Inexpensive cal standards

[Dan Moos]

I have hacked my Siglent analyzer to a VNA with all the options (thanks to all who figured out the hack!)

The only calibration standards I have are the cheap ones that came with my NanoVNA.  Assuming just hobbiest needs, are these sufficient? What limitations do they have?

[tszaboo]

They might not be 50 Ohm at all the frequency. It's also possible their delay is not the same for the 3 parts.
But probably the biggest issue with them is the quality of the connector. SMA (especially cheap quality) has a limited number of connections before the threads are used up, and you can run though it with just a few measurements.

[Dan Moos]

Actually, that brings me another question. Are ANY loads 50 ohms at all frequencies?

I'm just a hobbiest, and even though the analyzer goes up to 3.2 Ghz, I doubt my needs go beyond 100s of Mhz. Surely they are ok for  that? Proper standards from Siglent are expensive, and if they are overkill for my use, I don't want to waste my money.

[tszaboo]

High end SMA calibration kits are good till 9GHz. Good ones are usually go to 6GHz.
Those cheap ones are great value, and the frequency you are interested in, I don't expect any issues.

[joeqsmith]

Actually, that brings me another question. Are ANY loads 50 ohms at all frequencies?

Are ANY resistors 50 ohms at ANY frequency?  The correct answer, everything has errors.

[EggertEnjoyer123]

I compared the loads I got from the calibration kit supplied with the LiteVNA against commercial 50 ohm loads, and I found no difference up to 6 GHz.

If you're worried a 50 ohm load from Amphenol is available for around $4. (https://www.digikey.com/en/products/detail/amphenol-rf/132360/1011928). You can buy a few of them and choose the one which measures closest to 50 ohms. Of course this is no match for an actual calibration kit but it's good enough for you.

[Berni]

VNA Cal standards can be made DIY too.

For example i often ended up actually wanting a opposite gender cal kit. So what i did  is take some trough hole PCB mount SMA connectors. Filled one of them into a solid block of solder on the back for a short, cut away the pin for an open, put two precise 100 Ohm SMD resistors in parallel in opposing directions from the center for a 50 Ohm load.

Gave them a measure on a real HP VNA boatanchor and they looked pretty much the same against a real proper cal kit. To be fair it is only up to 4GHz since that's how far that VNA goes. But certainly good enough for a little NanoVNA. 4GHz is not that terribly high as long as the connectors and cables are half decent.

The number of insertion cycles is just the guaranteed number, they work just fine well past it. If they do wear out, toss them away and make new ones since it only costs you 3 connectors and a bit of work. While the VNA should always be using a sacrificial cable that wears out instead of the VNA connector itself.

[tszaboo]

I compared the loads I got from the calibration kit supplied with the LiteVNA against commercial 50 ohm loads, and I found no difference up to 6 GHz.

If you're worried a 50 ohm load from Amphenol is available for around $4. (https://www.digikey.com/en/products/detail/amphenol-rf/132360/1011928). You can buy a few of them and choose the one which measures closest to 50 ohms. Of course this is no match for an actual calibration kit but it's good enough for you.

OK, if you mix one terminator with another kit's open and short, how do you make sure the electrical length  is the same?

[joeqsmith]

I doubt my needs go beyond 100s of Mhz

I wonder how many hundreds of MHz. 

I compared the loads I got from the calibration kit supplied with the LiteVNA against commercial 50 ohm loads, and I found no difference up to 6 GHz.

If you're worried a 50 ohm load from Amphenol is available for around $4. (https://www.digikey.com/en/products/detail/amphenol-rf/132360/1011928). You can buy a few of them and choose the one which measures closest to 50 ohms. Of course this is no match for an actual calibration kit but it's good enough for you.

OK, if you mix one terminator with another kit's open and short, how do you make sure the electrical length  is the same?

It's a terminator, not a short or open.  There is no reflection. 

[alan.bain]

Normally the cal kit data includes the electrical length of the offset short/open which can be different as well as a polynomial for fringing capacitance as a function of frequency for the open (and maybe a similar polynomial for inductance for the short - but certainly sub 3GHz the short data isn't so important). 
Clearly no electrical length is needed for the broadband load!

I have used Rosenberger SMA fittings as a make shift SMA cal kit along the lines of https://www.sdr-kits.net/calibrationkits (because my real cal kit is 3.5mm).

[joeqsmith]

Normally the cal kit data includes the electrical length ..

True.  I'll admit that I often ignore all of that and assume everything is perfect.  Even worse, many times I will not calibrate the VNA.  The PNA does a good enough job (same for my LiteVNA64) that I can tell if things are working without doing anything.   Calibration is pretty much the last thing I do.  Granted, not every VNA is going to be the same.  No way I could do this with my original NanoVNA.   

For the low cost VNAs, I think the only time I have used anything other than the ideal model (assuming all the standards are perfect) was my experiments with rectangular waveguides in the X-band.  Similar to coax,  I was using homemade standards for the offset & short.  Load and mismatch load were made by Narda.

[selcuk]

That calibration kit will be fine for most of the hobby tasks. I have liteVNA. There was a SMA short, open and load in the VNA kit. I bought two identical SMA 50 ohm 6GHz loads to run isolation calibration when needed. And a kit with N-type connectors to work with N type cables and equipment easily. These are not mandatory but help working with the VNA. I guess your Siglent analyzer has N type connectors, doesn't it?

[virtualparticles]

The only calibration standards I have are the cheap ones that came with my NanoVNA.  Assuming just hobbiest needs, are these sufficient? What limitations do they have?

The biggest issue will be the 50 ohm load. The cheap load standards might have worst case return loss of 15 dB. I measured 12 dB on on a real VNA. If the load is 15 dB then your uncertainty will be +/- 3.3 dB for 5 dB S11 measurements rising to infinity at 15. If you can invest in a good load, you'll make much better reflection measurements.

I wrote this a while back. You might find it handy.
https://coppermountaintech.com/introduction-to-the-metrology-of-vna-measurement/

[小太]

As other people have mentioned, the cheap kits will generally be fine up to a few hundred MHz.
In fact, a lot of the more expensive kits sacrifice low frequency (<500MHz) performance for better high frequency performance.

If you want something better than the cheapo kits but not too expensive, check out the SDR-Kits kits already mentioned earlier (US$20~150 depending on options).
Beyond that are the kits from Kirkby and Applied EM ($US600~700) which are "properly" characterised with calibration coefficients, but still SMA.
Beyond even that are the "real" 3.5mm calibration kits from the major test equipment vendors (>US$1k).

[cncjerry]

Lately there have been some great deals on minicircuits labs 50ohm terminators.  Search in mcl and load or terminator. I think they are a level above those you find in cheap cal kits.  I bought some recently for an average of about $6 per.  I think they are not knockoffs, but I would exercise caution as always.  I thought I had posted this here already, sorry if it is a dupe.

Jerry

[Wallace Gasiewicz]

I have a bagfull of H&S SMA Loads that are new.Test great to 3 GHz which is all that I can do.

[Marsupilami]

It's a fun project to make a T-Checker and try it for yourself.
https://scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_application/application_notes/1ez43/1ez43_0e.pdf
I vaguely remember last time I could even find software for that, which now I can not, but either way it's a single equation to implement. Pretty easy.

[joeqsmith]

Lately there have been some great deals on minicircuits labs 50ohm terminators. Search in mcl and load or terminator. I think they are a level above those you find in cheap cal kits. I bought some recently for an average of about $6 per.  I think they are not knockoffs, but I would exercise caution as always.  I thought I had posted this here already, sorry if it is a dupe.

Jerry

The specs are not that great.  One of the eight I have tested very good when compared against a set of Agilent standards.  You may have better luck.   

https://www.eevblog.com/forum/projects/20db-rf-attenuator-seeking-feedback-to-improve/msg3093968/#msg3093968

[switchabl]

The six you have measured here look pretty good to me:
https://www.eevblog.com/forum/projects/20db-rf-attenuator-seeking-feedback-to-improve/msg3093675/#msg3093675

Especially in the context of this thread (and considering they are <15$ list price). They are all extremely well behaved until at least 2-3 GHz, so if you select one solely on DC resistance, chances are it will be very good. Four of them are better than 40dB to 3 GHz which is already "proper" cal kit territory. If OP were to buy a budget Siglent-branded cal kit, it might actually be worse than that.

Contrast that with the one labelled "nanovna" (https://www.eevblog.com/forum/projects/20db-rf-attenuator-seeking-feedback-to-improve/msg3093968/#msg3093968) where you can immediately tell that the construction is not suitable for high frequencies. Even if you find one that is 50.00 ohms, it will never be good beyond 1 GHz because of the large parasitics.

Beyond 3 GHz the story is a little more complicated. That's where (quality) commodity loads tend to become less predictable and this seems to be borne out by your data. Another caveat is that your Agilent load is "only" specified to 38dB in that region. So uncertainties will be pretty large. If you are really serious about characterizing loads beyond 40 dB over the whole frequency range, I'm afraid you will need to look into TRL calibration or get your hands on an ECal.

Personally, I would be fine using any of the six loads to 6 GHz for non-critical work. 35 dB is still plenty good for a lot of things.

[joeqsmith]

The original NanoVNA was only useful to 300MHz and the supplied load was fine for that.  Data you mention started out 3X higher than that.   It's not very good for that.   

For the mini-circuits parts, I would just  go by the worse case numbers in their data sheets.  I was looking for something better and why I was sorting them. 

[switchabl]

For a commercial-grade termination, guaranteed >30dB @ 4GHz is almost as good as it gets. I have a bag of Huber&Suhner SMA terminations which I think are >32dB but they are significantly more expensive. For someone looking for a cheap "good enough" working standard, the MC seems easy to recommend. And if you buy 3, chances of getting one close to 40dB to ~3 GHz should be pretty good.

If you need guaranteed, tracable specs, then there's no way around getting a proper, characterized cal kit (and paying $$$ for that). But having some working standards, selected from inexpensive off-the-shelf parts, is a thing even in professional labs. Especially for design work or a quick pre-check, a dB or two usually doesn't matter and you generally want to avoid unnecessary mating cycles on the expensive stuff. Sometimes you can get away without calibration (and maybe automatic port extension) but sometimes you can't.

[joeqsmith]

If you need ...

For me, it's just a hobby so everything is a want, not a need.  I had borrowed a few sets of HP & Agilent metrology grade standards to characterize mine.  For the low cost VNAs, I don't bother.  Normally, I don't even bother to calibrate...   

[Tishers]

I have a couple of spec-an's and VNA's in the middling price ranges ($500-$5000 ea) and was scrimping on the connectors, cal standards and cabling.

One day I was browsing the most popular auction site and found some 3.5mm components and was curious. Looking at the seller I discovered it was an electronics recycling center about five miles from my house.

I called them up to see if I might be able to stop by and see what they had sitting around their warehouse. It was a big place and they had a bunch of minimum wage folks chopping off the gold leads, tabs, etc.. from electronic devices and sorting by metals and putting it all in bins. It made me sick to my stomach to see a stack of HP test equipment reduced to chopped up pieces to get after anything that was gold in color.

I wandered around in the darker recesses of this warehouse and found giant wooden crates filled with small electronic parts. They had a shipping address as coming from one of the major universities in Alabama. It looked like they had taken the entire tools, test equipment and parts boxes out of the RF labs and just sent them off for recycling. I scavenged through this giant wooden crate (four feet high and probably eight by eight feet inside. I had to climb on top of other equipment to get inside.)

I was not prepared that trip to be digging through the dirt and electronic flotsam that day but I kept at it for about two hours. I also found one of those black, conductive-plastic bins that we use to hold PCB's in manufacturing. I filled the bin up with about ten pounds of every imaginable RF component; little circulators, return loss bridges, more than a hundred SMA attenuators, biconical antennas in the microwave band and half-completed project boxes with SMA connections on each end and space for components to be soldered in the jig.

I took it all up to the front desk and the receptionist was prepared to count each part and sell them to me... I left the dirt and grub all over them and looked a complete, dirty mess when I walked in to her office. She was reluctant to even touch the box. I said "how about I give you $150 for the entire bin?" She immediately said YES (anything but to not go digging through that plastic bin of dirty stuff).

Taking it home I inventoried what I had; About 2/3 was still good. APC-7 connectors and adapters still in their plastic tubes, SMA and 3.5mm, attenuators, fixed and variable, delay lines.. I threaded them all together end on end and it was more than two feet long of just attenuators. It all just needed a good cleaning.

I still have my 'protected' cal standards for when I do real-work but I make use of the pieces-and-parts on a regular basis.
---
One thing I saw when I was leaving the building was a giant stack about ten feet tall of wall sections for Lindgren Faraday cages. I only had my car and by the time I got back there with my pickup truck a few days later they had stripped all of that down for the copper mesh and threw the rest away.

The place is now out of business; Like all electronics recycling centers they seem to follow a seasonal circus route and move to a new town.