|
Re: NanoVNA Under The Covers
Hi Tuck(?)
Whether a load shows a return loss of 50 dB, 30 dB or 70 dB depends upon what load was used to calibrate the VNA. As you can see from the chart, all three loads exhibited better than 70 dB when they were used as the calibration source. On the other hand, these very same loads all exhibited in the 30's when checked after calibration with the other load. The nanoVNA divides the displayed frequency range into 101 "bins". The normal return loss for an open, short, and a nominal load are all known factors. So when you tell the VNA you are using a short it checks each of the 101 bins and determines what correction factor is needed to make that value 'normal". This is done three times, once each for open, short, and nominal load. Then those bin by bin correction factors are applied to any measurement you make on the theory that the calibration procedure has "nulled' out any inaccuracies. So there is no defined return loss inherent to a particular load. There is only loss as measured against a calibration standard. Anything can be a calibration 'standard'.... but if you use one that is reactive or of a non-standard resistance, the VNA assumes it is correct and reports all subsequent measurements against that standard. This principle, by the way, can be used to your advantage. Say you have devices to be measured at the far end of a piece of coax. If you apply the calibration standards at the far end of the coax instead of at the instrument you will effectively "null out" the coax and the values reported will be as if the far end device was connected directly to the VNA. It is a very useful technique. This VNA is an incredible value. I am thrilled with mine! Warren Allgyer WA8TOD PS: "Pete" thinks he is still the lap dog he was as a 10 pound puppy. At 105 lbs he is a lap full! |
|
Re: "Hand capacitance"
Hello Peter,
On the device itself. So I will elaborate a little more, as I would like to use one nanovna at the field, as a simple SWR measurement tool, and of course operate it handheld. Maybe it will not occur in a proper HF antenna? Would you isolate the S21 port, or simply the "open" calibration tool will suffice? |
|
Re: NanoVNA Under The Covers
Thank you Warren. This is no argument over a pin head. Do you believe any of your loads are 50dB or better or even 38dB or better. How do these figures degrade with Frequency?Sent from my Samsung Galaxy smartphone.
toggle quoted message
Show quoted text
-------- Original message --------From: Warren Allgyer <allgyer@...> Date: 05/08/2019 13:46 (GMT+01:00) To: [email protected] Subject: Re: [nanovna-users] NanoVNA Under The Covers I am not sure if I am adding to or further confusing this discussion which has descended into, in my opinion, an argument over the number of angels on a pin head.First, when a three point calibration is done on any VNA, the result will be to null the return from the load used in the calibration. When the calibration is finished the display will be showing the noise floor of the instrument, not the return loss from the calibration load. The only inference you can draw about the return loss of the load in this is that it is below the noise floor of the instrument.As an illustration, I have attached a chart showing the results of three different 3 point calibrations. I used two random "50 ohm" terminations from my junk box and the calibration load supplied with the nanoVNA. After each of the three calibrations I recorded the indicated return loss for all three loads in succession.As you can see from the chart, the reading for any of the the three loads when used as the calibration load is the noise floor i.e. "<-70 dB". What can also be seen from the chart is the symmetry of the readings with different calibration loads. For example, when I calibrate with Load 1, Load 2 indicates -53.6 dB. When I calibrate with Load 2 then Load 1 indicates a similar -54.3 dB.The point is that, by definition, it is impossible to get an accurate measurement of the return loss of a calibration load on the instrument that was calibrated by that load. The instrument has been set by the calibration to assume that load is perfect and will return only its noise floor level.The nanoVNA cannot and should not be expected to match the performance of a $15,000 lab instrument. But it does a remarkable job for its cost. In my "lab" I do not care about the difference in return loss between -40.5 and -41.0 dB. What I care about is at what frequency the return loss is best and, at that frequency, whether or not the load is reactive or not. I also care about having an instrument that can show a filter profile and allow me to optimize it for the performance I want. I care about what the input to my linear amplifier looks like for return loss and impedance. For these things I find the performance of the nanoVNA to be more than adequate and the value to be tremendous!
|
|
Re: NanoVNA Under The Covers
I don't? think this vna calibration just arbitrarily sets the 50ohm dummy to the noise floor. If it is based on a 3 load measurement then it must use Thomas Baier' s DG8SAQ procedure which is to find 3 parameters a, b,c in his article which will be a sort of curve fitting for S11 and S12. In some of the cheaper vna' s like the mini or the F4GOH version there is only one calibration, by the open load.My view is that like all instruments, it is only as good as the return loss bridge sitting inside the box, not the noise floor or number routines or other ekwctronicd. I have built several RLBs myself from HF to 1 GHz. Without elaborate shielding at HF and with the right fetrite cores you can get 60dB which drops to about 20dB at 1GHz. I do not imagine the SA602 bridge is better than this. There are minicircuit directional couplers that can do better but the cost is not worth the gain.?Sent from my Samsung Galaxy smartphone.
toggle quoted message
Show quoted text
-------- Original message --------From: Warren Allgyer <allgyer@...> Date: 05/08/2019 13:46 (GMT+01:00) To: [email protected] Subject: Re: [nanovna-users] NanoVNA Under The Covers I am not sure if I am adding to or further confusing this discussion which has descended into, in my opinion, an argument over the number of angels on a pin head.First, when a three point calibration is done on any VNA, the result will be to null the return from the load used in the calibration. When the calibration is finished the display will be showing the noise floor of the instrument, not the return loss from the calibration load. The only inference you can draw about the return loss of the load in this is that it is below the noise floor of the instrument.As an illustration, I have attached a chart showing the results of three different 3 point calibrations. I used two random "50 ohm" terminations from my junk box and the calibration load supplied with the nanoVNA. After each of the three calibrations I recorded the indicated return loss for all three loads in succession.As you can see from the chart, the reading for any of the the three loads when used as the calibration load is the noise floor i.e. "<-70 dB". What can also be seen from the chart is the symmetry of the readings with different calibration loads. For example, when I calibrate with Load 1, Load 2 indicates -53.6 dB. When I calibrate with Load 2 then Load 1 indicates a similar -54.3 dB.The point is that, by definition, it is impossible to get an accurate measurement of the return loss of a calibration load on the instrument that was calibrated by that load. The instrument has been set by the calibration to assume that load is perfect and will return only its noise floor level.The nanoVNA cannot and should not be expected to match the performance of a $15,000 lab instrument. But it does a remarkable job for its cost. In my "lab" I do not care about the difference in return loss between -40.5 and -41.0 dB. What I care about is at what frequency the return loss is best and, at that frequency, whether or not the load is reactive or not. I also care about having an instrument that can show a filter profile and allow me to optimize it for the performance I want. I care about what the input to my linear amplifier looks like for return loss and impedance. For these things I find the performance of the nanoVNA to be more than adequate and the value to be tremendous!
|
|
"Hand capacitance"
Hello team,
While I was measuring an handheld VU antenna, I noticed that the nanovna was suffering of "Hand capacitance" influence. My unit is an original one, shielded. What can I do to improve it? Have you noticed the same? What about calibration, do you think your hand will influence it? Cheers! LL |
|
Re: NanoVNA Under The Covers
I am not sure if I am adding to or further confusing this discussion which has descended into, in my opinion, an argument over the number of angels on a pin head.
First, when a three point calibration is done on any VNA, the result will be to null the return from the load used in the calibration. When the calibration is finished the display will be showing the noise floor of the instrument, not the return loss from the calibration load. The only inference you can draw about the return loss of the load in this is that it is below the noise floor of the instrument. As an illustration, I have attached a chart showing the results of three different 3 point calibrations. I used two random "50 ohm" terminations from my junk box and the calibration load supplied with the nanoVNA. After each of the three calibrations I recorded the indicated return loss for all three loads in succession. As you can see from the chart, the reading for any of the the three loads when used as the calibration load is the noise floor i.e. "<-70 dB". What can also be seen from the chart is the symmetry of the readings with different calibration loads. For example, when I calibrate with Load 1, Load 2 indicates -53.6 dB. When I calibrate with Load 2 then Load 1 indicates a similar -54.3 dB. The point is that, by definition, it is impossible to get an accurate measurement of the return loss of a calibration load on the instrument that was calibrated by that load. The instrument has been set by the calibration to assume that load is perfect and will return only its noise floor level. The nanoVNA cannot and should not be expected to match the performance of a $15,000 lab instrument. But it does a remarkable job for its cost. In my "lab" I do not care about the difference in return loss between -40.5 and -41.0 dB. What I care about is at what frequency the return loss is best and, at that frequency, whether or not the load is reactive or not. I also care about having an instrument that can show a filter profile and allow me to optimize it for the performance I want. I care about what the input to my linear amplifier looks like for return loss and impedance. For these things I find the performance of the nanoVNA to be more than adequate and the value to be tremendous! |
|
Re: NanoVNA Under The Covers
You are in cloud cuckoo land to think that I believe I am measuring 70dB. Read my post carefully and tell me what are we all measuring when the instrument says 70dB. And stop pushing your expertise and your products.?Sent from my Samsung Galaxy smartphone.
toggle quoted message
Show quoted text
-------- Original message --------From: "Dr. David Kirkby from Kirkby Microwave Ltd" <drkirkby@...> Date: 05/08/2019 12:39 (GMT+01:00) To: [email protected] Subject: Re: [nanovna-users] NanoVNA Under The Covers On Mon, 5 Aug 2019 at 09:56, tuckvk3cca <tuckvk3cca@...> wrote:> But several people here claim that 70dB is impossible, even with lab> equipment. Sent from my Samsung Galaxy smartphone.In order to measure a return loss of X dB? with an uncertainty of 1 dB youneed a load with a return loss of of around X + 10 dB. So to guarantee areturn loss of 70 dB you would need a load with a return loss of around 81dB.You are really living in cloud cuckoo land? if you believe that you canmeasure such a return loss.Do the maths and work out the dimensional tolerance of a connector neededto maintain the impedance to that necessary to have a return loss of 70 dB.Assume that the outer is perfect, but work out what the minimum and maximumdimensions on the inner. Then measure the diameter of the inner conductorsof various SMA plugs and tell us what you find.On precision 3.5 mm and N connectors, the inner conductor doesn¡¯t have 4 or6 slots. There are no slots, but instead thin internal fingers whichreduces the expansion in diameter.? (I would assume that the same is trueof metrology grade smaller connectors.)When the male SMA pin enters the female, the female expands in diameter ina way that you can¡¯t predict. The male pin is not like to be dead centre,which again will change the impedance.There is a scientific paper where the authors try to make correctionsbetween different connectors which will mate1) 3.5 mm and 2.92 mm2) 2.4 mm and 1.85 mm3) SMA and 3.5 mmThey worked out corrections for1) 3.5 mm and 2.92 mm2) 2.4 mm and 1.85 mmbut concluded it was not possible to do this with SMA & 3.5 mm due to thevariability of the semi precision connectors.You are living in cloud cuckoo land if you believe that junk about 70 dBreturn loss.I have observed |S11| < -100 dB when making connections with SMA. I don¡¯tbelieve the return loss is 100 dB or even 50 dB.Dave.> -------- Original message --------From: hugen@... Date:> 05/08/2019? 10:52? (GMT+01:00) To: [email protected] Subject: Re:> [nanovna-users] NanoVNA Under The Covers Limited by the effective dynamics> of SA602 and AIC3024, external noise cannot be eliminated and NanoVNA can> only measure 70dB return loss.> >> --Dr. David Kirkby,
|
|
Re: NanoVNA Under The Covers
I am afraid your statement is contradictory. 70dB is the minimum return loss calculated by this instrument. The noise floor is the minimum amplitude voltage the detectors can measure sure but they are both jobs and the same. The instrument cannot just ignore the dummy load and return 70dB if it is actually a 40dB dummy load, surely. I have another rather good dummy load which actually read 42dB. Are you saying that if I had used this for calibration it will now read 70dB?Sent from my Samsung Galaxy smartphone.
toggle quoted message
Show quoted text
-------- Original message --------From: alan victor <avictor73@...> Date: 05/08/2019 11:49 (GMT+01:00) To: [email protected] Subject: Re: [nanovna-users] NanoVNA Under The Covers Good day,I believe what you are saying is the noise floor of the instrument is 70 dB below the 0 dB reference line. The 0 dB reference line would represent 0 dB return loss or complete reflection of the incident voltage; a reflection coefficient of unity. Hence, the noise floor of the instrument would prevail and limit the minute reflected voltage that could be detected if the load were a perfect termination to the instrument. Hence a reflection coefficient of zero. This noise floor is 70 dB below the 0 dB reference line... Sure, I agree.I think if you drop the notion that the noise floor is the return loss value you are measuring will be on the same page.Have a great day,Alan________________________________From: [email protected] <[email protected]> on behalf of tuckvk3cca <tuckvk3cca@...>Sent: Monday, August 5, 2019 8:07 AMTo: [email protected] <[email protected]>Subject: Re: [nanovna-users] NanoVNA Under The CoversHere is the RLB that I have got. I have checked with manual measurements that at the HF end I can get 60dB return loss. I do not expect the nano vna can out perform this unit, but I want to make sense of its 70dB return loss. Is that so difficult for people to understand?Sent from my Samsung Galaxy smartphone.-------- Original message --------From: Peter Gottlieb <hpnpilot@...> Date: 05/08/2019? 03:52? (GMT+01:00) To: [email protected] Subject: Re: [nanovna-users] NanoVNA Under The Covers How would you measure that kind of return loss accurately?PeterOn 8/4/2019 9:50 PM, Dr. David Kirkby from Kirkby Microwave Ltd wrote:> On Mon, 5 Aug 2019 at 00:32, tuckvk3cca <tuckvk3cca@...> wrote:>>> That 1.005 swr is up at the 5GHz end I presume. At 500MHz you can get>> 70dB.?? I have seen some people complain that some such loads have a bump>> at 1GHz, origin unknown.Sent from my Samsung Galaxy smartphone.>> Why do you assume that you can get 70 dB at 500 MHz? I think you can be> pretty sure if Keysight believed they could get a 70 dB return loss up to> 500 MHz, it would be in the data sheet.>> I very much doubt NPL, NIST, METAS or any other national standards> laboratory could measure 70 dB return loss.>> Dave.>>
|
|
Re: NanoVNA Under The Covers
But several people here claim that 70dB is impossible, even with lab equipment.?Sent from my Samsung Galaxy smartphone.
toggle quoted message
Show quoted text
-------- Original message --------From: hugen@... Date: 05/08/2019 10:52 (GMT+01:00) To: [email protected] Subject: Re: [nanovna-users] NanoVNA Under The Covers Limited by the effective dynamics of SA602 and AIC3024, external noise cannot be eliminated and NanoVNA can only measure 70dB return loss.
|
Warren Allgyer
CT2FZI
F4WCV
