Anyone (else) using a vector voltmeter?

[pdenisowski]

Just wondering: is anyone (else) here using a vector voltmeter these days (traditional analog or VVM functionality in a two port VNA)?  If so, what are you using it for?

In my experience, the biggest application (by far) is for navaids (ILS, VOR) and anything else that requires very precise cable lengths / phasing, but I'm curious what other people might be using them for

[shabaz]

By coincidence I wrote an article on hyperbolic navigation a couple of weeks ago! It's just some notes that I decided to write up because I found it interesting, and hope maybe others do too.

I have a few analog meters if I ever get around to experimenting a little. It would be interesting to recreate a simple experiment using a sig-gen.

[Roehrenonkel]

Hi pdenisowski,
 
yes i do (did) - and with the exact HP-model. :-))
The fixed probes are a pita....
Used it to compare two 10MHz-sources (GPSDO and Rubidium).

Best regards

[coppercone2]

it needs a mod. I am tired of this probe cable causing such a problem.

Its likely because its resistive probes and terminating it is problematic. I wonder if you can braze it into a normal connector with a tiny induction or resistance heater  to get the center pin on and then use it with a N connector by getting fine silver wire putting it into the center pin along with some flux. If you can spot heat it fast enough you should be able to solder to the fine resistance wire without a problem. I heard its hard to solder. Or to get a small brass pin with a hole drilled in it, that can be brazed to the fine center conductor with a minimum of wire, then solder that into the N connector.


This is also a problem with HV probes. They use some bizzaro ceramic + metal crimped thing for their resistance wire in the leads.

It sounds a little complicated but the discussion about how to crimp that to a metrology grade is also probably insane.



*************

I am almost certain a small pace resistweeze can solder the fine steel resistance wire from a dampened lead to a tiny brass rod that is drilled out with silver, so that it can be treated as a conventional wire. It just needs a long N connector with good strain relief because its going to be a little dodgy. It can make things glow red on high settings. Brass should be acceptable because it has a fair resistance and lower thermal conductivity which makes soldering to it easy. Then you would have a tiny brass nub on the end of that steel inner conductor that can be soldered into a normal connector with the minimum change of impedance. THen you could free your probes instead of having that god damn zerg structure in your lab



I feel like that resistance probe cable is like a massive nuisance to people that wanna do serious electronics in the high AC range. Then you don't need the last known in existence resistive wire crimper, found in some tektronix stock room, to make anything custom.



I don't have one, because the probes are hard wired to it. So you can't put it on a shelf when not in use. Its preposterous. I almost bought this thing 10 times over the last 10 years. 




Of course,  it would be more stable if you left it alone instead of butchering it to make it modular. 

it sounds like any loss of measuring capabilities would be a blow to the fine work its being used for



But there must be damaged units with ripped cables if its build like that. Maybe with a little bit of silver and panel work, its possible to fix them without trying to find new cables, especially if its frayed near the input.



My hunch is
-resistance wire is crimped into a little ceramic coaxial connector, then the outer shield is crimped over it. The inside cable might not be critical , if the long outside cable is all resistance, so it might be possible to replace with normal coaxial if its damaged with minimum effect on  the measurement. If you made the cable a little shorter, you might get a hair more resonance. It might be more then acceptable and possible to calibrate out.

Well that one looks like twisted pair, but it should still be possible to fix, even if its two resistance wires, by using a twinaxial BNC.

old resistance wire coaxial probe, another road block

[G0HZU]

Maybe you should have a look at the service manual to see what is inside the probes and the probe cables. It's a lot more complicated than a piece of coax made with resistance wire.

There are ground connections and also four other wires inside each probe cable. The probe is a sampling gate (similar to that used in a sampling oscilloscope) and there are fine sampling pulses fed up the cables to a diode based sampling gate (followed by a JFET) in the probe tip and the sampling pulses are about 300ps wide. There is a 20kHz IF signal fed back down the probe from the sampler at the tip of the probe. So there is a lot going on inside these probes.

[G0HZU]

The probes are a weak point in that they are easy to damage. The needle pin at the probe tip can break off in the same way that a scope probe tip pin can break. It is possible to unscrew the pin and screw another one on if you can find spares.

It is always wise to keep the BNC tip adaptors on the ends of the probes when storing or moving the VVM to another location. This is the time when it is most likely that the probe tips will be physically damaged.

The other limitation is that the probes can only deal with signals that are smaller than the amplitude of the sampler pulses fed up the probe cable. Otherwise, the input signal can upset the sampling gate action.

So that means you can realistically only use it for signals up to about 1.5 V peak.

There are dividers extensions that can be fitted to the probe tips and these divide by ten. This means larger amplitude signals can be measured. However, the accuracy and bandwidth is significantly compromised if you use the dividers.

In my experience, it's best to use the probe tips directly and only probe small amplitude signals with it. I first used one of these meters when I was a student and I used them some more in my first job. I also bought one about 11 years ago and recently bought another one that was made in the late 1980s and it is in very good condition.

[G0HZU]

If anyone decides they might like to buy one of these meters, there are a few caveats.

These meters date back to about 1966. My older meter was probably made in 1967. HP 8405A VVMs from the 1960s and the 1970s will probably be suffering from lots of age related problems. One of the most significant issues is that the meter dial transparencies begin to peel away from the white background of the meter. This causes the needle to jam against the peeling transparency. This issue also affects other HP gear from that era. My first meter developed this issue soon after I purchased it and I had to take the meter apart to fix the issue.

Some people stick new dials inside the meter. In my case, I made up a water based adhesive solution and used it to stick the dials back down. Not a nice job to have to do and it took me two attempts to get it all stuck down.

Also, the diodes in the probe sampler can fail through overload and the original diodes are long obsolete. There are suggested alternatives for these diodes, but I don't think they will work as well as the originals.

The early meters really are becoming unreliable now and there are lots of spares/repair units available these days on ebay etc. However, it is likely the meter will have some issues, either the probes will be damaged or it won't be able to lock properly or maybe the meter dials will be peeling slightly.

It's generally best to find a later model like the one pictured in the original post. These were made in the 1980s and have an LED instead of a neon for the lock status and they have nicer looking meters with more detailed scaling. These should be a lot more reliable.

[G0HZU]

If you want to see an example of the meter peeling issue then take a look at the video below. If you look closely below the meter mirror in both dials you can see that the transparency is beginning to lift and curly slightly.

Once this begins to happen, the amount of actual peeling will depend on temperature changes in the room and eventually, the peel will be big enough to jam the meter needle. It's best to do something about it asap rather than let it get worse and worse.




In the video, the HP 8405A measures the short term drift of two Agilent E4433 signal generators each set to 10MHz and 0dBm. Both generators have the 1E5 oven option.

The meter shows 1 degree of phase shift in about 8 seconds and this means that the generators are adjusted to be within about 0.00035Hz or 3.5E-11 of each other.

If the recorder output at the back of the meter is used instead of the big dial meters, then it's possible to measure phase differences down to below 0.1 degree. See the image below to see some more info about this technique.

[pdenisowski]

It's generally best to find a later model like the one pictured in the original post. These were made in the 1980s and have an LED instead of a neon for the lock status and they have nicer looking meters with more detailed scaling. These should be a lot more reliable.

Mine has been completely reliable (and accurate) with no peeling, etc. 

But to be honest, I normally use one of our portable VNAs (the ZNH) when I need to make vector voltmeter measurements.  The reason I was asking is that I'm putting together a presentation / video on this topic and want to be sure I'm not missing any "interesting" applications of VVMs.  There are tons of cool things you can do with a VVM (and I'm sure I don't even know half of them), but as a practical matter, I don't see many people using VVMs (old school or modern VNA version) outside of very specialized applications.

[coppercone2]

Maybe you should have a look at the service manual to see what is inside the probes and the probe cables. It's a lot more complicated than a piece of coax made with resistance wire.

There are ground connections and also four other wires inside each probe cable. The probe is a sampling gate (similar to that used in a sampling oscilloscope) and there are fine sampling pulses fed up the cables to a diode based sampling gate (followed by a JFET) in the probe tip and the sampling pulses are about 300ps wide. There is a 20kHz IF signal fed back down the probe from the sampler at the tip of the probe. So there is a lot going on inside these probes.

it don't matter how many wires so long its possible to re-terminate them in a reliable manner. That style of cable gets damaged near the inlet where the strain relief is, and it can end up being mad difficult just to trim it down 2 inches and reconnect it, if you have that steel or nichrome wire crap for any conductor.

[rf-messkopf]

Mine has been completely reliable

I'm surprised that you don't use the R&S version: https://www.holzleitner.com/el/rohde-schwarz-zpv/index-de.html. 

The ZPV can directly indicate circuit parameters and group delay, and has a tuned receiver. There is a tuner plug-in which can measure down to 10 Hz and has 1 megohms BNC inputs, which can be very useful.

Should be a great toy. Maybe I'll snap one up if I come across one cheap.

[pdenisowski]

I'm surprised that you don't use the R&S version: https://www.holzleitner.com/el/rohde-schwarz-zpv/index-de.html. 

The ZPV can directly indicate circuit parameters and group delay, and has a tuned receiver. There is a tuner plug-in which can measure down to 10 Hz and has 1 megohms BNC inputs, which can be very useful.

Should be a great toy. Maybe I'll snap one up if I come across one cheap.

I would have used a ZPV, but our North American instrument loaner pool .... strangely ... doesn't stock any instruments manufactured before German reunification

But yes, the ZPV is very cool - much more like a "real" VNA than the HP8405

And good luck finding a used ZPV these days - old HP vector voltmeters (both digital and analog) are pretty easy to find, but I have yet to find a deal (d.h. ein "Schnäppchen") on a ZPV.  And I'm looking ....

[rf-messkopf]

And good luck finding a used ZPV these days

There are almost always some on German eBay, but often for crazy prices and in questionable condition. I don't see me spending 1500 Euros just to play with one when there is already a ZVB sitting on the bench. I'm also currently trying to reduce the pile of essentially unnecessary equipment 

Here is one listed at 500 Euros + tax from a dealer, but unclear if that includes a tuner plugin. They are listed separately at 400 to 500 Euros each.

[G0HZU]

There's a couple of ZPVs on UK ebay for about £160 each but both are spares or repair items. The seller is well known and often sells surplus test gear as untested.

The HP 8405A has the appeal that is it very simple and intuitive to operate, and the big dials are nice to look at when looking at phase drift etc. It also offers recorder outputs at the rear to connect to a decent DVM for logging etc.

However, the IF phase meter and the IF voltmeter in the HP 8405A both use mid 1960s technology and one has to keep the frequency range control set correctly to keep it in lock. So it is a bit fiddly to use it at times. It's probably best viewed as an educational tool today.

For many years, the HP 8405A was also used for measuring crystals. If a synthesised (accurate to <1Hz) sig gen is used with it and a suitable test fixture is made, it is possible to measure the resonant frequency of crystals very quickly and easily. This can also be done with a modern VNA, but it isn't as rewarding as tuning the sig gen and watching for zero phase on the big analogue phase meter on the 8405A.

It's possible to measure to within about 1Hz when using this technique. If the crystal is designed for a 20pF load capacitance, then simply fit a precision 20pF cap in series with it and measure it for series resonance. It's very quick and easy to measure a large batch of crystals using this method. I think some crystal manufacturers would have used this setup.

I'm not sure what the modern equivalent will be, but it will probably be a VNA locked to a decent OCXO frequency reference.

[G0HZU]

Maybe you should have a look at the service manual to see what is inside the probes and the probe cables. It's a lot more complicated than a piece of coax made with resistance wire.

There are ground connections and also four other wires inside each probe cable. The probe is a sampling gate (similar to that used in a sampling oscilloscope) and there are fine sampling pulses fed up the cables to a diode based sampling gate (followed by a JFET) in the probe tip and the sampling pulses are about 300ps wide. There is a 20kHz IF signal fed back down the probe from the sampler at the tip of the probe. So there is a lot going on inside these probes.

it don't matter how many wires so long its possible to re-terminate them in a reliable manner. That style of cable gets damaged near the inlet where the strain relief is, and it can end up being mad difficult just to trim it down 2 inches and reconnect it, if you have that steel or nichrome wire crap for any conductor.

I suspect that one reason there is no connector there, is that the differential pair that carries the 300ps wide sampling pulses up to the probe tip needs to have minimal discontinuities all along its length. Adding a detachable connector part way along the probe would degrade this to some degree. So the best solution is a continuous and pure cable pair from the source and then through the cable and up to the resistors at the diode sampling gate in the probe tip. Also, don't forget that this is mid 1960s technology.

[coppercone2]

maybe a modern high quality connector can handle it in a mod. But differential connectors are fairly alien to me. They do keep saying connector technology moved forward.

But the manufacturing precision for connectors is seriously impressive. Those RF connectors are like +-500nm tolerance for center drilling in the microwave connector area.


I wonder what connector would work here.


something along the lines of this, but this is data
https://www.rosenberger.com/products/connectors-adaptors/data-connectors/

[G0HZU]

Here's a link to an old HP document that lists some typical applications for the HP 8405A.

https://www.hpmemoryproject.org/an/pdf/an_91.pdf


Here's some info in the link below about testing crystals with the HP 8405A and an HP 8640B sig gen.

https://www.hpmemoryproject.org/an/pdf/an_171-1.pdf

The short term frequency stability and (very) close to carrier phase noise and jitter of the old HP 8640B sig gen will be poor compared to a modern synthesised sig gen and I think this is why the app note uses the LPF and the FM input of the 8640B. This is probably there to clean up the close to carrier jitter and noise of the 8640B sig gen and it may also help with tuning it reliably and easily in small increments of a few Hz.

A modern synthesised sig gen will be much cleaner and more stable (with minimal jitter) compared to the old 8640B at frequencies 'very' close to the carrier. It will also be much easier to tune up and down by just a few Hz compared to the 8640B. So it won't need the  LPF at the phase recorder output. The modern sig gen would need to be able to tune in steps of 1Hz or less but this is usually not a problem. I have used this crystal measuring technique with my 8405A and a modern sig gen several times now. I have a few homebrew test fixtures for this stuff. It is very rewarding to do it this way when compared to trying to do it with a modern VNA.

[paul@yahrprobert.com]

Back in the 90's I worked in a plasma lab where we did high power RF plasma heating.  The big problem was impedance matching to the plasma, which was a difficult load.  We had matching networks made out of big vacuum variable capacitors and fixed inductors.  The key was to measure the complex voltages from well calibrated directional couplers, then using computer programs and models of the matching network and antenna to adjust the capacitors. We used an HP 8508 vector voltmeter to calibrate the DC's, and then built our own multichannel fast vector voltmeters to measure the load when the plasma formed. We were working in the range 7 to 20 MHz.  I've always wondered why nobody seems to make a good vector voltmeter these days.  All you can buy is an expensive vector signal analyzer that gives you only one channel.

[coppercone2]

that is very interesting

[Bud]

The home made N2PK VNA (DDS based) has a Vector Voltmeter mode with a fraction of a degree and amplitude resolution.

[G0HZU]

I had a go at logging my newer HP 8405A  VVM (made in 1986) to see how stable the phase is over time. This is done using a DVM connected to the phase recorder at the back of the meter.

The result is quite good and it appears to have drifted about 0.005 degrees in 200 seconds. This is much better than my older HP 8405A but I have recently improved the stability of the 20kHz reference oscillator on the older meter. So it might now be similar to the plot below.

[G0HZU]

I'm not sure if this counts but I also use an Analog Discovery 2 using its network analyser mode. This measures and displays voltage and phase. It can also measure impedance using a series sense resistor. The main disadvantages with the AD2 are that it only works well up to a few MHz and it isn't very sensitive.

I also have a similar test fixture for the HP 8405A that uses a series sense resistor in a test fixture.
This was the first fixture I made for it as I wanted to measure negative resistances up at VHF using the IV method.

I also made a cheap clone of the HP 'tee' used with the HP 8405A. These can be difficult to buy cheaply now so I designed my own. The aim of the tee is to preserve a low VSWR up into UHF when the probes are used in a 50R system. The aim is to absorb the 2.5pF loading capacitance of the probe into the tee so it maintains a low VSWR when connected.

I also designed some simpler (compensated) terminations for my 8405A and these work well up at a few hundred MHz.

[quince]

By coincidence I wrote an article on hyperbolic navigation a couple of weeks ago! It's just some notes that I decided to write up because I found it interesting, and hope maybe others do too.

Just a heads up, on desktop your page renders with painfully large fonts and images+text that span the entire screen.

The page is much improved by removing all of the font-size specs in the css style section and changing all the image widths from width="100%" to something like 40 or 50%. You could also add a max-width value for text sections so they aren't a meter across.

[shabaz]

You're right, I'm not happy with that either. I put in the sizing hack because when I asked a small group for feedback, the response was it was too tiny for comfortable reading on mobile. I'm not very knowledgeable in this area, and tried finding a way to scale the text, but the settings currently only work for the paragraph text, and the headings (and images) explode in size on desktop as you've experienced. Mobile view looks great I feel, at least, it appears very comfortable on my mobile, and on the phones of a few people I asked to check. I'd like the desktop view to be comfortable and not crazy-huge as it is now. I'll ask a web developer friend to help out and create a pull request with a proper method that works for all. The workaround for now on desktop is to resize the browser window narrower, otherwise there are huge headings/images : (

[shabaz]

(Hopefully) resolved pretty much! I phoned a friend, and a suggested method was to use 'media queries'. Now it's fairly readable on mobile and desktop. I'll tweak the settings further to make it more graceful in the sizings, but for now hopefully it's a lot more readable than before (you might have to do Shift-F5 etc to clear cache if it currently still looks massive).