.
Cristiano, You are welcome.
Larry, W0QE
Thanks Larry. With your callsign, I'm wondering if your intended audience are hams? At my local (very active) ham club, I would estimate that less than 5% understand ohms law. This material would be viewed as rocket science. Sad but true.
Looking forward to that nanoVNA video, great job, as usual ...
Thanks Larry for another informative video
The cal standard coefficients in the VNA use a time delay, then frequency dependent polynomial capacitance for the open, and inductance for the short. So what you can do is measure your standard, then fit a model to it, then plug those parameters back into the VNA as a custom cal kit.
I tried bootstrapping an SMA cal kit by creating zero length shorts with mercury, but didn’t work out too well. Ended up finding a pristine Maury 8050J 3.5mm kit on eBay.
Louis,
I too have seen the 4th order polynomial (IIRC) for the fringing capacitance but that is way beyond my capabilities to measure. The minimal change I made to the "open" was a simple but worthwhile improvement (as you can see in the plots) and not intended to be a precision solution. I am open to any suggested improvements.
Larry, W0QE
Would be interesting to see how bad of an calibration is still use full, for someone without experience it is hard to figure out how good it has to be. Maybe show how big the erros are. I know it depend on what you measure, in wich area of the smith the DUT is. But it would be handy to know and see what a cheap standart compared to a expensive one. Again thanks for the amazing videos always learn something new, SimSmith is the LTSpice of RF!
Heinrich,
The worst errors due to the "open" being slightly too long will occur with impedances on the right side of the Smith chart. If you are trying to use the calibration standards at 1GHz my old design gave an approximate 1 degree error in Gamma. This error is in addition to what may be inside the analyzer itself but I thought a 1 degree error was pretty reasonable for home made loads. The slight shortening for the "open" reduced this error significantly and seemed like a worthwhile modification for those who can grind off .012" or .30mm to the "open". Your question of "how good" has no definitive answer but I try to trade off difficulty and cost versus accuracy. Earlier in life electronics was a profession and now it is a hobby which definitely has an effect on what I consider "accurate enough".
Larry, W0QE
I know there is no definitive Answer, but i have the feeling that sometimes the people think they need extremely high accurate standards, but what they want to know is a result of +-5 or more percent. The modification is a good idea. I think it is important to know how accurate of an measurement you need, for standard current voltage and Frequency it is easy to tell if i have to know it down to the mV but for RF i never got the feeling. most of the time even as a professional it only has to be good enough chasing the last µm is not important and will cost allot of time and money.
Heinrich,
I agree with you 100% but achieving a certain level of accuracy at 10 MHz will be easier than achieving it at 100 MHz and hugely easier than achieving it at 1 GHz. For some the hobby can become an obsession but if you enjoy what you are doing all is well. I never know what level of RF understanding my viewers have which is another consideration. Thanks for the comments and good or bad they are welcome.
Larry, W0QE
Thanks!
Thankyou so much, I found that extremely interesting.
Great
Thanks. Good explanations and video. Is there a way to do a good home made calibration set without having another reference calibration kit ? Let's says using a high quality 50ohms load (for up to 26GHz for instances) and DIY the open and the short because they're simpler to build by yourself ?
Oliver,
What I built did not require a reference calibration kit. The reference kit was used to verify what I built. I would have no idea how to build anything to go to 26GHz. I don't think that the short and especially the open are easier than the load due to fringing capacitance of the open.
Larry, W0QE
@@w0qe Thanks Larry for taking time. Comparing the result of your DiY kit to the calibration of the high quality calibration kit, your determined that the fringing capacitor of the open of your DIY kit makes it longer that is should be. That's why you've grinded the central pin to improve it and make it closer to your better calibration kit. To an extend, can we assume that building an home kit using your technique we will end up with such an open that is too long and that we should also grind it's central pin ?
My intend what not to use the 26GHz load for designing a calibration kit to such a frequency but knowing that this load has a 1.2 SWR up to 26GHz , I can sure assume it's going to be very good at 3Ghz on the nanoVNA . Seing no decrepancy (both being a "spot" at the center of the smith chart) between by home built load and the 26GHz load, i think I can assume the home built is "good" compared to the reference 26GHz one at least up to 3GHz. It 's to note this is due to the fact that a "spot" rotated is still a "spot" and any phase difference (delay) on a perfect load on a smith chart can not be seen. Please correct me if anything wrong in my understanding. Regards Olivier
@@olivierschmidt2344
Some old Hewlett Packard documentation I have refers to the correction of the "open" with a fairly complex polynomial that described the fringe capacitance so just shortening the pin a little is probably not the solution even though it did help at the lower frequencies. A lot of the old HP documentation is available on line and that might help you. If nothing else it is very interesting to try and understand what these engineers struggled with as the frequencies went higher and higher.
Larry, W0QE
@@w0qe Kurt Poulsen de OZ7OU did some very detailed work on the subject. See for instance www.hamcom.dk/VNWA/SOT259%20calibration%20standards.pdf. he did even some electro magnetic simulation on the subject. Have not yet taken time to analyse and review his paper. The DG8SAQ VNA seems to be able to take into account some "modeling" of the load in order to take into account maybe some capacitiv effect in the resistor loads. It would wish that nanovna-saver took into account into their SW. By the way I mounted my SMD resistor on the loads a little bit different . As the resistor film on the SMD is mounted on the upper side (just under the marking), I flipped them before soldering on the flattened back surface of the connected. Doing so, the resistor film is just at the calibration plane...maybe the matter of 1mm...
Larry, great video:)
Great video, TNX for the upload!
73 N8AUM