Keyboard Shortcuts
Likes
Search
Affordable VNA Cal Kit in 12~15Ghz
|
Hi there, I need to measure and tune some of the RF passive components i.e. filters, DC, ring resonators etc I made for my hobby projects. I would like to get some more precision but I don't have $5000+ laying around for a calibration kit. Looking at ebay, there are some DC-26.5g HP calkit offered at $2500 range and claiming "fully working". How would I know if it is "working" without sending it to Keysight for a verification? Is there any Cal Kit in the 12~15 ghz range that doesn't cost an arm and a leg, and provide definition files that I can load to a 8510C VNA? Have anyone tried to make their own open, short, load, through, then rent a calibrated VNA to characterize them to make their own poorman's cal kit? Thanks, Calvin |
开云体育Hi Calvindepending on your frequency range of interest, you can make your own cal kit. I made a SMA kit which works fine up to 6GHz, and a N kit which works to approx. 7GHz. However, to calibrate your own cal kit, you need to have access once to a 'real' cal kit,
then measure your own standards.
If you are interested, I can tell you more details about this.
Tobias HB9FSX
-------- Original message --------
From: "guancalvin@... [hp_agilent_equipment]"
Date: 10/7/17 05:03 (GMT+01:00)
To: hp_agilent_equipment@...
Subject: [hp_agilent_equipment] Affordable VNA Cal Kit in 12~15Ghz
?
Hi there,
I need to measure and tune some of the RF passive components i.e. filters, DC, ring resonators etc I made for my hobby projects. I would like to get some more precision but I don't have $5000+ laying around for a calibration kit.
Looking at ebay, there are some DC-26.5g HP calkit offered at $2500 range and claiming "fully working". How would I know if it is "working" without sending it to Keysight for a verification?
Is there any Cal Kit in the 12~15 ghz range that doesn't cost an arm and a leg, and provide definition files that I can load to a 8510C VNA?
Have anyone tried to make their own open, short, load, through, then rent a calibrated VNA to characterize them to make their own poorman's cal kit?
Thanks, Calvin |
|
Can't you send a DIY cal kit to a calibration lab for measurement? I don't have a VNA yet, but that was what I was planning to do unless I manage to stumble across someone who will do it for me for free.
I was reading papers on VNA cal last night trying to find copies of Engen & Hoer, 1979 and Eul & Schiek, 1988 that didn't cost an arm and a leg from IEEE. Sadly now joy there though I did find some good papers. Take a look at: Advances in Radio Science (2003) 1: 21–25 Calibration of vector network analyzers on the basis of the LRR-method Rolfe and Schiek, which is available via google scholar. They show 4 microstrip calibration standards that should be very amenable to DIY construction. There's a huge amount of algebra in the paper , but it's all trivial. It's just presentation of all the details of least squares (L2). One can do better using least summed absolute error (L1). In either case, reading the math is far worse than actually doing the computation. Have Fun! Reg |
|
Hi,
It's not a simple job to measure a cal kit, most cal labs will not "calibrate" an unknown set of components. I second the recommendation for Kirby Microwave. Dave's kits are excellent and a much better option than some unkown overpriced kit bought on ebay. Search the group for more information, it's all been said before. Robert G8RPI. |
|
I hope this won't be taken personally. But I think the OP is being done a disservice by being told he needs to buy something that he is clearly skilled enough to make himself.
It was said is that someone like Transcat won't measure S parameters for a set of passive components over a specified frequency range. That may be true, but I find it rather amazing if it is. NIST certainly will, but I expect that they're a lot more expensive. This is vastly less work than an instrument calibration, so it should be cheaper. There are no adjustments to be made. As for the feasibility, that is not an issue as was noted by others. For bleeding edge work there are references that get sent from lab to lab. The measurements are then reviewed and discussed at length. I'm quite certain that David's kits are a much better choice than name brand stuff bought on eBay in unknown condition. I would expect he provides a set of calibration sweeps to compensate for the unavoidable errors in the standards. He certainly is no more able to make perfect standards than anyone else. Are David's calibration sweeps as good as Keysight's? Quite unlikely. They can afford to get in house references measured by NIST. Keysight's price for their references reflects the costs and the sales volume. If the test fixture is designed to probe microstrip devices, then the article I cited is clearly the way to go. I was driving most of the day and doing algebra in my head without paper and pencil is unreliable at best. So I've not worked out the algebra for multiple devices with multiple connectors. An assumption of connector equivalence may be required to account for connector terms. At the moment I don't know. There is a vast literature on the problem of quantifying instrument and reference errors. All you need is as many measurements as there are errors in the references. And a willingness to let yourself get beat up for awhile by the algebra. It is certainly true, that in a commercial setting buying a set of references from Keysight is the prudent approach. But for a hobbyist it's silly. A story from a different realm. A salesman was calling on a company. While chatting with one of the employees he asked what they made. "Gears of all types." "Really? I thought you had to buy those." Yeah, I'm a reflection seismologist, not an EE. But 1D (transmission line) problems are the first thing we have to learn. After that we spend most of our time on 2D and 3D issues because the earth is not 1D. It's not EM, but the algebra is the same. I've been quite surprised at how intimidated people are by matrix algebra. Have Fun! Reg - |
开云体育I'm interested. I think I can get access to the a cal kit through work. 6GHz would be plenty since I don't have equipment that goes past 3GHz. Ed On 10/7/2017 3:52 AM, Tobias Pluess
tobias.pluess@... [hp_agilent_equipment] wrote:
??? |
|
Mel Swanberg
I'll make a third recommendation for Kirkby Microwave cal kits. I've purchased two of them, and the proof is in the verification tests. When my measurements closely match the supplied verification device, it just doesn't get better than that.
toggle quoted message
Show quoted text
One of my kits even came with custom coefficients for a number of different frequencies that I specified. When compared to what else is out there, you'd be hard pressed to find a better option. Mel - WA6JBD -------------------------------------------- On Sat, 10/7/17, robert8rpi@... [hp_agilent_equipment] <hp_agilent_equipment@...> wrote:
Subject: Re: [hp_agilent_equipment] Affordable VNA Cal Kit in 12~15Ghz To: hp_agilent_equipment@... Date: Saturday, October 7, 2017, 7:55 AM Hi, It's not a simple job to measure a cal kit, most cal labs will not "calibrate" an unknown set of components. I second the recommendation for Kirby Microwave. Dave's kits are excellent and a much better option than some unkown overpriced kit bought on ebay. Search the group for more information, it's all been said before. Robert G8RPI. |
|
On 10/07/2017 08:13 PM, Mel Swanberg wa6jbd@...
[hp_agilent_equipment] wrote: I'll make a third recommendation for Kirkby Microwave cal kits. I've purchased two of them, and the proof is in the verification tests. When my measurements closely match the supplied verification device, it just doesn't get better than that.I have two as well, good stuff. -Dave -- Dave McGuire, AK4HZ New Kensington, PA |
开云体育Thank you for all responses guys! I did look at Dave’s cal kit but the SMA version doesn’t cover my frequency range. Reginald, I will look at the paper you indicated. I’m not afraid of math. I used to do J.D. Jackson’s Classical Electrodynamics end chapter problems to do brain workout although I’m not a physicist:) I use my VNA to do some casual measurement of some 15ghz filters and couplers, surprisingly the uncorrected measurement result matched the manufacturer’s data sheet nicely. I guess I could just live with that. I understand that I can’t reach the full potential of my 8510C without a cal kit, oh well…I’m not doing academic research nor making a filter for a missile:) Regards, Calvin
|
|
Hi,
It depends on what you want to do. If you need measurements you need a cal kit with known coefficents, if you just want to adjust for "best" performance you can get away with less. I don't think anyone was saying you can't make your own cal kit, just that Dave's kit is probably a more cost effective option overall. You could aquire all the components, check them as best you can and then send for calibration only to find one or more was out of limits and off you go again. Don't forget you need a calibration measurement device at least 3 times better than your required accuracy of the cal kit. I had a play with this to produce just a open / short / load (OSL) set for scalar measurements (no phase or length) to 1GHz. I had access to a modern 8GHz VNA and cal kit (a Kirby one) and was amazed how bad anything other than a proper calibration load (and most the used calibration ones I tried) were in terms of return loss. Dave will do higher frequency SMA kits to special order You should contact him. No connection to Dave, just a happy customer. Robert G8RPI. |
|
"I used to do J.D. Jackson’s Classical Electrodynamics end chapter problems to do brain workout although I’m not a physicist:)"
That's a pretty serious amusement! The following is more for the benefit of other readers. You obviously will figure all the rest of this once you read the paper. The paper is solving classic least squares solution of: Gm=d where d=data measurements m=correction parameters G=system of equations to be solved There are probably MATLAB programs for doing this that already have the G matrix built. It seems to me a pretty obvious graduate level homework problem. Building the G matrix is the hard part. But the paper gives all the details, although I'd strongly suggest that you work through the algebra by hand to verify there are no typos. Make up 2-4 of the stripline boards shown in the paper as close to identical as possible using your normal construction techniques and measure them. Making 4 will cut the variance of your calibration coefficient estimates in half. Then do a short write up for QEX. If you decide to do this, get in touch with me directly and I'll see what the L1 solution looks like. I drop in and out of the list because of the traffic level and shifting interests. I also like to play guitar: jazz, blues, folk and traditional country ballads. Another alternative would be the 6 port network of Engen & Hoer. Once I get a VNA, making and characterizing standards will be one of my first projects. The EEVblog metrology group might be a good place to check on the matter. Have Fun! Reg |
|
Robert,
I would certainly agree that making an accurate load is the hard part. The techniques I'm referencing bypass that. They were developed by research staff at the various standards institutes and rely on clever algebra and first principles. Take a look at the paper. It's short and there is no need to read the math. Just look at the figures and read the introduction. The problem with making an accurate load is getting the reflection to be zero. The connector guarantees that it will never be zero. The TRR paper avoids that by allowing the reflection coefficients to be unknowns which are eliminated algebraically. OSL is convenient, but has the serious issue of requiring a calibrated instrument to characterize the load. Before you can do that accurately you need a calibration method that does not depend upon OSL. There are a slew of techniques that have been published. Most I can't access. Of the ones I could, TRR seemed the easiest to implement. I've not been able to obtain a copy of Engen & Hoer's 6 port paper, but it appears to be a tapped transmission line which should also be easy to implement but I think it requires maintaining tolerances of 0.001" at 15 GHz to achieve ~1% accuracy. Doable, but definitely not easy. Prior to Joseph Whitworth, no on knew how to make a flat surface. Whitworth became justifiably famous for showing how to do that in the 19th century along with consistent measurements to a millionth of an inch. The construction of a surface plate is a physical exercise in algebra. You have to make 3 at once and compare and correct them in a specific order. Have Fun! Reg |
|
Hello,
toggle quoted message
Show quoted text
contact me offlist with Papers which interest you. I won a lawsuit some years ago...as long as you are Content with paper printouts, Bob should be your uncle. --- With best regards Tam HANNA (emailing on a BlackBerry PRIV) Enjoy electronics? Join 6500 other followers by visiting the Crazy Electronics Lab at Am 8. Oktober 2017 15:55:36 MESZ schrieb "Reginald Beardsley pulaskite@... [hp_agilent_equipment]" : ? |
|
Calvin,
For your VNA measurement (12-15 GHz), and considering the budget, I would get a sliding load and a set of HP shorts/opens on ebay. I say "HP" because the calibration constants for those are published. Maury Microwave also made them and those numbers can be obtained too. All this can be purchased at reasonable cost if you get the parts as individual pieces or incomplete kits. Sliding loads are rarely used these days, so they show up at reasonable cost. HP and Maury made them. Also, get the free software for editing and sending calibration coefficients to the 8510 via HPIB. It's called VNA Cal Kit Manager. Your biggest problem will be understanding the terminology and conventions regarding the data entry into the analyzer. They didn't make it simple. Also beware of the "male"/"female" conventions with regard to port connectors. They changed them at some point. The 8510 was the last of the VNAs with high performance test sets. For many applications, the analyzer was good enough to use without calibration. However the test port cables are always a problem and this is where you are stuck. Vladan |
Dave McGuire