Checkin' out dummy loads with a VNA...

Tell your VNA you want to measure S11; it won't even touch the second port (feel free to hook it up to a high-gain antenna and point it at the nearest cell phone tower in your neighborhood...)

Unless your VNA is 20+ years old, it'll be happy to display the S11 measurement in the form of a Smith chart, real/imaginary impedances, VSWR, etc...

Hello Paul,

A quick and dirty one would be to use a resistor from VNA-out to dummy load, then another from dummy load to VNA-in. 50 ohms or less. With next to nothing reactive, of course. All nicely RF fashioned, chip resistors maybe. It's not going to be matched but you can have the analyzer do the math to calculate out the dummy load. Mind the power limits.

Sometimes I do this with an HP resistive splitter when the impedance analyzer is down (like it is right now, again....%*&#!!).

Regards, Joerg

Hi Paul,

I don't mean to pick on you, but... you really should become more familiar with VNA measurements in general. You should be able to find documentation on the web; try the Agilent site for example, and look for VNA ap notes. I downloaded some recently, so I'm pretty sure they are there. A VNA is a wonderful piece of test equipment, but it won't give you answers you can trust if you don't understand it. I would trust what they tell you (or similar from R&S or other reputable manufacturer) more than I would trust replies here to be not only accurate but complete.

For example, you should understand that your dummy load could be considered a two-port device where S21 and S12 will be very low--hopefully vanishingly close to zero. If you connect it to port 1, you should get a valid indication of S11. You should expect S22 to be unity or very close to it, if that port is left open or shorted. (But why would you even bother to ask for S22, or S21, or S12?)

But there's another problem here. HOW do you calibrate your VNA?? You ideally should have a precision short, a precision open, and a precision load--best if you have calibration data for the loads. If you measure your load, how will you know whether an imperfect S11 is due to the load, or to the VNA? Of course, it will be to some extent both, but it would be nice to know that the VNA is accurate to within some tolerance.

Cheers, Tom

Okay guys, thanks. Yes, as you've surmised, I'm only just beginning to get to grips with this complex piece of kit and there is *much* to learn indeed. In my defence, I have to say although I purchased this thing 5 months ago, I've only now been able to assemble the ancilliary cables, loads, splitters and whatnot to be able to check it out properly. All I could do back in December was to switch it on and check for signs of life. Now, however, things start to get rather more complicated ( but hopefully stimulating, too).

Hello Paul,

Just curious, which VNA are you using?

Tentec has a really nice one. Only goes to 100MHz but for around $700 it seems like a good deal. Some of my clients could have used these but unfortunately this VNA does not seem to have an industry-standard calibration scheme. Wish they did, then I could probably get rid of my heavy 4191 in the lab.

Regards, Joerg

It's an HP 8754A (4 to 1300Mhz) and luckily old enough to be reasonably serviceable and repairable! This is just as well since there is a fault with the rectangular trace, which just appears to 'flatline' with nothing visible other than the sweep + markers. So no nice bod plots of filters, I'm afraid. However, the polar and phase displays seem absolutely fine. If I can just get that one fault fixed, it'll be a great piece of test gear to have around. I've never worked on anything this complicated before, though, so it'll be something of a challenge!

Hello Paul,

That looks like a problem in the digital section. Just hope that none of the EPROMs suffered a memory loss. I am not sure whether they'd report a checksum error in the 8754. I had found one that had become unseated on another analyzer and the instrument produced only garbage output but reported nothing on self-test. That disappointed me a bit. The only reason I found the cause was that I opened it and looked at it long enough.

Complicated they are. However, while on the troubleshooting trail in my HP4191 I sometimes came upon a large conglomeration of discretes and banged my head, asking myself "Why didn't they do that with a uA733?". And on the directional bridge they seem to have left about 10dB of dynamic range on the table for the test channel versus the ref channel. I'd have put a wee amp before the sampler.

Regards, Joerg

Thanks, Joerg. I think this item pre-dates a lot of this digital stuff; I doubt there's a single eprom in it - and no custom chips, either! It was built in about 1980 and doesn't go through any self-testing routines. One has to carry out manual tests oneself. This wonderful simplicity should hopefully enable me to keep it running indefinitely, once I can get it fixed. Serviceability is one of my main criteria when choosing an item of test equipment.

Hello Paul,

The HP4191A is from that era as well. However, it is one heck of a complicated machine with a computer board as large as a family-size pizza. Lots of PROMs and stuff :-(

Regards, Joerg

Hi Joerg, Well I'm sorry to hear that. As an analogue guy, I don't much care for PROMS and stuff. I can fix up anything analogue in good time, given a schematic and a decent range of test equipment, but am off my territory with digital stuff. But I'm still not sure what the problem is here. Are you saying spares for the digital parts are hard to obtain or rather that the encoding on those chips may be hard to reprogram?

Hello Paul,

Not sure yet what is broken. The analog sampling stuff and all seems to work but is reads out garbage. The PROMs are really old style. They look posh with gold plating and all but I guess that non-programmed spares would be next to impossible to find. But again, maybe they are still ok. Hopefully...

Regards, Joerg