|
Here is an interesting video from W0QE where he shows the effect just a fraction of an inch (a few thousands in fact) dimensional calibration error makes in your measurements due to the delay error:
#90: Building VNA Calibration Loads - Revisited: Others of his videos dealing with cal loads: #78: Building VNA Calibration Loads: #89: VNA Calibration Through a Cable - What Can Go Wrong: In you are looking for a good series on Smith Charts and their use, his series is worth it. Just start with #1. |
|
Dr. David Kirkby from Kirkby Microwave Ltd
On Fri, 11 Oct 2019 at 21:21, Mat Breton <ab8vj@...> wrote:
Here is an interesting video from W0QE where he shows the effect just a You gave two identical links. I looked at only the first. I don¡¯t follow his logic. I think however he has failed to set up his 8753 properly for the calibration kit with the result he is fooling himself. If he has set up the 8753 properly, he certainly doesn¡¯t say so. He uses an HP 85033D calibration kit. Attached is a photograph of what the Smith Chart should look like if he calibrated with his 85033D kit and measured his 85033D kit. The short should *not* be a dot, but an arc that starts at the 9 o¡¯clock position on the Smith Chart and ends at about the 11 o¡¯clock position. On a NanoVNA at 1.5 GHz you would an arc half that length I got this photograph after setting up my 8753ES properly to use the HP 85033D kit he has. I calibrated with my 85052B kit, which has exactly the same coefficients as his kit, then measured the short. -- Dr. David Kirkby, Kirkby Microwave Ltd, drkirkby@... Telephone 01621-680100./ +44 1621 680100 Registered in England & Wales, company number 08914892. Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom |
|
David,
toggle quoted message
Show quoted text
I did mention in the video at 1:43 showing a slide where the 85033D calibration loads produced the arcs that you and the manual mentions in chapter 5.? I then added electrical delay so viewers would feel more comfortable with dots for the O/S/L loads. If you disagree with anything I have done you are welcome to address these concerns directly to me. Larry, W0QE On 10/11/2019 6:13 PM, Dr. David Kirkby from Kirkby Microwave Ltd wrote:
On Fri, 11 Oct 2019 at 21:21, Mat Breton <ab8vj@...> wrote:Here is an interesting video from W0QE where he shows the effect just aYou gave two identical links. I looked at only the first. |
|
Hello Larry,
Adding the delay is not required and inappropriate. If the calibration kit is input to the VNA correctly, all C and L parasitic coefficients, offset delay and losses as defined by the kit, then you remeasure the same, so called validation, the "arcs" David mentions need to be present! Not because there is an error, but because that is EXACTLY what those calibration pieces represent. To be sure that a proper cal is accomplished, an independent verification kit is required. A sliding load or perhaps an APC-7 open and short which has zero delay would be appropriate. If you were to measure such an element, you would find viola' that the zero delay validation piece would display correctly, a DOT. By adding this extension say 60 pS or 70 pS to produce a DOT from a 3.5 mm kit, you have in fact built an error into your measurements going forward! I believe this issue, particularly with the 3.5 mm cal kit is presented in a number of application briefs from HP, Keysight, R&S and Wiltron. I'll see if I can find these references. Alan |
|
Here is a cut and paste from the Microwave101 web site.
I hope by reference to THEM, I am observing the proper copyright method!! Regards, Alan ======================================= Why doesn't an open or short look like a dot on the Smith chart? After performing a one-port or two-port calibration on your network analyzer, what do you expect to see when you measure the open standard from the calibration kit? Some people expect to see a perfect dot at the edge of the Smith chart. After all, a perfect open has a reflection coefficient of one at an angle of zero degrees at all frequencies. And since you are measuring the open that you calibrated with, it should look perfect, right? This sounds reasonable, but in fact the measurement of the open will look like an arc instead of a dot. This is a common source of confusion for many VNA users. This measurement result is actually correct - there is nothing wrong with the network analyzer system or the calibration. You see, the cal kit's open standard is not an ideal open. It has nonlinear fringing capacitance, electrical length, and possibly some loss due to radiation. All of these imperfections are included in the model of the open as described in the calibration kit. Because the goal of the calibration is to make your measurement system give accurate results, and the open standard is not ideal, it would be inaccurate for the calibrated measurement system to report that it is an ideal open. In most cal kits, this same argument applies to the short standard as well. For mechanical construction reasons, the actual short circuit is a small distance away from the calibration plane, so the short standard looks like a short circuit at the end of a very short transmission line. When you measure it, the result is also an arc on the Smith chart. Author : Barry Source : VNAHelp.com ====================================== |
|
Alan,
toggle quoted message
Show quoted text
I understand what you are saying well.? Since any calibration standards will have some physical length you or the analyzer needs to subtract this length before you can make accurate measurements.? The 8753 analyzer family does this length correction automatically which is why the open and short cal. standards show as arcs.? I have never seen any small hobby type VNAs do this automatic correction. I was manually "un-correcting" by moving the measurement plane into the cal. standards to show that the original standards looked correct. Larry, W0QE On 10/11/2019 7:44 PM, alan victor wrote:
Here is a cut and paste from the Microwave101 web site. |
|
Dr. David Kirkby from Kirkby Microwave Ltd
On Sat, 12 Oct 2019 at 03:13, Larry Benko <xxw0qe@...> wrote:
Alan, No, neither the user nor the analyzer has to subtract any physical length. Doing that would give reasonable results, but certainly not as accurate as the 8753 is capable of. In fact, with proper support in the firmware, and correct use, the NanoVNA might outperform the 8753 in some measurements if one just subtracted a delay on the 8753. Just considering the open, there are 3 parameters that need to be taken into account. In order of decreasing importance these are 1) Offset length 2) Capacitance, which is frequency dependant 3) Loss, which is frequency dependant. The 8753 analyzer family does this length correction automatically which is why I don¡¯t know what you mean by ¡°automatically¡±. The 8753 has no way to automatically determine the offset length, capacitance or loss of the standards. The VNWA will show arcs for the calibration standards, and so will some Deepace products. They are all aimed at the amateur market Larry, W0QE Dave, G8WRB --Dr. David Kirkby, Kirkby Microwave Ltd, drkirkby@... Telephone 01621-680100./ +44 1621 680100 Registered in England & Wales, company number 08914892. Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom |
|
Dr. David Kirkby from Kirkby Microwave Ltd
On Sat, 12 Oct 2019 at 02:37, alan victor <avictor73@...> wrote:
Hello Larry, Alan, I concur with that. Certainly for any kits I have, the effect is most pronounced with the 3.5 mm kits. On the HP 85052B 3.5 mm kit, the arcs will do a full 360 degrees around the Smith Chart by about 16 GHz. Neither my 85050B (18 GHz APC7), nor 85054B (18 GHz N) show the effected as pronounced as the 3.5 mm kit. Dave --Dr. David Kirkby, Kirkby Microwave Ltd, drkirkby@... Telephone 01621-680100./ +44 1621 680100 Registered in England & Wales, company number 08914892. Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom |
|
Dave,
toggle quoted message
Show quoted text
Please educate me.? Suppose I build a product that has a SMA-F connector on it and you and I are going to measure S11 for my product.? The only way we are able to get the same answer is if there is an agreed upon definition as to where within the mated SMA connectors we are going to call the reference plane.? Agreed? The most accurate way to calibrate my measuring equipment is to connect a cable to it that is terminated in an SMA-M connector and then calibrate it with calibration O/S/L standards that have SMA-F connectors.? This eliminates any adapters being needed.? Agreed? Since my O/S/L standards have some physical length I will need to move the reference plane closer to the measuring equipment by some small amount.? To my way of thinking this can be done either by the measuring equipment automatically of manually by doing some line extension. Is this incorrect? Larry, W0QE On 10/11/2019 10:07 PM, Dr. David Kirkby from Kirkby Microwave Ltd wrote:
On Sat, 12 Oct 2019 at 03:13, Larry Benko <xxw0qe@...> wrote:Alan,No, neither the user nor the analyzer has to subtract any physical length. |
|
Dr. David Kirkby from Kirkby Microwave Ltd
On Sat, 12 Oct 2019 at 02:44, alan victor <avictor73@...> wrote:
Here is a cut and paste from the Microwave101 web site. To get another view of this same topic, see I would welcome any comments, either positive or negative. Dave. -- Dr. David Kirkby, Kirkby Microwave Ltd, drkirkby@... Telephone 01621-680100./ +44 1621 680100 Registered in England & Wales, company number 08914892. Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom |
|
Dr. David Kirkby from Kirkby Microwave Ltd
On Sat, 12 Oct 2019 at 05:53, Larry Benko <xxw0qe@...> wrote:
Dave, The MIL standard for the SMA connector defines the reference plane. The most accurate way to calibrate my measuring equipment is to connect a cable to it that is terminated in an SMA-M connector and then Yes. agreed.
*No, that¡¯s wrong.* You should 1) Configure the 8753 to use the 85033D coefficients. From memory, that¡¯s something like CAL (physical button on 8753) Select Cal kit (soft key) Press the soft key corresponding to the 85033D. Return Now just calibrate the instrument properly, then make your measurement. There¡¯s no need to mess around adding any delays - as Alan pointed out, that will screw your results up Larry, W0QE Dave, G8WRB -- Dr. David Kirkby, Kirkby Microwave Ltd, drkirkby@... Telephone 01621-680100./ +44 1621 680100 Registered in England & Wales, company number 08914892. Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom |
Mat Breton
