Suggestions wanted for HP/Agilent VNA to cover at least 6 GHz, ideally 20 GHz.

I'm looking to purchase a used HP or Agilent VNA. My initial interst is antennas, but I'd like to extend that to filters too.

I don't really want to spend more than ￡10000 or S15000, and would like to keep it a bit less, but if there were specific advantages in spending a little over ￡10000/S15000, I would. But that's the sort of figure I am looking at. I need coverage to at least 6 GHz, but 20 GHz would be a bit more future-proof. I don't really care too much about coverage below 50 MHz - if need be, I could buy a cheap USB unit which plugs into a PC for that.

I'm interested in any thoughts on the following models, or any other suggestions.

• 8720D
• 8753ES
• 8719ES

I'm keen to hear of any specific problems or "gotchas" with these, such as

• You need accessory X, and they are hard to find
• Model Y gives a lot of problems with dry joints
• The connectors on the front of model Z are are easily damaged and impossible/expensive to get fixed

Dave

YOu pay for accuracy.. How accurate? When I was testing filters for Satphone I used a S100K Anritsu with 0.1dB accuracy after jig calibration. 0~6GHz
The 8719ET/8720ET/8722ET is a great one with Error Correction option.

To be frank, it is not something I'd give much consideration too, though I don't think absolute accuracy is going to be that critical to me. My main interst is antennas, and certainly for that accuracy is not that critical. I think dynamic range will be more critical to me, as I've had issues in the past of testing antennas when the attenuation between test antennas is too great to make meaningful measurements. I note that the more modern instruments seem to have much better dynamic range than the older ones.

I'm a bit surprised you needed to spend S100,000 to get 0.1 dB I must admit.

I did look at a few Anritsu models, but my previous experience has been support is not as good as with HP/Agilent. A friend has an HP VNA too, and it would make sence to write code which will work with either. But his is an older 3 GHz model, which is not much use to me.

Dave

Typical prices:
8720A with NIST calibration S10.9K
8720C with NIST Cal S14.4K
8720D with Nist Cal S21.9K
E8362B with Nist Cal S27.9K

I personally would not go any older than a 8720D, as you can get a lot of different calibration types with that, the scope screen will probably not be shot, and the connectors are still probably serviceable.

I would seriously consider the options! 1 Hz step size and time domain capability do not add much to a used NA price, but can really come in handy!

You can, of course, go cheaper on ebay, etc, but on something as delicate as a network analyzer, it might be a poor choice. The older ones are no longer repairable by Agilent, so if it dies...you might be stuck.

I have a HP 8753. A old work horse and very recommended if on low budget and if yearly calibration not is necessary. But I would think twice before buying a unknown used 8753. You can get them relative cheap at Ebay, but if you need to repair something can that cost just as much as an used instrument. I even doubt that new spare parts for HP VNA instruments are available any longer.

What internal calculations a VNA can do is of less interest for me, as it can be better done in my computer, and that is where I want to save the results anyway. Do not even use the VNA display as all live data is transferred to the computer screen over GPIB.
For me, doing practical antenna work at some distance from VNA, is it also a advantage to be able to look at a big computer screen, getting much more information then is possible to show on the relative small VNA screen. Also must most of my work be well documented. Without GPIB I would have to copy data to diskettes, and then copy that data to the computer, renaming the file as it not is practical to set file names in 8753 and without good naming, after 10 savings, I do not remember which file is which.

These old instruments are a bit noisy due to its fans, and cause a lot of heat, and I prefer to have as much as possible of the VNA control in the computer. For that reason is something like this of interest as my next VNA: http://www.coppermountaintech.com/pr...LANAR%20804/1/

I saw the copper mountain stuff at a recent show, and it certainly is affordable. But it seemed slower do to external laptop processing, and the guy demonstrating it had trouble getting the TDR to work. I would get a loaner and see if you like it first.

Interesting, for the kind of antennas I make, is response wanted at the same moment as shape is adjusted. From HP8753 do I get full S-parameter update on computer screen, including some processing, 3-4 times pro second, and slower then that is not wanted.
Was these Copper Mountain VNA's slower then that?

I do not remember the speed, but I remember thinking that it should be a little faster. Also, it might have had something to do with the laptop the guy was using and if it was USB 2.0 or 3.0. No reason it should not do 10 updates per second--there is not that much data going by and USB 3.0 can do 100 Mbps!

Isn't ist that the speed differences are mainly caused by different default settings for number of frequency points?

What use is a 1 Hz step size? I can see the TDR option being useful, but I'm less convinced about 1 Hz steps.

Dave

It is partly true for HP8753. Number of points are linear to amount of data that should be transferred to computer and more points costs more sweep time. 201 points/trace is often default setting and I do seldom need more. But there are other factors that affect total screen update rate such as how data is collected from programming view.
Slowest and most common is polling for data. VNA then starts a separate frequency sweep not related to internal screen data stream -> A/D -> adding several correction factors -> output buffer filling -> set GPIB ready signal, and finally can computer receive GPIB data. It takes more then 3 times longer time then fastest possible method that I know about and it is all about internal VNA delay.
Final amount of transferred data is relative small so that goes rather quick either it is GPIB or USB.
If it only is for saving data or printing, is that delay less important but in my case, it is like trying to fine tune frequency knob of a FM radio, and have to wait a second to know result. Tuning setting will sway back and forth.

If buying a new VNA will update speed on computer screen be an important factor for me, and if cost is reduced for the VNA, if it not includes thing that already exist on the ordinary computer, such as screen or keyboard, that is an advantage.
Also have some other USB boxes: signal generator, Aaronia 9 Ghz spectrum analyzer, USB RF attenuator. It is very practical to be able to control them all from one computer. With some scripting is it easy to switch between complex measurement scenarios, which for me is a real time-saver.
Drkirby have perhaps other priority's, but I had hoped that Copper Mountain VNA's was a combination of in every aspect good performance at cheap cost, that combination is often to good to be completely true, I know, but I never learn.
Best economy is perhaps to buy another HP8753 at Ebay to be prepared with spare parts when current VNA dies.

For a low-end solution, I thought this was quite interesting.

http://miniradiosolutions.com/minivna-pro

though I note there are no specifications of accuracy on it.

I have an application where I'd like to measure RF power at a connector in the air. But running a cable down to the ground will not be possible. Hence sticking a VNA up in the air, with Bluetooth communication to a laptop on the ground, is quite attractive.

I've made a dealer an offer on an HP 8720D. We shall see how that works out.


Dave

Yes, you can slow down the VNA by setting more frequency points or narrower IF bandwidth (for more dynamic range).

I've used the 8753C + S-parameter testset with doubler (-6GHz) when it was released. Something that I often missed was the fourth receiver channel that would have been required to use TRL calibration and other advanced calibration methods. Having SOLT calibration only was very limiting for SMD component measurement, where TRL would have been much more accurate. From what I have seen, the newer models such as 8753ES seem to have TRL* (TRL star) as a derivative which works with three receiver channels.

For the 8720D, there is an option 400 to have four (instead of three) receiver channels for true TRL cal capability.

I might have mispoken about 1Hz step size, maybe it is 1 khz? But there is a reduced frequency step size option that is VERY useful if you ever want to measure a crystal or saw resonator, or a crystal filter. Otherwise, you step right over the resonance point.

- - - Updated - - -

Gee, that is interesting to know. I thought all 8720D's could do TRL calibration--a very useful tool for modern tiny circuits!

I would risk buying an hp from ebay with even lower budget
i searched the marked few months ago for my friend. he also needed a vna for antenna measurements. anritsu had a 20ghz handheld sa+vna model with of course lan+gpib interface. you can check that and use with a pc monitor.
dynamic range was a bit low, i cant remember the numbers but worse than RS and agilent. bu the price was half. around 20k usd incl tax.

- - - Updated - - -

http://www.anritsu.com/en-GB/Products-Solutions/Test-Measurement/RF-Microwave/Vector-Network-Analyzers-Handheld/index.aspx
thats the link if i am not mistaken.
they also have 6ghz model

economy SA, NA or Sig Gen vendors exist, but i could not find any VNA, so please let us know the developments.
i even found a company in lithuania, probably old russian vendor. but their frequencies were too high.

Fortunately, I only borrowed the Demo unit for 6 mos (!) and the Iridium project was cancelled so the savings in purchase offset of development travel expenses for 3 mos in Cali and West Palm Beach, when our sister plant C-MAC there was setting up to build MOT's chip for their SatPhone. So it helped pay my billing of S70K for all the work I did.
Here I needed 0.01dB resolution and 0.1dB accuracy for measuring the chip switch loss in the LAN/PA chip. But it was extremely fast in performing bi-directional S parameters, 3rd order intercepts, IM and band pass/reject characteristics in sharp MLM ceramic filters that MOT designed into the chip in Phoenix. I was pretty experienced in SA/VNA's so I opted for a 1 day calibration training support at Cali Anritsu HQ, while I was stationed in Winnipeg , Canada and their Field Engineer in Fla was great.
Canadian HQ support was under-staffed at the time, but I was not dependant on them except for their over-generosity in the demo unit. :)

ALthough for what you need, it was overkill, as you need is portability, sensitivity, 1 way active ports with good directivity and plotting functions. So 10K is reasonable and anything less is on low budget.

From my personal experience on Rician Channel fading, which dominates fringe WiFi error rates and ability to obtain the max bit rate at low levels< −75 dBm you want to make sure your side lobes are well suppressed so that Rician scatter noise creates as few directional dropouts as possible. Some people don't appreciate the significance of loss of mobility on antenna with high directivity but for static tuned direction, the gain is significant for distance.