US$200 VNA kickstarter launched

[xaxaxa]

EDIT: The kickstarter has now been launched:
https://www.kickstarter.com/projects/1759352588/xavna-a-full-featured-low-cost-two-port-vna

--------------------------------
I've been working on a very low cost VNA project for some time and was thinking about doing a kickstarter; see original thread:
https://www.eevblog.com/forum/rf-microwave/low-cost-vector-network-analyzer/

I've already adapted the design to a 4 layer board and eliminated all the off board wiring. I also ran through DFM checks and have gotten quotes from the PCBA fab (elecrow) + component suppliers. Most of what's left is to prototype and verify the 4 layer design and work out logistics. This is for the 140MHz to 2.5GHz T/R version; all the other versions will require more R&D effort.

A few people have mentioned that HF support and full two port support would be useful, but I'm curious about how important people find these features to be vs the price, so I decided to post this poll.

A T/R VNA can measure a device in one direction at a time: reflection coefficient on port 1 and through coefficient from port 1 to 2; to measure the other direction (port 2 reflection and port 2->1 through) you need to physically reverse the DUT (swap ports 1 and 2). A full two port VNA has directional couplers on both ports and can measure both directions without having to reverse the DUT. The main difficulty with implementing a full two port VNA is that the board will need to be bigger to accommodate two directional couplers, as well as keeping good isolation. PCBs larger than 10cm x 10cm get very expensive.

Note: the upper frequency limit of 2.5GHz can be increased to 4.4GHz with slight additional cost; if this is useful please comment.

My original idea with this project is to commoditize the VNA, since everything currently on the market are expensive and proprietary; since this project is entirely open source and all designs (schematics, pcb layout, simulation files, fpga logic, software) are published, eventually people of aliexpress could clone it and maybe even offer it at a lower price. I was thinking there should be a market for this if it's inexpensive enough, since many people still use antenna analyzers and a VNA can give you much more information.

[Howardlong]

I think you’ve identified the three broad use cases, and three different markets: that of the non-ham Wifi antenna tweaker, the ham antenna tweaker, and the RF circuit builder.

I would think it would be reasonable that your average ham antenna tweaker would also want to go down to 3.5 if not 1.8MHz though.

I have an old Anritsu Sitemaster VNA that’s single port but because it’s battery powered and handheld, it’s very useful for antenna tuning and setup outside in he field, coverage is 25MHz to 3GHz.

On the bench I have an HP 8753A, 300kHz to 3GHz, which, with an accompanying TR test set, will do two port as well as one port measurements. For practical antenna measurements, except for small aperture microwave antennas it’s less than ideal because it’s hardly portable!

In my experience, don’t underestimate the size of the ham market. If you have a unique product at a reasonable price, they will come, but without HF capability it will rather limit that market. Word of mouth is strong. The non-ham market is undoubtedly much smaller. You need to make sure you extol the key differentiators between your product and any competitors, as there are already products like miniVNA and PocketVNA out there which will fragment your market.

[trevwhite]

If this would allow me to tweak antenna in the range of 433Mhz, 868Mhz, 915Mhz and 2.4Ghz then I would very much be interested.

[VK5RC]

+1 re the Amateur radio market and including HF, I would add that the software for these devices is often the deal maker (or not).
There are several others out there - Array Solutions AIM UHF , N2PKs efforts  and miniVNA to name a few.

[xaxaxa]

Thanks for all the suggestions.

The kickstarter is now live:
https://www.kickstarter.com/projects/1759352588/xavna-a-full-featured-low-cost-two-port-vna

Taking into account the poll results, I introduced a full two port version; HF support will be introduced if the stretch goal is met.

[TK]

Are you planning on adding the capability of setting your own calibration kit parameters that can be saved/recalled?

[xaxaxa]

Are you planning on adding the capability of setting your own calibration kit parameters that can be saved/recalled?

Yes; the software is under heavy development and you can expect all the standard features of a professional VNA to be implemented.

[usagi]

keep in mind you're competing with existing products like these:

http://miniradiosolutions.com/54-2/
https://sdr-kits.net/index.php?route=web/pages&page_id=68_68

[xaxaxa]

keep in mind you're competing with existing products like these:

http://miniradiosolutions.com/54-2/
https://sdr-kits.net/index.php?route=web/pages&page_id=68_68

I've seen those ones; the existing low cost VNAs are much more expensive and have accuracy and temperature drift problems;
see last page of: http://www.gsm-modem.de/M2M/wp-content/uploads/2014/04/IoT-M2M-Cookbook-Harald_excerpt.pdf
https://www.eevblog.com/forum/rf-microwave/pocketvna-any-idea-what-its-like/msg1171730/#msg1171730

[usagi]

i guess the only thing now is to decide who your primary target market is.

ham market will be considerably different than typical RF builder. ham guys dont care much for two port. and no <10mhz from the beginning (no stretch) will limit ham interest. also, most hams rarely care about >500mhz.

[pigrew]

One advantage of the full two port VNA is that the full twelve-term error model can be used. With only T/R, my understanding is that the cables cannot be fully de-embedded.

I've mostly used VNAs for characterizing amplifiers.

[xaxaxa]

i guess the only thing now is to decide who your primary target market is.

ham market will be considerably different than typical RF builder. ham guys dont care much for two port. and no <10mhz from the beginning (no stretch) will limit ham interest. also, most hams rarely care about >500mhz.

The intended market is RF engineers using the ISM bands, i.e. 433MHz, 863MHz, 915MHz, 2450MHz, and as a plus it can also cover UHF ham bands plus 144MHz;

Because of the way the LO synthesizer is fed into the mixers it requires a lot more extra circuitry to insert a switch and prescaler in there. I decided that adding full two-port support is enough feature creep for now and to limit the scope to reduce risks.

I could replace the synthesizer with the adf4351 or max2871 and cover down to about 30MHz, but not sure how useful that is.

[rfeecs]

One advantage of the full two port VNA is that the full twelve-term error model can be used. With only T/R, my understanding is that the cables cannot be fully de-embedded.

I've mostly used VNAs for characterizing amplifiers.

You can use the full twelve-term error model with T/R.  Requires two measurements, one in each direction.
Some calibration methods like TRL require the full two port to have 4 detectors to be fully accurate:

"Applying Error Correction to Vector Network Analyzer Measurements"
http://literature.cdn.keysight.com/litweb/pdf/5965-7709E.pdf

[Bud]

i guess the only thing now is to decide who your primary target market is.

The intended market is RF engineers...

RF engineeres are too busy to toy with arduino-style kits. To them you have to offer a finished solution.
Also, "calibration kit requires soldering" does not add credibility to the product.

[xaxaxa]

RF engineeres are too busy to toy with arduino-style kits. To them you have to offer a finished solution.
Also, "calibration kit requires soldering" does not add credibility to the product.

It isn't a kit as the board is ready to use;

The supplied calibration kit is something like this:

(plus the thru reference)

The "requires soldering" just means adding a blob of solder for the s.c. reference and soldering two resistors (provided) to the load reference.
I will update the kickstarter to clarify this.

The provided sma connectors will be genuine ones from digikey. It's mainly just to save you from having to order it yourself and paying for shipping again.

[TK]

Mouser has some stock male SMA termination, open and short plugs.  They are male, not female like your homemade kit, but you can add a female-female SMA connector and measure the new parameters and use them as calibration kit. 

Short: http://www.mouser.com/ProductDetail/Amphenol/132331/?qs=%2fha2pyFadui%252bsOUeTj9xiXC2xh5fwe1gjFO04t3RQW9wolrRJhIfTw%3d%3d
Open: http://www.mouser.com/ProductDetail/Amphenol-RF/202112/?qs=sGAEpiMZZMvlX3nhDDO4ANxAIWTV9VDXYOGp73w%2fDZI%3d
Load: http://www.mouser.com/ProductDetail/Amphenol-RF/132360/?qs=sGAEpiMZZMvkloRJ4SuqXuLVgwQXDW1k58OMTbQUIq8%3d
Adapter: http://www.mouser.com/ProductDetail/Amphenol-RF/132169/?qs=sGAEpiMZZMs022Iw%2foIyC4%252bc6P8fzLVU

Open seems to short when closed too tight, or maybe it is just the piece I received.  I just avoid screwing it too much.

If you know how to browse mouser, you might be able to find the equivalent female standards.

[xaxaxa]

Mouser has some stock male SMA termination, open and short plugs.  They are male, not female like your homemade kit, but you can add a female-female SMA connector and measure the new parameters and use them as calibration kit. 

Short: http://www.mouser.com/ProductDetail/Amphenol/132331/?qs=%2fha2pyFadui%252bsOUeTj9xiXC2xh5fwe1gjFO04t3RQW9wolrRJhIfTw%3d%3d
Open: http://www.mouser.com/ProductDetail/Amphenol-RF/202112/?qs=sGAEpiMZZMvlX3nhDDO4ANxAIWTV9VDXYOGp73w%2fDZI%3d
Load: http://www.mouser.com/ProductDetail/Amphenol-RF/132360/?qs=sGAEpiMZZMvkloRJ4SuqXuLVgwQXDW1k58OMTbQUIq8%3d
Adapter: http://www.mouser.com/ProductDetail/Amphenol-RF/132169/?qs=sGAEpiMZZMs022Iw%2foIyC4%252bc6P8fzLVU

Open seems to short when closed too tight, or maybe it is just the piece I received.  I just avoid screwing it too much.

If you know how to browse mouser, you might be able to find the equivalent female standards.

I checked all the suppliers and haven't found any sma female terminations that have specified return loss/swr and is also reasonably priced;
using an adapter is possible but the S parameters could change over time because of the extra connector interface.
The homemade termination can achieve return loss >30dB up to 3GHz (according to graphs on hforsten's site).

Do you think it's worth it to switch to pre-made terminations? Most people who would buy a VNA sure have a soldering iron and don't mind soldering two resistors right? 

[Jay_Diddy_B]

Hi,

For the calibration standards you have to understand the concept of a reference plane.

The load at the reference impedance is easy, there should be no reflections, if everything is properly matched so the distance from the reference plane to the load doesn't matter.

The short will have a little inductance and this can be measured and it needs to be known. The reference plane for the short is fairly easy to establish. All the incident power should be reflected by the short.

The open is actually the most difficult standard. Because of fringing capacitance and the open is a short distance beyond the back of the open connector. The open will also reflect all the incident power.

The difference between the open and the short is phase. There should be a 180 degrees difference between the open and the short.

Distance / velocity = time

and time delay results in phase shift



One of the measures of how good the calibration standards is how much the phase, between the open and the short, differs from 180 degrees at the maximum frequency.

Jay_Diddy_B

[Jay_Diddy_B]

Hi,

I would check this link:

https://www.sdr-kits.net/index.php?route=product/product&product_id=50

These connectors have been characterized.

A m/m or f/f connector is used for the open. The open listed earlier, by TK, is really a dust cap, to keep dirt out of a connector.

Jay_Diddy_B

[xaxaxa]

Hi,

For the calibration standards you have to understand the concept of a reference plane.

The load at the reference impedance is easy, there should be no reflections, if everything is properly matched so the distance from the reference plane to the load doesn't matter.

The short will have a little inductance and this can be measured and it needs to be known. The reference plane for the short is fairly easy to establish. All the incident power should be reflected by the short.

The open is actually the most difficult standard. Because of fringing capacitance and the open is a short distance beyond the back of the open connector. The open will also reflect all the incident power.

The difference between the open and the short is phase. There should be a 180 degrees difference between the open and the short.

Distance / velocity = time

and time delay results in phase shift



One of the measures of how good the calibration standards is how much the phase, between the open and the short, differs from 180 degrees at the maximum frequency.

Jay_Diddy_B

Yes that is the other issue with using an adapter on the cal references; you'd need to measure the combined S parameters of the termination + adapter (using another vna) and input it into the calibration algorithm.

[TK]

Do you think it's worth it to switch to pre-made terminations? Most people who would buy a VNA sure have a soldering iron and don't mind soldering two resistors right? 

I think the impedance of the connector can change if heated, so even when you ship the same connectors, each user might end up with different set of calibration kit and make the standard kit parameters that you include with your instrument wrong.  They might need to enter their own parameters.

[xaxaxa]

I think the impedance of the connector can change if heated, so even when you ship the same connectors, each user might end up with different set of calibration kit and make the standard kit parameters that you include with your instrument wrong.  They might need to enter their own parameters.

It's all relative; you can make fairly good measurements with no cal kit parameters, just pretending the references are ideal; in fact all the graphs shown on my page are done this way since I don't have access to an expensive VNA or calibration kit.

The dielectric characteristics may vary after heating but that's likely less significant than the fringing field issue which can be corrected identically for all units.

I'm going to offer the option of pre-soldered cal kits (at no extra charge); just leave a remark when pledging. [updated kickstarter]

[exe]

Stupid question, why it cannot go down to, e.g., 100KHz? I'm asking because I'd like to characterize capacitors and filters, but even 10MHz is already way to much.

It looks like there are two types of VNA: designed low frequencies (say, below 0.5GHz) , and for RF work (several GHz).

[xaxaxa]

Stupid question, why it cannot go down to, e.g., 100KHz? I'm asking because I'd like to characterize capacitors and filters, but even 10MHz is already way to much.

It looks like there are two types of VNA: designed low frequencies (say, below 0.5GHz) , and for RF work (several GHz).

There are generally two architectures used in VNAs, the single conversion type where the signal gets downconverted to low frequency IF and get digitized, and the spectrum analyzer style architecture where the signal gets upconverted first (above the maximum input frequency). The former can't receive signals down to anywhere near its IF frequency because of the filtering, and the latter requires much more circuitry and much more care to preserve signal integrity.

Both the hforsten VNA and mine are single conversion because it's much easier to get good accuracy this way. The miniVNA seems to be the upconverting type and can cover down to near DC. It's mainly a tradeoff between accuracy, frequency coverage, and cost.

[exe]

The former can't receive signals down to anywhere near its IF frequency because of the filtering

Is it possible to bypass the filter?

[xaxaxa]

Is it possible to bypass the filter?

That's only half the problem; you also need a synthesizer that can reach near DC, and those are nearly nonexistent. The max2871 has one of the widest frequency coverage and even then it only goes down to 23.5MHz.

It is possible to use a prescaler or DDS solution; if i use a lower IF as well i might be able to do <1MHz. That is what I might consider for the next version once i get the main product delivered.

[xaxaxa]

Looks like my kickstarter has flatlined;

If anyone could help me spread the word or give me some suggestions about where else to advertise it I would really appreciate it!

[TheSteve]

It is probably too late but one option is to send a prototype to someone who produces youtube content. Dave is an option, or Shahriar from the Signalpath (if they are interested of course). They are obviously very busy though and don't produce instant content. The biggest thing you have going for yourself is the price. I have two network analyzers already or I'd likely be interested. Either way I am routing for you as you're local to me.

[Jay_Diddy_B]

Hi,

You can see a chart of the progress on this website:

https://www.backerkit.com/projects/1759352588/xavna-a-full-featured-low-cost-two-port-vna

This is their prediction at the moment:



I would keep the screenshots and the progress reports coming. I think you need a few more points on some of the measurements to get smoother curves.

Regards,

Jay_Diddy_B

[xaxaxa]

Hi,

You can see a chart of the progress on this website:

https://www.backerkit.com/projects/1759352588/xavna-a-full-featured-low-cost-two-port-vna

This is their prediction at the moment:

I think that graph might be a bit overly optimistic 
Currently it's already dropped off to less than one backer per day which means I need a wider distribution channel if it is to reach goal.
Any suggestions?

[trevwhite]

Hi,

You can see a chart of the progress on this website:

https://www.backerkit.com/projects/1759352588/xavna-a-full-featured-low-cost-two-port-vna

This is their prediction at the moment:

I think that graph might be a bit overly optimistic 
Currently it's already dropped off to less than one backer per day which means I need a wider distribution channel if it is to reach goal.
Any suggestions?

You could tag it as an Internet of Things tool for everyone who wants to tune their antenna? Get it mentioned in hackaday? See if Elektor mag would do a comment on it?

Can you do a video of you tuning an actual Wi-Fi board antenna or a Bluetooth board antenna? That might tag some interest/generate traffic. It would show a really useful example of how people can use it and experiment with it.





Thanks

Trev

[phil]

+1 on the video.

You could try and submit it to https://www.rtl-sdr.com. All this cheap(ish) SDR stuff is very popular with the maker/hacker folks. There are quite a few out there building their own front ends or antennas and looking for an affordable VNA.

[usagi]

+1 on the video.

You could try and submit it to https://www.rtl-sdr.com. All this cheap(ish) SDR stuff is very popular with the maker/hacker folks. There are quite a few out there building their own front ends or antennas and looking for an affordable VNA.

lack of <140mhz is going to dampen interest in this device by the sdr crowd.

[TK]

In my opinion, Antenna tuning can be done using a spectrum analyzer with noise generator or tracking generator and it is top priority for most HAM and SDR users.  VNA is a desired equipment but not top priority.  So, they will first buy a nice spectrum analyzer, then more HAM equipment, and lastly a VNA.   You need to show them they NEED one so it makes it top priority.

Your project is open source and if you make it attractive for the the user community to hack and improve it, you will have a differentiation.  But you need to be prepared for the clones from China if you are successful.

You can take a look at HACKRF (Great Scott Gadgets) and how they promote the item online.  There are lots of clones now selling on eBay as it is also open source.

[homestea]

I think if it had a nice enclosure (sheetmetal, molded, etc) to make it look more like a finished product it would have way more appeal.  I know that this will add cost but you could make it a higher cost tier backing for including the enclosure.   If you need help I’m a Mechanical Engineer with extensive experience with designing enclosures.  I’d be willing to help in return for a free kit or something like that. 


Sent from my iPhone using Tapatalk

[trevwhite]

I think if it had a nice enclosure (sheetmetal, molded, etc) to make it look more like a finished product it would have way more appeal.  I know that this will add cost but you could make it a higher cost tier backing for including the enclosure.   If you need help I’m a Mechanical Engineer with extensive experience with designing enclosures.  I’d be willing to help in return for a free kit or something like that. 


Sent from my iPhone using Tapatalk

I think that Is a good idea and a decent offer if the enclosure can be made cheap enough. Might be cheaper to drill an off the shelf enclosure though  if numbers are low.

Maybe something could be 3d printed?

Thanks

Trev

[xaxaxa]

I've submitted the project to hackaday and rtl-sdr blog;

@homestea is helping me with the enclosure, and I plan to introduce a version with a proper metal enclosure. The default version will most likely have an acrylic case too, but I didn't promise that on the campaign just in case it doesn't fit in the budget. It is mentioned on the kickstarter page.

The hardware and software are both open source and community involvement is welcome; especially on the fpga hdl part there are a lot of possibilities (maybe a sdr mod?)

The video is a good idea; will do that when I have a good lab environment (i'm currently out of town).

[rhb]

A VNA in a plastic box is an oxymoron.  You should design the board around a suitable standard enclosure taking into account cooling and shielding.  I very much doubt that someone who would be interested in owning a VNA would be willing to house it in a plastic box.  They certainly *ought* to know better.

I bought a LimeSDR and am still struggling with the problem of packaging it properly.  I'd be much more interested in a more capable unit than a VNWA 3E in a well designed enclosure at the same price point than a $200 board that will require considerable thought and effort to package properly.  You can always sell the board without the enclosure, but at least offer a suitable extruded enclosure with punched and screened end panels.

The LimeSDR enclosure offerings were ridiculous, the acrylic was stupid and the metal was absurdly expensive and not suited for general experimentation.

I think your $200 price point is too ambitious.  For a prospective buyer the difference between $200 and $400 is less important than getting a usable product.  It needs to become a regular product which means you need to factor in a reasonable profit margin for whomever is going to sell them.  If it's an orphan product there won't be any bug fixes.

[xaxaxa]

As mentioned, there will be a version with a metal enclosure. Bare board with standoffs and cover is also perfectly usable since the product will come in a cardboard box that you can easily carry it around in. The terasic fpga dev boards are a good example of this.

Not everyone can afford or justify $400+ and to some people (like me) it is a huge difference. Plus, the main selling point is not the price but the much improved accuracy and temperature drift compared to existing products. I'm not going to quote specific numbers here but you can be assured that my profit margin is not razor thin.

A VNA in a plastic box is an oxymoron.  You should design the board around a suitable standard enclosure taking into account cooling and shielding.  I very much doubt that someone who would be interested in owning a VNA would be willing to house it in a plastic box.  They certainly *ought* to know better.

I bought a LimeSDR and am still struggling with the problem of packaging it properly.  I'd be much more interested in a more capable unit than a VNWA 3E in a well designed enclosure at the same price point than a $200 board that will require considerable thought and effort to package properly.  You can always sell the board without the enclosure, but at least offer a suitable extruded enclosure with punched and screened end panels.

The LimeSDR enclosure offerings were ridiculous, the acrylic was stupid and the metal was absurdly expensive and not suited for general experimentation.

I think your $200 price point is too ambitious.  For a prospective buyer the difference between $200 and $400 is less important than getting a usable product.  It needs to become a regular product which means you need to factor in a reasonable profit margin for whomever is going to sell them.  If it's an orphan product there won't be any bug fixes.

[xaxaxa]

mini-update: iteration 2 prototype ready to be sent to the fab. Attached is the top view of the PCB.

[HalFET]

Why would a VNA in a plastic box necessarily be worse? If you manage to shield the critical signal paths within the inner layers of the PCB I don't see why you couldn't get an acceptable performance level? Don't expect a $200 unit to perform like a $2000 one. But realistically speaking, do you need it to? There's no need to beat Keysight's FieldFox units I think. But actually, another question: Did you check prices at any other board house? Most of the EU and US outfits charge quite a bit extra, we've gotten pretty far by just going to the Chinese board houses directly. Might be worth shopping around once you got the design finished.

[xaxaxa]

Why would a VNA in a plastic box necessarily be worse? If you manage to shield the critical signal paths within the inner layers of the PCB I don't see why you couldn't get an acceptable performance level? Don't expect a $200 unit to perform like a $2000 one. But realistically speaking, do you need it to? There's no need to beat Keysight's FieldFox units I think. But actually, another question: Did you check prices at any other board house? Most of the EU and US outfits charge quite a bit extra, we've gotten pretty far by just going to the Chinese board houses directly. Might be worth shopping around once you got the design finished.

All the sensitive signals are routed as a stripline between two ground planes; the directional coupler balun will have a shield can over it as well.

A vna isn't particularly sensitive to outside interference because the receivers are extremely narrowband (with the exception of those cheap analog phase comparator based ones) and the signal levels involved are fairly high compared to e.g. a spectrum analyzer; in my tests you can even transmit 100mW into the S11 measurement port and not affect measurements much unless your signal coincides with one of the measurement frequency points.

I've looked at all the major pcb fabs and will be using a fairly reputable one from china to ensure quality.

[HalFET]

Sounds very reasonable to me. Additionally if people are so worried about noise they can always line the plastic case with copper or aluminium tape 

Mhhh, Elecrow isn't very cheap in my experience though, or did you switch over to someone different?

[xaxaxa]

Sounds very reasonable to me. Additionally if people are so worried about noise they can always line the plastic case with copper or aluminium tape 

Mhhh, Elecrow isn't very cheap in my experience though, or did you switch over to someone different?

I have a quotation from elecrow but I'm looking into a few other manufacturers as well; I might actually make some test boards to see how good their impedance control is, and then decide.

[krazatchu]

I'm tempted to pull the trigger on the xaVNA two port but the lower end wouldn't cover the 13MHz RFID design I sometimes do.

Any chance we will see an xaVNA two port with a lower bottom range?

Michael

[xaxaxa]

I'm tempted to pull the trigger on the xaVNA two port but the lower end wouldn't cover the 13MHz RFID design I sometimes do.

Any chance we will see an xaVNA two port with a lower bottom range?

Michael

In the far future (several months) I'm planning to develop a "lite" version that covers LF to 150MHz using DDS, since I personally also need to use it for some stuff around 100MHz. It will most likely be T/R or one-port though, and cost less, since the main users of this frequency range are hams and it makes the hardware significantly more complicated to do full two port support. I'll put clock input/output ports to make it possible to lock two of them together to form a two port VNA.