Measuring low impedances with a VNA

[precaud]

I recently refurbed and calibrated an Anritsu MS420K network analyzer, with the intent to use it to measure impedance (milliOhm range) of pcb power distribution networks (PDN), in powered and passive states, as well as separate components. This is the basic setup, used/recommended by HP/Agilent, Ridley, and others:



I did this successfully in the late 90's with passive components (without the blocking-C) on a test jig using an HP 3570A and 3330B combo, with a Tek current probe in place of the 1 Ohm resistor.

Now I'd like to do this same measurement on a pcb, to see the VR output impedance. There will be no room for a current probe in most cases so the 1 Ohm R will be used.

Desired measurement range is 1m Ohm to 10 Ohm from 10Hz to 10MHz.

A major disadvantage of this configuration is that the isolation xfmr is basically working into a 1 Ohm load. Most of the available isolation xfmrs were not designed for this.

So I thought the place to start would be to examine the transformer freq response from a 50 Ohm source into a 1 Ohm load over the range of interest.

Attached is the FR of six candidates for the job, and a pic of the units. Some of these are recommended for measuring loop gain/phase in switching supplies.

All of these are dual-winding xfmrs except for the Jensen, which is a video isolation "humbucker". Other notes:

The Tokin common mode choke is the best of several such chokes I tested. It is maybe useful to 20kHz. Maybe.
Bandwidth-wise, the Jensen humbucker and the three North Hills units perform pretty much the same. Using them to 1MHz would be realistic; 10 may be pushing it under actual conditions.
Apart from the dip at 20kHz, the Ridley looks good. It measures flat well beyond the 10MHz target. Except for a ~3dB peak @ 22MHz it is usable to 30MHz.

Your comments are welcome, especially if you have experience with this technique (or similar).

  John

[Pitrsek]

I started with impedance measurements with S21 and vanilla 2 port VNA(no dedicated generator port), when we bought 3 port VNA, we bought current injector as well. So I never used the injecting transformer for impedance measurements(only for stability). The setup is for me more cumbersome compared to I/V. But I sure can understand that that's what you are equipped for. Colleague of mine experimented with it, I'll try to ask him about it.

Few hints for S21 and I/V measurements:
With S21 it's necessary to watch out for DC voltage levels!(even if you have AC coupled  VNA inputs, generator output is usually DC coupled). If you need to measure higher voltage than is allowable at ports of your VNA, you can add resistors in series with the ports. This will also shift your measurement window:
https://www.edn.com/electronics-blogs/impedance-measurement-rescues/4439326/Increase-range-in-2-port-impedance-measurements
For direct I/V measurements, take a look at AN 104 from LT and late Jim Williams - quite nice voltage controlled current load.
You can buy something similar from Picotest(that's what we use). With current injectors, you cannot obviously measure impedance with power off. 
When measuring low impedance at low frequency, there is a problem of braid induced error:
http://www.electrical-integrity.com/Paper_download_files/DC10_12-TH3_Novak-Mori-Resso_slides.pdf
Picotest is offering CM transformer to counter it(that we do not have), but you can roll your own. I Had a good experience with "slightly smaller than palm" sized EMC ferrite from Wurth. Also Vacuumschmelze is offering some cores with crazy high permeability. Depends on how low impedance and how low frequency you want to measure. If you want to go really low with frequency, you will need differential input. Differential probes works well, but most of them are 10x and will eat into your SNR. Use of short semi-rigid cable with DUT mounted as close as possible to port 1 also helps.
I usually provide two SMA per power plane on my boards for impedance measurements. SMA are 90deg. apart, so there is no collision between injector and voltage probe. Injector is mounted directly to PCB with adapter - no cable. Easy setup, great signal integrity. Just connect voltage probe, current injector, and you are set. 

For separate component measurements, I use S21(Z<1Ohm) shunt with CM transformer. Or dedicated fixture from VNA manufacturer(Omicron)
I would recommend reading up on papers from Istvan Novak, Steve Sandler (articles, videos on YT + live classes, I've been to one and it was really nice  ) and take a look at app notes from Omicron Labs (makers of our VNA). If you are interested, I can try to find PN of cores for CM transformer and snap a few pictures of the setup I use.

On a related note, it's possible to load your measured data into LT spice - both transfer functions, and impedance. So you can simulate with measured data.

[precaud]

I started with impedance measurements with S21 and vanilla 2 port VNA(no dedicated generator port), when we bought 3 port VNA, we bought current injector as well. So I never used the injecting transformer for impedance measurements(only for stability). The setup is for me more cumbersome compared to I/V. But I sure can understand that that's what you are equipped for. Colleague of mine experimented with it, I'll try to ask him about it.

Thanks Pitrsek, that would be great. BTW, I'm not married to this setup, I am open to other methods if they are effective.

Few hints for S21 and I/V measurements:
With S21 it's necessary to watch out for DC voltage levels!(even if you have AC coupled  VNA inputs, generator output is usually DC coupled). If you need to measure higher voltage than is allowable at ports of your VNA, you can add resistors in series with the ports. This will also shift your measurement window:
https://www.edn.com/electronics-blogs/impedance-measurement-rescues/4439326/Increase-range-in-2-port-impedance-measurements

I will rarely be looking at anything with greater than 20VDC, but I'll look into it.

For direct I/V measurements, take a look at AN 104 from LT and late Jim Williams - quite nice voltage controlled current load.
You can buy something similar from Picotest(that's what we use). With current injectors, you cannot obviously measure impedance with power off. 

Thanks for the AN reference. I was looking at the Picotest J2111A specs last night. A bit pricey...

When measuring low impedance at low frequency, there is a problem of braid induced error:
http://www.electrical-integrity.com/Paper_download_files/DC10_12-TH3_Novak-Mori-Resso_slides.pdf 

Yes, that one I've been aware of. My previous setup avoided it, broke ground the loop, by the use of the current transformer, but it will be a factor now.

Picotest is offering CM transformer to counter it(that we do not have), but you can roll your own. I Had a good experience with "slightly smaller than palm" sized EMC ferrite from Wurth. Also Vacuumschmelze is offering some cores with crazy high permeability. Depends on how low impedance and how low frequency you want to measure. If you want to go really low with frequency, you will need differential input. Differential probes works well, but most of them are 10x and will eat into your SNR. Use of short semi-rigid cable with DUT mounted as close as possible to port 1 also helps.

I guess it depends on what we define as "really low". My desired low limits are 10 Hz and 1 milliOhm.

I usually provide two SMA per power plane on my boards for impedance measurements. SMA are 90deg. apart, so there is no collision between injector and voltage probe. Injector is mounted directly to PCB with adapter - no cable. Easy setup, great signal integrity. Just connect voltage probe, current injector, and you are set. 

That is smart. My applications will be to existing boards, mounting connectors won't always be possible. Sometimes I just have to solder the leads to the traces.

For separate component measurements, I use S21(Z<1Ohm) shunt with CM transformer. Or dedicated fixture from VNA manufacturer(Omicron)
I would recommend reading up on papers from Istvan Novak, Steve Sandler (articles, videos on YT + live classes, I've been to one and it was really nice  ) and take a look at app notes from Omicron Labs (makers of our VNA). If you are interested, I can try to find PN of cores for CM transformer and snap a few pictures of the setup I use.

Thanks for the references, I have some homework to do    Any pics of your setup would be great.

On a related note, it's possible to load your measured data into LT spice - both transfer functions, and impedance. So you can simulate with measured data.

Yes, I do that with software I've written. Once I settle on the measurement setup, I'll modify it to accomodate.

[precaud]

I just wanted to add, the purpose of this thread is not just to explore this one technique... If you're doing low-Z measurements with a VNA, pls share your setup.

[Pitrsek]

Picotest injectors - yes, they are pricey. If you go for a bundle it gets a bit more reasonable. There is one "feature" of the injector that i did not manage to wrap my head around. The current sense port is "feedforward", ie. the signal is not about what is there, but about what should be there. Current sense output works even with disconnected load. I have on idea why did they designed it this way.  On the other side, it's nice for calibration

10Hz and 1mOhm - it's quite tough. Depending on dynamic range of your VNA, you might probably need to build a preamp. Ideally with differential input, to get rid of the braid error(If you feel like building something like that - let me know, I'd be glad to help - I could use it too, same goes for the current load). If I may ask, why do you need to go so low?  There is usually not much happening...

Apart from two sma(really nice if you can accommodate the space, I use a lot of edge mounted ones), I used diy probes from semirigid coax and some needles(as recommended in "right the first time" book) - it sucked.  I have a new version with spring loaded gold plated tip in works, I'll post pictures when it's done.

I've attached pics of my setup - short semi rigid line +  common mode transformer. Of course it could be improved, ie. directly from coax to bnc(no adapters) and semi flexible coax for the CM transforem(+box). PN of the core is W518-03. Measured are 4x0.1R 0603 in parallel. I'll try to get something smaller, to show you the braid error with this setup. Actually this is quite a bit better than old setup with WE emc ferite core.

[Pitrsek]

Managed to find 10mOhm resistor(LVK12R010FER). Below 10Hz you can see braid error kicking in. Unfortunately I do not have anything smaller at hand. With worse CM transformer, the transition would just happen higher in the frequency domain.

[precaud]

Thanks for the details and pics, Pitr. That CM xfmr is a monster!

Picotest injectors - The current sense port is "feedforward", ie. the signal is not about what is there, but about what should be there. Current sense output works even with disconnected load. I have no idea why did they designed it this way.

I'm puzzled by this.

10Hz and 1mOhm - it's quite tough. Depending on dynamic range of your VNA, you might probably need to build a preamp. Ideally with differential input, to get rid of the braid error

Yes, preamps with diff inputs, a source that can comfortably drive a few watts into 1 Ohm at 30MHz... they're on the list of things to build some day.

Dynamic range with the Anritsu is better than 90dB with 30Hz IF BW, and even better than that at lower BW's. It has isolated inputs, though "how much" they float off common is not specified. Perhaps I should determine that, both inter-channel and channel-to-source-out. I have read several ways to deal with the braid error issue. Diff inputs are one. Isolation xfmrs are another. Keysight does in their VNA's by lifting the input grounds by 30 Ohms, shown starting on p. 19 of this PDF. It's pretty effective:

literature.cdn.keysight.com/litweb/pdf/5990-5902EN.pdf

If I may ask, why do you need to go so low?  There is usually not much happening...

Some Vregs dip even below 1mOhm in places. To accurately do simulations combined with measured data, the data needs to be accurate...

I've attached pics of my setup - short semi rigid line +  common mode transformer.

Very interesting.This looks good for SMD parts. I see in the circuit setup, there is no connection to the R(ef) input of the Bode 100. Where does it get its reference? Do you make a "thru" sweep first? Or open/short/load sweeps which the instrument stores and applies to each measurement?

I am going to have to read further and experiment with this "shunt-thru" setup.

I can see, it is going to be a challenge to do on-board probing that is good to 10MHz. Unfortunately I won't be able to solder SMA's to the necessary test points most of the time. Ideal would be to clip onto the test points.

Managed to find 10mOhm resistor(LVK12R010FER). Below 10Hz you can see braid error kicking in. Unfortunately I do not have anything smaller at hand. With worse CM transformer, the transition would just happen higher in the frequency domain.

Yes, it is an interesting result. Is the behavior above 100kHz entire the DUT, or is some of it residual L in the setup? I would expect a device that small to not have so much.

Thanks again for sharing your setup and results.

[Pitrsek]

Floating inputs are nice  . The Anritsu looks nice. Thx for the link to Agilent VNA - an interesting solution.
Unfortunately yes, some regulators have ultra low output impedance. In my experience, you can usually see the lowes Z out of regulator before the braid error kicks in(so you have your R an L for model). Anyway, with your floating inputs, it's a moot point anyway . Take a look at some presentations from Steve Sandler about flat output impedance and why it is good idea. Might be interesting for you.

Ref channel is connected internally. Bode is quite flexible in this regard(see attachment). For S21 shunt, I calibrate for trough only. It should be ok for Bode 100 bandwidth - info from Mr. Sandler - i forget to ask why(so many information in one presentation... ). For higher bandwidth, full calibration is recommended.

If SMA is too big, you can solder coax directly to capacitor Pads - capacitor should be removed, if you keep it, the ripple current flowing trough vias will impair your measurements. There's an article about it from Istvan Novak somewhere on the internet.

Or you can make impedance probes - piece of semi-rigid coax(first one I built were from semiflex, not a good idea, too soft/bendable), for a ground connect you can use a needle, or better gold plated spring loaded pin - the problem is that center connector is still quite soft copper, but it works.
On the left it's a new design - proper semi-rigid calbe + center conductor replaced with Ingun test needle (50mil series, with housing, tip is gold plated blade type, changable).It's work in progress. The probe is quite short, as the center conductor is replaced with test needle housing(after pulling out the center conductor, the hole needs to be increased slightly to accommodate needle housing). Both contact spring loaded. It's a still work in progress. Or alternatively, you can take a look at Ingun, they do offer a HF test probes, which could be "misused". I will probably try them in future.

Actually the 10mOhm resistor is surprisingly good in this regard. The sharp rise of the blue trace above 100khz is phase. I would say that impedance starts to rise a decade later, around 1M.  It ain't easy to have a flat and low impedance to MHz range. The inductance of package will bite you in the ***. For example 0.314nH will exhibit cca 0.010R imedance at 50Mhz, since the resistor is 10mOhm on it's own, it needs to be way smaller not to impact resistor impedance significantly. For comparison, mounting inductance for various packages:
http://www.sigcon.com/Pubs/news/6_09.htm

Glad to help

 

[precaud]

Unfortunately yes, some regulators have ultra low output impedance. In my experience, you can usually see the lowes Z out of regulator before the braid error kicks in(so you have your R an L for model).

So, far my experiences agrees, lowest Z has always been above 100Hz.

Take a look at some presentations from Steve Sandler about flat output impedance and why it is good idea. Might be interesting for you.

Thanks. I've watched some of his presentations and he definitely knows his stuff. I am already convinced of the need.

If SMA is too big, you can solder coax directly to capacitor Pads - capacitor should be removed, if you keep it, the ripple current flowing trough vias will impair your measurements. There's an article about it from Istvan Novak somewhere on the internet.

I've either been clipping or soldering to the test points, which quite often coincide with the output capacitor(s). Exact location sort of depends on where the regs tie into the ground plane (assuming one exists!). I'll have to look up that article, thanks.

Or you can make impedance probes - piece of semi-rigid coax(first one I built were from semiflex, not a good idea, too soft/bendable), for a ground connect you can use a needle, or better gold plated spring loaded pin - the problem is that center connector is still quite soft copper, but it works.
On the left it's a new design - proper semi-rigid calbe + center conductor replaced with Ingun test needle (50mil series, with housing, tip is gold plated blade type, changable).It's work in progress. The probe is quite short, as the center conductor is replaced with test needle housing(after pulling out the center conductor, the hole needs to be increased slightly to accommodate needle housing). Both contact spring loaded. It's a still work in progress.

Very nice. This appears worth experimenting with. I bet you never guessed that one of your required engineering skills would be to have a "steady hand" to hold the probe during the Z sweep! 

Or alternatively, you can take a look at Ingun, they do offer a HF test probes, which could be "misused". I will probably try them in future.

I've never heard of them, will have a look , thanks.

Actually the 10mOhm resistor is surprisingly good in this regard. The sharp rise of the blue trace above 100khz is phase. I would say that impedance starts to rise a decade later, around 1M.

Wow, so at that low impedance, the L component becomes visible already at 100kHz.  Hmmm...

It ain't easy to have a flat and low impedance to MHz range. The inductance of package will bite you in the ***.

So I am seeing...

For example 0.314nH will exhibit cca 0.010R imedance at 50Mhz, since the resistor is 10mOhm on it's own, it needs to be way smaller not to impact resistor impedance significantly. For comparison, mounting inductance for various packages:
http://www.sigcon.com/Pubs/news/6_09.htm 

Interesting. Beside the trace width, the direct correlation is between the via spacing and the inductance.

Does the Bode100 box have open/short/load compensation routines available? Back in the 90's, my component Z measurements with current probe didn't work very well until I added those routines to the program. The short and load, in particular. At low Z's, we can pretty much forget about the open condition.
 
Thanks for bringing all these stinky details to my attention.     This is going to be quite a challenge, indeed.

[Pitrsek]

Yes, Bode supports open,short,load calibration. I like Bode 100, it's a handy piece of kit, and techsupport from manufacturer is top notch
Stinky details, that's a rather fitting description... It reminded me about switchable probes, it's ultra easy to accidentally switch probe attenuation, when trying to connect it to a pcb that was not really prepared for probing. Bought dedicated 1x probes - saves the hassle with checking probe settings, provides better bandwidth than switchable probe, and as a bonus, none is interested in measly 1x 40MHz probe....  so  nobody is "borrowing" them
Good luck and let us know how your measurements turned out. 

[rx8pilot]

Joining the thread.....

[precaud]

Thx for the link to Agilent VNA - an interesting solution.

There are actually several different approaches described in that article. It covers most of the bases in pretty good detail. I read thru it again today, and while the the shunt-thru is good for component testing, it may not be the best choice for measuring regulator Z. Because the terminal impedances are so low (50 Ohms), the DC-blocking C's have to be huge (and bipolar) 'lytics. And that brings yet another bag of worms to the table, with capacitor self-L becoming a factor at quite low freq's (<20kHz) in big 'lytics...

On the left it's a new design - proper semi-rigid calbe + center conductor replaced with Ingun test needle (50mil series, with housing, tip is gold plated blade type, changable).It's work in progress. The probe is quite short, as the center conductor is replaced with test needle housing(after pulling out the center conductor, the hole needs to be increased slightly to accommodate needle housing). Both contact spring loaded. It's a still work in progress.

I like the idea, but I'm thinking that contact resistance would be variable with that sort of setup, which would be fatal for millOhm-range measurements. Are you seeing that?

I'm also wondering about the jig made with rigid coax for testing SMT parts. As Johnson shows, "parasitic inductance" is basically related to electric length, i.e. spacing between conductors, vias, or test leads. Since that measurement setup has a fixed distance between the leads, they can't be brought together for an accurate Short compensation measurement. So short will include that inherent inductance, and it will be erroneously removed from every measurement by the post-processing math.

Do you have a chart of "useful Z measurement range vs frequency" for the Bode 100? Here's one for an AP Instruments AP200 VNA. This chart is from an application guide for the instrument.

  It ain't easy to have a flat and low impedance to MHz range. The inductance of package will bite you in the ***.

That is SO true...

[precaud]

I made a 50 Ohm splitter today and did first experiments with the shunt-thru method using the MS240. See first attached image for how everything is connected. Because I used short test clips between the DUT and the inputs, and axial-leaded current sense resistors, I knew that above 100kHz it would not be correct. I consider this a "test of concept" anyway, and also very much wanted to see if the MS420's isolated inputs do in fact eliminate the braid loop error at low freqs.

Second image shows 10Hz-10kHz sweeps of 1 Ohm, 100mOhm, 10mOhm, and short DUT's. (Formulas for finding the impedance value:  Zdut= 25 x S21 / (1 - S21), where S21= Vt / Vr. Convert dB to voltage ratio by 10^dB/20). The Z's calculate out correctly, so the method works. And there is no sign of braid loop error, so the isolated inputs do the trick.

There are a couple problems, however. First, the minimum IF bandwidth is 3Hz, so at 10Hz it is still tracing out the shape of the IF filter at low impedances. You can see this in the 10mOhm and short plots. This results in a minimum usable freq at 10mOhm of about 20Hz, and at 1mOhm it would be about 30Hz. That's not bad, but not what I was hoping for. And second, in order to keep sweep times at low freqs reasonable, the MS420 has a mode (which I used) where it automatically adjusts the resolution bandwidth to the frequency as it sweeps (at 1-3-10... intervals above 100Hz). Wider bandwidths allow faster sweep times but result in a higher noise floor. In the 10mOhm and short sweeps, you can see the noise rise above 1kHz and again at 3kHz. The workaround is to use a fixed narrow IF bandwidth for the entire sweep. The penalty for doing so very is looooong sweep times. Using a 3Hz filter, it would take several minutes per sweep.

So right off the bat we see the MS420 is not bad but not the ideal instrument for my goal of 10Hz to 10MHz and 1mOhm to 10 Ohms. At low freqs my HP 3577A would be even worse, with its 10Hz minimum IF BW and non-isolated inputs. For measurement down to 10Hz, a 1Hz IF filter BW is needed.

Another workaround would be to use the VNA for 100Hz and above, and use an FFT-based dynamic signal analyzer for the low freqs.

[Pitrsek]

Yes, the 25R(50R of generator and 2 port are in parallel) of loading is limiting the S21 usefulness. Yet, if you go for DC block on port 2, the cap need not to be ultra large, and it can be calibrated out. Another trick is to increase your system impedance with series resistor:
https://www.edn.com/electronics-blogs/impedance-measurement-rescues/4442173/1/Fix-poor-capacitor--inductor--and-DC-DC-impedance-measurements
This will decrease your DUT loading, but will shift your measurement window as well(in many cases, this might not be a problem at all)
I use the increased system impedance technique for measuring inductors and chip-beads - otherwise one would need to combine S21 and S11 data to have accurate measurements trough whole impedance range. The S21 measurement for regulators got one really nice advantage, you can measure just passives/caps, then turn the psu on, measure it again, and observe the difference....

Yes, contact resistance will be variable, but a) its gold plated blade pin, b) the sensitivity of S21 to contact resistance is quite low
The contact resistance is THE reason to use S21 instead of S11 for impedance measurements. For S11, contact resistance is in series with DUT and you are not able to tell it apart from dut. With S21, the contact impedance is in series with ports. So you have 50ohm + mOhm of contact impedance, then dut to ground, and parallel to DUT is another contact impedance in series with 50ohms of port 2.
see "Why S11 VNA Measurements Don’t Work for PDN Measurements" and "Accuracy Improvements of PDN
Impedance Measurements in the Low to Middle Frequency Range "

For the rigid coax jig, I only use trough calibration. You are correct about the package inductance. In some cases, it gets calibrated out.
I've seen this even in some books, I need to ask the book author(Sandler) why he did it this way. 

Not for the S21 measurements. Only for use of their component fixtures(b-wic and b-smc).

I've inquired about price of Ingun HF probes, I'll let you know when I hear back.

 

[precaud]

Yes, the 25R(50R of generator and 2 port are in parallel) of loading is limiting the S21 usefulness. Yet, if you go for DC block on port 2, the cap need not to be ultra large, and it can be calibrated out. Another trick is to increase your system impedance with series resistor:
https://www.edn.com/electronics-blogs/impedance-measurement-rescues/4442173/1/Fix-poor-capacitor--inductor--and-DC-DC-impedance-measurements
This will decrease your DUT loading, but will shift your measurement window as well(in many cases, this might not be a problem at all)

Yes, but that addresses a different problem. It shifts the measurement window up, extends S21 usefulness into higher impedances, at the expense of lower. I can see how it would be useful for your beads and inductors at high freqs, though.

The S21 measurement for regulators got one really nice advantage, you can measure just passives/caps, then turn the psu on, measure it again, and observe the difference....

Yes... very convenient!

Yes, contact resistance will be variable, but a) its gold plated blade pin, b) the sensitivity of S21 to contact resistance is quite low
The contact resistance is THE reason to use S21 instead of S11 for impedance measurements. For S11, contact resistance is in series with DUT and you are not able to tell it apart from dut.

Good points, I forgot about that.

For the rigid coax jig, I only use trough calibration. You are correct about the package inductance. In some cases, it gets calibrated out.
I've seen this even in some books, I need to ask the book author(Sandler) why he did it this way.

I saw a video where the residual L was being manually "tuned out" by making successive sweeps and tweaking the L constant for the software until it looked good. I'm not fond of this approach. It's probably OK for medium impedances. I've seen it used in several instruments, even 4-wire Z meters (like HP 4192A). The problem is, the residual L is not pure and so it varies with freq. HP abandoned that approach in their next generation of Z meters. A better approach is to take correction data at numerous freqs across the whole range, and interpolate between them. I'm pretty sure your Bode 100 box does that. The AP VNA's use 15 points.

I've inquired about price of Ingun HF probes, I'll let you know when I hear back.

No doubt there will be some sticker shock involved   

[Pitrsek]

Yes, but that addresses a different problem. It shifts the measurement window up, extends S21 usefulness into higher impedances, at the expense of lower. I can see how it would be useful for your beads and inductors at high freqs, though.

Yes, indeed. The think is that in many cases, this will not be a problem. Usually when the DUT loading is the problem(low power LDO, under Amp), the Z out of LDO is quite high anyway, so there is no harm in shifting measurement window up and crating voltage divider for VNA prots in the process. The same goes for higher voltage regulators, usually the Zout is not really low so, there is no harm in adding series resistors. I'm not saying that there are no 15V 3mOhm power supplies, not at all. I'd just like to point out that in some cases(quite a few in my experience), simple addition of the resistor might be very helpful  .

For low frequency measurements a good sound-card  could be suprisingly potent, off course bandwidth is quite limited.
I used it for Bode measurements before we bought Bode, it could be used for impedance measurements with current load.

Are you interested in stability measurements as well?

[precaud]

For low frequency measurements a good sound-card  could be suprisingly potent, off course bandwidth is quite limited.
I used it for Bode measurements before we bought Bode, it could be used for impedance measurements with current load.

I agree. Most of my work is within the audio bandwidth, and I have good tools for it. HP 3567A (a PC-based 3562A with more power and 16 bit DACs and ADCs), and HP 4276A, a highly underappreciated Z meter and the most capable unit ever made within its freq range.

Are you interested in stability measurements as well?

Not at this point.

[Pitrsek]

The Ingum probes are not cheap(who would have thought ), but I'm yet to find a solution that is cheaper(picotest and sequid are way more expensive). The price is cca 170eur for one. One of the type was cheaper, 140eur, but it had a wider spacing. Not cheap, but compared to others, actually not that bad either. I'll probably try to finish the spring loaded DIY probes...

 

[precaud]

Are you looking at what they call "dipole" test probes, in the RF section?

[Pitrsek]

Yes, I was interested in following:
HFS-810 204 051 A 02 V1-AS3 - cheaper one
HFS-810 201 051 A 06
HFS-810 358 080 A 02 V2-00S
HFS-810 201 051 A 29 V2
HFS-810 307 100 A 02 V2-36S

[precaud]

Well they're beautiful. 2GHz is way beyond anything I'd need.

I have some thinking to do about my objective. Given that the spacing between the test points (and, therefore the probe terminal spacing) determines the highest useful frequency (before inductance dominates), my 10MHz target is looking unrealistic for what I want to do with it. Most of my measurements will be made to existing boards, with mostly thruhole parts. In most cases I can't control the location or spacing between the test points. Most of the time I will have to use either clip (or solder) leads to the test points.

To see the impact of short leads, here's a plot of shunt-thru using the MS420K from 1kHz to 10MHz, of two "DUT's", one a short using 21-ga wire, the other an axial-leaded 25mOhm current sense R, with different test setups. One setup uses short (3") clip leads for the DUT. The other setup, the DUT is soldered directly into one end of a BNC tee, which maintains the 50 Ohm environment. As you would guess, the smoother curves are the ones using the tee.

For either setup, 1mOhm is out of the question. It's right at the noise floor. If I drop the IF bandwidth another step and slow the sweep, it might clean it up a little. 10mOhm would be usable, but only to 100kHz max.

So just like the low-freq test a few posts back, 1mOhm is right on the bleeding edge, and imposes frequency limits at both ends. Perhaps it's time to take a look at the transformer I/V method to compare its results to the shunt-thru.


If I'm limited to 10mOhm and 100kHz, I don't need a VNA for that. An HP 3562A will work just fine, and much faster.

Hmmm...

PS - The Ingun website is the slowest I've encountered in years. Is it that way in EU too?

[R_G_B_]

I'm not sure if there are people who are aware that digilent have released an update to their analogue discovery waveforms 215 that now includes an impedance analysis tool.

[Pitrsek]

Could you please post pics of your test setup?
I might be able to replicate, so we can compare results. What is representative spacing on your boards?
I would say that the inductive rise is not about limit of the test setup, but more of a DUT limit. Axial-leaded and low inductance don't go well in one sentence. Off course, it's connected. If the DUT is big, it requires big spacing, it's inductive(regardless of test probe spacing). The thing is, if it's THT board, there are no mOhms at 10MHz anyway.... If the DUT was small, but your setup forced use of unnecessary big spacing, then we could talk about exces inductance of the test setup.
Depending on noise performance of the analyzer, simple opamp based low noise pre-amp might help with the dynamic range.
Have you tried the semi-rigid test setup with small SMD resistor? To see where is the limit is?
Soldering coaxial pigtails to vacant capacitor pads is ok an works surprisingly well.
Ingun page is ok in EU.

[precaud]

Could you please post pics of your test setup?

Sure, see attached pics. 50 Ohm output goes thru 50 Ohm divider. One side goes into the Ref channel termed 50 Ohms, the other to the Tee and into the Test channel termed 50 Ohms. The short is just a wire soldered to the shell. The .025 Ohm current sense R is indeed axial, so some lead length is involved, as you can see.

I plan on cutting the shell flush with the center conductor. It will shorten the path length and I can then solder smaller parts to it.

To test the effect of cable length, I inserted a 16" extension BNC to each cable in turn, and sweeping the .025 Ohm R (to 10MHz) to see if there was any change. Nothing. So one nice thing about the shunt-thru method is, it is relatively insensitive to cable length (to 10MHz, anyway) . Very conducive to use with probes.

I might be able to replicate, so we can compare results. What is representative spacing on your boards?

I'd say 5 cm would be typical.

I would say that the inductive rise is not about limit of the test setup, but more of a DUT limit.

You may be right. I'm not sure yet. (Trust but verify...) The inputs on the MS420 go through several relays and RC trims before reaching the FET buffer. I remember comparing Z measurements with it to the HP 3570A years ago and I got better results with the HP. The 3570A input goes right into the FET. Very clean signal path. I attributed the better results to the input stage being much simpler. But I could be wrong.

Back in the day, Anritsu sold an impedance probe for the MS420, called the MA413A. It had a xfmr in the probe body right behind the tip. Rated impedance range was 1Ohm to 1MegOhm. So it seems they too got messy results at lower impedances, and so they just spec'ed it above the mess 

Axial-leaded and low inductance don't go well in one sentence. Off course, it's connected. If the DUT is big, it requires big spacing, it's inductive(regardless of test probe spacing).

Yes, that's essentially saying the same thing I said last post. The spacing (of probe, test points, DUT) basically sets the upper freq limit.

The thing is, if it's THT board, there are no mOhms at 10MHz anyway....

Hmmm... really? Even with a good ground plane and distributed capacitance? I guess it's fair to say, that is why I want to do this measurement - to see what you already know 

The other response I would make is: and therein lies the opportunity.

If the DUT was small, but your setup forced use of unnecessary big spacing, then we could talk about exces inductance of the test setup.

Yes, that is not the case. I know to avoid that...

Depending on noise performance of the analyzer, simple opamp based low noise pre-amp might help with the dynamic range.

I was thinking that yesterday. There's a 20-40dB level drop to the T channel (depending on the impedance). A 10MHz 20dB preamp could be put together.

Have you tried the semi-rigid test setup with small SMD resistor? To see where is the limit is?

I don't have any yet. But I'll be parting out a Tek 7000-series mainframe in the next couple days, it has an analog delay line made from coiled rigid coax and then I'll have a large supply of it 

Do you have SMA on the back end of your DIY probe? I'll probably stay with BNC. It's convenient.

Soldering coaxial pigtails to vacant capacitor pads is ok an works surprisingly well.

Makes sense. I'll be doing that sometimes, for sure.

Ingun page is ok in EU.

Ah, ok. They might have someone in their basement hosting it for the US 

Thanks for your input on this!

[precaud]

Today I made a setup for the injection transformer method, using one of the North Hills xfmrs.The setup is "leggy" (the leads are long) and some are unshielded, but again this is proof-of-concept and will be refined if it works. The setup for this method is in the first post of this thread. With DC blocking C of two back-to-back 3300uF lytics, it is ready to be used on a Vreg.

This is definitely a larger and more cumbersome setup than the shunt-thru. But to be fair, the latter didn't have DC blocking C's yet. And two are required.

This method must be used with short or short/load compensation in post-processing. Otherwise it is wildly inaccurate.

I have read that, for best accuracy, it is important to minimize the distance between the 1R current sense resistor and the DUT, so I attached it to the back of the grabber.

The reason for the banana plugs/jacks is that the test channel lead needs to be moved. It connects to the Vr test point for the thru sweep (to calibrate the analyzer and test jig). Then it is moved to the DUT for short compensation, and for the actual measurement.

And that in itself creates a problem. The thru sweep connects to the other side of the xfmr, so it is always 180º out of phase to the measured sweeps. That is why the phase is 180º in the attached graph. The fix is: I'll have to read the thru-sweep into the computer, flip the phase, and then shove it back into the analyzer.

The last attachment is Z/phase sweeps of two axial 3W current sense R's, .025 and .050 Ohms using this setup. As you can see, despite its messiness, it is actually better-behaved than the shunt-thru chart (see a couple posts earlier). The R050 is actually worse at high freq's, because it is further away from the short condition which is used to correct the data. To pull it into line, I'd need to add load compensation too. Very doable.

So shunt-thru has some competition! 

[precaud]

Well the last set of plots was too good to be true (an axial-lead R025 measuring flat Z to 2MHz ? Nope!). After correcting the phase inversion in the thru sweep routine, the curves look more realistic now. The first set of plots shows 25, 50, and 100 mOhm axial R's with short compensation. Above 100mOhm the resistance starts to swamp the reactance mess, and the curve becomes increasingly flat and extended.

The second set of curves shows three 10mOhm current sense resistors with different "leads"; a radial, an axial, and an SMD.The inductance of the axial package is clearly evident. Most interesting to me is that the radial-lead one was nearly as low-inductance as the SMD.

So as it sits, this setup is usable to 200kHz from 10mOhm up. 2MHz from 100mOhm up. If everything but the xfmr, clip leads, and Rsense was moved to a pcb, and the plugs/jacks were replaced by a low-impedance switch, it would no doubt extend higher in freq.

EDIT: I have added a pic of the three 10mOhm R's used in the second plot. In the plot, the SMD and radial parts measured quite similarly. And in the pic, we can see that their terminal spacing is pretty close. So as Pitrsek (and others) have said, the spacing sets the baseline inductive profile.

[precaud]

Going back to the shunt-thru method... I made a "fixture" to test SMD parts by cutting off the barrel of one end of a BNC tee flush with the center receptacle, so SMD's can be soldered directly across. See first pic. I have plenty of these things laying around doing nothing... finally putting them to good use 

Second pic shows the Z/phase in the MS420K using the shunt-thru method, axial vs SMD R025's. Much improved for the SMD. There is still some residual inductance in there, though; the Z rises too far and too fast. And the phase curve sits a bit above 0º even down at 1kHz. Makes me think it might be a setup error. Or perhaps short correction needs to be used. Thoughts?

Pitrsek, when you do your "through" measurement, is it simply the test setup without a DUT? That's how I did mine.

I'm 3/4 done making the rigid coax probe. Just gotta find a test point to use for the ground. Sad to say, I threw away a perfect specimen a few weeks ago. Sigh...

[R_G_B_]

I was looking at this method:inserting a resistor between each VNA port to increase dynamic range "TEE" configuration.

I have A VNA 8714  and anritzu network analyzer, 10H 30Mhz network analyzer,   and analogue discovery with the updated software that now supports impedance analysis. I was following this thread and I would like to see how to implement this and how each of these compare when measuring small ESR resistance.

The Omicron looks good its signal source like the analogue discovery is not permanently connected to the channel 1  which makes it versatile.

see the following:

https://www.omicron-lab.com/fileadmin/assets/Trainings_Events/Webinar_2014/141119_Webinar_Impedance.pdf

https://www.edn.com/electronics-blogs/impedance-measurement-rescues/4458562/Accurately-measure-ceramic-capacitors-by-extending-VNA-range

https://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&uact=8&ved=0ahUKEwiI__Ch_O3XAhVMKsAKHe80DRsQFghBMAQ&url=http%3A%2F%2Fwww.picotest.com.tw%2FDownload_File%2FINJ_um%2FApplication%2520Article%2FMeasuring%2520Output%2520Impedance%2520with%2520the%2520J2130A%2520and%2520Bode%2520100.pdf&usg=AOvVaw1_LB8_cPMuYVKD1YSgxJMU

 

 

[precaud]

Hi R_G_B,

Feel free to share your process and results, I look forward to it.

I have chosen not to explore the "adding series resistors to shunt-thru" approach because it adds measurement range (and accuracy) where I don't need it; at the low-frequency end. If you look at the charts in the EDN article you linked to, adding series resistors adds nothing to the measurement accuracy above 1MHz. The improvement it makes is all entirely below 100Hz. I don't need to see the entire impedance range of a 1nF capacitor from 1Hz to 50MHz. I have defined my range of interest as 10Hz to 10MHz, 1mOhm to 10 Ohms. Even this range is not easy...

Cheers.

[Pitrsek]

Still there, just not enough time to chime in. I'll be definitely back when time permits with some comparative measurements.
For coax s21 fixture, I do only trough calibration. That I can get by because the fixture inductance is so low. For the T setup, I would definitely consider performing short/load calibration as well. Even if you file down the T, there still a decent loop that will not get calibrated out with trough calibration, so you basically add this loop to you DUT.
Actually I have DIY SMA "calibrators" with cal. plane just at the back of the connector(you can use spare SMA connector and solder dut at pins). I'll try it and let you know.

[precaud]

Still there, just not enough time to chime in. I'll be definitely back when time permits with some comparative measurements.

OK, we'll wait. 

For coax s21 fixture, I do only trough calibration. That I can get by because the fixture inductance is so low.

Interesting. Is that for the coax probe, too?

For the T setup, I would definitely consider performing short/load calibration as well. Even if you file down the T, there still a decent loop that will not get calibrated out with trough calibration, so you basically add this loop to you DUT.

OK here are the same two 25mOhm resistors as measured last time, plotted with and without short compensation. It's quite an improvement. You can see by the phase curve that it is actually over-compensated a bit, the phase is in capacitive territory. That suggests to me that we may be up against the limit for this technique. It appears that using BNC T as a Z test fixture is OK up to 1MHz. For many things, this will be quite good enough.

The unanswered question, of course, is: what should the curve be for a single R025 of this size? Your four paralleled 0.1 Ohm R's gave nearly flat Z and phase all the way to 10MHz in your coax setup (remarkable, really). How much worse is just one .025 supposed to be?

It would be interesting to see the front end of your Omicron analyzer. I bet they have taken great care to maintain the 50 Ohm characteristic impedance right into the FET buffer.

BTW, I was told some time ago that HP/Agilent's 4-terminal standards had four paralleled SMD R's soldered across two copper plates for the lower values.

I found a ground pin late this afternoon, need to check if it takes solder. If so, I should have time to complete the coax probe tomorrow, unless work interferes 

Actually I have DIY SMA "calibrators" with cal. plane just at the back of the connector(you can use spare SMA connector and solder dut at pins). I'll try it and let you know.

Thanks. Yes, SMA has smaller loop and less inductance and will always be better than BNC in that department.

[R_G_B_]

I had a go with the analogue discovery.
 
I looked at both the impedance analyser and the network analyser
Function.  The impedance analyser sweeps up to 25Mhz from approximately 0Hz.

 The VNA has an option to sweep up to 50Mhz
And you can use an external signal source.

On the VNA signal source is good up to 5Mhz strange enough the output remains flat up to around 25Mhz. Maybe they are using compensation on the output amplitude as the output phase tracks the fall off in input  amplitude.
I had a difficult time getting the impedance analyser to work because the way it had to be calibrated using resistor values from 10, 100, 1K, 1M to calibrate for open and short compensation.

The resister I measure was a 0.05ohm current sense resistor. The noise you see is from the test setup and would not calibrate out.

[precaud]

The range being plotted is too wide to see what's going on. With a bottom range of 1mOhm, try a top range of 1 Ohm. Then we can see how it's handling the 50mOhm R.

Also, I would suggest get it working correctly at low frequencies (< 1MHz) first, and then open the measurement up to higher freqs.

What resistor value dis you end up using for the cal? I would think the lower one is better for 50mOhm DUT.

[R_G_B_]

I was comparing the analogue discovery with the ESR 70 capacitor meter which uses a test frequency of 100khz. The resistor I used for the cal of the analogue discovery was 10 ohm.
The ESR 70  measured the resistance of of the sense resistor as
0.02 ohms @ 100Khz. It's resolution is 0.01ohms.

[precaud]

I finished the rigid coax probe last night (see attached pics). A BNC female on the connector end. While soldering the ground pin on, the insulating material started oozing out. Ooops. I changed my approach and only worked one spot at a time instead of heating the whole thing up. It looks funky but turned out OK. To connect to the system, put a tee on the end, one lead goes to the source out, the other to the T input. Simple.

To test it, I soldered a 25mOhm SMD R onto pcb pads and cut the traces to them to isolate it. That way I can compare it to the 25mOhm R measurements made previously.

The measurement generally had the correct profile (relatively flat to 1MHz and inductive rise from there), but was way off (about 7mOhm too low), and with lots of spikey noise, especially at 1MHz and above. (I accidentally overwrote the file so I don't have a plot to show.) At any rate, the results are unacceptable, for three reasons: 1. I didn't get a good short across the test points, so the short measurement was too high, which in turn made the short compensation inaccurate. 2. I may have made the probe too long. (that is correctable, of course). 3. It is frankly a pain in the arse to use. Holding a long probe perfectly still with steady pressure on the pcb pads for the 23-second-long sweep is no fun and not easy. In a couple dozen tries I'm not sure if I got one entirely right.

So I will not pursue this probe any further at this time. A special shorting bar will have to be made, something that clamps on to the pins at the end and makes good contact. I would prefer having a probe that clamps or clips on so I don't have to be a hero and hold it perfectly still. So for now, I'll go back and see if I can make the BNC T fixture work better.

[nctnico]

For this kind of testing I like to solder a piece of coax to a board. Soldering an SMA onto the board is also an option.

[Pitrsek]

You need two probes. I'm sorry I did not mentioned it, I've never seen used the way you used it - it did not occurred to me to mention. You have two probes, one per channel. And the power plane is your "T splitter". I'll try to find a picture tomorrow.  If one is not fond of two probes(not enough free hands to do anything else), there is S21 probe from picotest which can be used as source of inspiration. Basically two probes side by side, with connected ends - ie. the T as at your probe tip. You trade a bit of inductance for more easier use.   

[precaud]

You need two probes. I'm sorry I did not mentioned it,

Now he tells me...     

You have two probes, one per channel. And the power plane is your "T splitter".

Understood. But hmmmm... last week I experimented using longer cables with the BNC tee, and they made no difference to the measurement. So this is a surprise. But at this point it doesn't matter, I am not excited about holding a probe on a pcb to make a measurement...

[Pitrsek]

Now he tells me...     

Yeah, I felt kinda stupid when I realized it...  Hey, I have a beer ready as an apology, if you are ever decide to visit Prague 

[precaud]

No worries, I was just having a little fun with it. I appreciated your help, period.

So I'm back working with the BNC-Tee-as-test-fixture. Referring to the plot in post #30, it bothered me that the phase had gone capacitive after the short compensation routine, while the impedance curve was clearly inductive. Obviously this can not be the case. I know the short comp math is good. It should leave whatever residual inductance is in the DUT. So what could cause this problem? Well, if the "ground reference" itself had exaggerated (more than actual) inductance at high freqs, the compensation math would not work as expected.

So I looked at the Tee with the ground wire, a single solid 20 gauge wire (see 1st pic). That could be the cause right there. So I soldered in three more wires to put the inductances in parallel (see 2nd pic), and ran the measurement again. (3rd pic). Voila. The original single-wire ground reference had enough inductance to throw the measurement off significantly. Compared to the plot in post #30 above, the impedance has been cut in half at 10MHz and the phase is now properly inductive and accurate, with 45º phase shift at the freq where the impedance is twice that of the level region, as it should be.

I'm becoming more impressed with this shunt-thru method. It's usable to 2MHz now. And so simple.

CORRECTION: I think I was pessimistic yesterday when saying this was giving good measurement to 2MHz. After looking at some graphs of Z measurements of single SMD current sense resistors of 30mOhm and below, they all have inductive rise to them starting a bit above above 1MHz. It's simply a function of the package dimensions (in this case, 1/4"). So I am more inclined to say that this technique is giving good info to 10MHz right now.

It would be nice to have a set of "reference" curves so one knows what Z/phase distribution to expect for a given lead/terminal spacing.

[R_G_B_]

I don't know if this will be helpful or interesting but i though i would post

[precaud]

Bob Pease was quite a character and always interesting.

[Pitrsek]

Hi,
I was finally able to run some more test. I started with my first semirigid fixture (Y.jpg), as you can see there is a severe braid error. Then I moved on to me new very short semirigid fixture. I've tested it with and without the mammoth CM coil. Final step was a setup similar to percaud's , but with SMA and CM transformer. One go with only trough calibration. The last one is with open,short, 50 and trough. As you can see, part of the mounting inductance got calibrated out, as well as some of braid error. To be honest the T setup turned out better than I expected.   

[Pitrsek]

I've used same coax cable for both runs with CM and without CM choke. I did not have a shorter one on hand for the without variant. Very short semi rigid would be off course preferable.  More pics + bonus/teaser pic for next time.
I'll be comparing test probes vs sma cables. Stay tuned 

[precaud]

Excellent tests, Pitrsek. Your small fixture is giving very good results, and with only a through cal. Do you think it is the "ground integrity" of the semi-rigid coax that makes the difference?

And of course the SMA Tee with compensations looks excellent. (At 10mOhm, I think you can skip the Open cal!)

With the Tee setup, when you do Through and Open/Short/Load sweeps. what is the difference (if any) for the Open and Through sweeps? Aren't they the same setup?

On the Anritsu, I run the Through with the fixture in place, but no DUT, which is the same as an Open. So to run it again as an Open would be redundant.

I look forward to your "probes vs Godzilla SMA cables" head-to-head.

The great value I see in using things like the Tee (vs soldering parts into a dedicated fixture) is efficiency. It enables you to solder the DUT in place, set it aside to cool down before testing it, and go on testing other devices while you wait. The question that remains for me is: how often will I use a setup that requires soldering the DUT in place? That is why I want some kind of drop-in fixture for my BNC Tee setup that will accept thru-hole radial parts without soldering.

[Pitrsek]

Small fixture - biggest thing going in favor is really low mounting inductance. You are usually able to solder SMD pat to it in a way that you do not add any(or very little) inductance. Short semi-rigid coax also helps with the braid error, but the biggest killer of braid error is the CM toroid.

You are absolutely correct, the the open cal could be skipped, and at that point open=trough, same setup. Did no occurred to me at the time.
If I was going to measure 0603 and smaller capacitors, I'd definitely go for soldering dut to the small fixture. I would say that key is the ratio of your fixture inductance to your DUT inductance. If your DUT is inductive as hell(IMHO all THT stuff compared to 0603), you don't really care for fixture inductance anyway.
For regular part testing, I use component fixtures that we have for Bode. I take out the small fixture only when I need to measure really low impedance/low inductance stuff - yes, soldering DUT is PITA.

Actually you can easily built fixture that you are dreaming about .
Small board, SMA on opposite edges, 50Ohm trace going in between. Mill-max pin receptacles for THT parts pins, in 50mil grid. See the picture.
I used mill-max receptacles for transformer when I was developing my last SMPS - great stuff, you design tight layout from the start and mess with transformer design all day long. No soldering necessary(a tip from Mr.Ridley). You can use multiple columns of the receptacles for different lead diameter. Also you can add series resistors if you'd like to shift the measurement range. And you could probably experiment with CM choke on the output as well, as I do not believe that for low MHz range you need CM choke with coaxial winding. This way you will be measuring device + mounting inductance. Depending on your use, it might be useful to do it on 4L board, to have board inductance as low as possible(ground plane in first inner layer). if you use same size resistor for calibration, you will calibrate out the DUT inductance and you will need to add it later based on the DUT geometry. Or you can provide 4 grounding millmax pins around the center(pattern of 5 on a dice) and create shorting plate. Do you have access to machinist shop? 

 

[precaud]

Small fixture - biggest thing going in favor is really low mounting inductance. You are usually able to solder SMD pat to it in a way that you do not add any(or very little) inductance. Short semi-rigid coax also helps with the braid error, but the biggest killer of braid error is the CM toroid.

I am fortunate (so far) to not have to worry about braid error...

If I was going to measure 0603 and smaller capacitors, I'd definitely go for soldering dut to the small fixture. I would say that key is the ratio of your fixture inductance to your DUT inductance.

That makes sense.

If your DUT is inductive as hell(IMHO all THT stuff compared to 0603), you don't really care for fixture inductance anyway.

I'm not so sure about that. The work I am preparing for involves modifications to existing boards, populated by a mix of SMD and THT parts. Cleaning up old messes, you might say. So I need to be able to characterize these THT parts in freq ranges where we know they do not behave well.

For regular part testing, I use component fixtures that we have for Bode. I take out the small fixture only when I need to measure really low impedance/low inductance stuff

I am hiding my jealousy  Can you post a close-up pic of the fixture?

Actually you can easily built fixture that you are dreaming about .
Small board, SMA on opposite edges, 50Ohm trace going in between. Mill-max pin receptacles for THT parts pins, in 50mil grid. See the picture.

Yes, I have some of the Mill-Max and thought of something like that, but I decided that the variable contact resistance of those connectors was not suitable for low-impedance stuff. And that's what I am concentrated on right now.

Above 1 Ohm is easy. You can do it without special fixtures, even. (Have you seen the YT video about Z measurement, soldering a radial 1nF cap with an axial 47 Ohm R and sweeping Z all the way to 65MHz? It looks so easy!) Low Z @ high freq is tough!

I used mill-max receptacles for transformer when I was developing my last SMPS - great stuff, you design tight layout from the start and mess with transformer design all day long. No soldering necessary(a tip from Mr.Ridley).

Yes, for those impedances, it's a viable setup.

And you could probably experiment with CM choke on the output as well, as I do not believe that for low MHz range you need CM choke with coaxial winding.

Fortunately I can skip the CM choke business. At least with the Anritsu...

Do you have access to machinist shop?

Only in my dreams 

Just for the helluvit, last night I put together a fixture based on the "typical" way of measuring impedance; measuring voltage on either side of a series resistor (see below). Not useful for SMT at all, but good for radial parts. I didn't take pics or save a plot (will do so next time and post), but the results were a bit worse than the injection xfmr method. Even with short signal paths and careful grounding, there is excessive inductance in the setup. Using 1 Ohm reference R, measuring the axial 25mOhm R used previous, without short compensation, it was accurate to maybe 30kHz. Short comp extended it to 100kHz or so. Not very useful.

Before I abandon it, I am going to try moving the reference R into the ground leg and measure the current there. It's a lower-Z node so the L should be lower and usable bandwidth higher.

[precaud]

Here are some measurements using the configuration shown in the previous post. The series R is 10 Ohms. I first tried it with a 1 Ohm R, but the results were not good at low or high frequencies, the 1 Ohm being too close in value to the impedance being measured. The test jig is set up for measuring radial-lead components, such as large electrolytic caps, with push-post terminals spaced at 8mm (5/16").

1st pic shows the setup, with a 20mOhm radial current sense resistor mounted.
Second one shows the 20mOhm R with and without short compensation. As you can see, it primarily cleans up some resonances above 1MHz, leaving a smooth inductive rise.
Third one shows three electrolytic capacitors, with the 20mOhm R for context. The ESR of the capacitors is accurately shown in their impedance at around 100kHz where the phase crosses zero.

The rising impedance slope above 100kHz is mostly due to the fixture spacing plus the length of the push posts. It does not calibrate out with the short compensation. Also of interest is that the 1000uF lytic actually measures lower than the 20mOhm resistor above 500kHz. The capacitor has a lower impedance up there than the resistor does!

So with its high self-inductance, this fixture is limited to 100kHz at 20mOhm, 40-50kHz at 10mOhm, and 1MHz at 100mOhm. Fairly similar to the "injection transformer" method. I think this configuration can be improved a little bit, with lower source impedance (not a 50 Ohm output) and shortening the posts. But I don't expect it to bring it anywhere near as good as the shunt-thru results.

Next is to move the current-sensing R into the ground leg.

[bson]

Isn't Z(f) unbiased in your fixture, and isn't this problematic with a polarized device like an electrolytic cap?

[precaud]

Isn't Z(f) unbiased in your fixture

Yes

and isn't this problematic with a polarized device like an electrolytic cap?

Not in my experience. I experimented with this in the mid-90's using an HP 4274A, and saw no change in impedance of a 'lytic between biased and unbiased states. That was up to 100kHz. Is it different at, say, 1MHz and above? That I can't answer. But I rather doubt it.

[bson]

Interesting...  That greatly simplifies the test setup! 

[precaud]

For testing 'lytics, yes it does.

Over the last couple days I've had some interesting email discourse with the designer of the Cleverscope. He has added significant capability for Z measurement (several topologies/techniques) in their software. And wrestled with the difficulty of measuring low-Z at high freqs. He put it in context this way:

1mm of pcb track between the test point and the DUT adds roughly 1nH of inductance, which at 1MHz has an impedance of 6mOhm. At 10MHz, is it 60mOhm.

I haven't confirmed if his figures are exact, but the point stands. And as I have let it sink in, methinks perhaps I shouldn't fret too much about board-level single-digit Z measurement above 1MHz. Other than identifying gross anomalies and resonances, which isn't as exacting.

[Pitrsek]

To keep inductance low, you need closely spaced power plane/ground plane pair on multi-layer board. Once you have that, life is way easier for layout as well. With regular 1,5mm double side board, the inductance is too high even with ground plane. Actually this would make nice demonstration as well,hm... I need to book the VNA for some tests  . 

[Hydron]

For testing 'lytics, yes it does.

Over the last couple days I've had some interesting email discourse with the designer of the Cleverscope. He has added significant capability for Z measurement (several topologies/techniques) in their software. And wrestled with the difficulty of measuring low-Z at high freqs. He put it in context this way:

1mm of pcb track between the test point and the DUT adds roughly 1nH of inductance, which at 1MHz has an impedance of 6mOhm. At 10MHz, is it 60mOhm.

I haven't confirmed if his figures are exact, but the point stands. And as I have let it sink in, methinks perhaps I shouldn't fret too much about board-level single-digit Z measurement above 1MHz. Other than identifying gross anomalies and resonances, which isn't as exacting.

A run through Saturn PCB Toolkit (highly recommended free software) shows this is in the right ballpark (see attached pic).

With ground planes you get much higher C/lower L than a thin track, but it's still tricky to measure anything down in the milliohms.

I have been using a Cleverscope for looking at <20mohm impedances in the 300+ kHz region and it's not easy to get to the point where you're completely sure whether stuff is coming from a fixture or the DUT - small test setup changes can make a fair difference. The isolated sig-gen and a DC-50Mhz current clamp (tek a6302, not cheap even decades old) lets you cheat braid error though!

[precaud]

A run through Saturn PCB Toolkit (highly recommended free software) shows this is in the right ballpark (see attached pic).

Thanks for that, will check out that program.

I have been using a Cleverscope for looking at <20mohm impedances in the 300+ kHz region and it's not easy to get to the point where you're completely sure whether stuff is coming from a fixture or the DUT - small test setup changes can make a fair difference.

Yeah, this seems to be the case regardless of the instrument used.

The isolated sig-gen and a DC-50Mhz current clamp (tek a6302, not cheap even decades old) lets you cheat braid error though!

Isn't the isolated output alone sufficient to kill the braid error loop? That isolated-output gen is one of the very attractive things about the CS, for sure. Which ADC option are you using? Bart is pretty convinced that the 10-bit (with averaging) is fine for Z/phase below 1MHz.

[bson]

I winged a quick and dirty test fixture using whatever I had on hand, and calculated the ESR of a brand new 100uF 35V Nichicon PW to ~50mohm.  Looks ballpark.  (Trace is unrelated, I was still experimenting and I had a 20dB pad on the A input.)  The bare wires are ground interconnects, in addition to solder bridges on the copper side.

Should be fine at least to a few MHz, then it rolls off instead of climbs indefinitely.  I might spin an actual board for this, but for electrolytics it seems to work fine as is.



VNA |Z| trace, wiggle due to 1s sweep time as I was setting things up.  The instrument isn't on GPIB right now or I'd pull the trace data and work on that instead.



Calculator:

import sympy as s
import math, sys

Ra  = 50.    # Input A impedance (50/1M)
Rdiv=9.989 # Divider resistor
Pad = 0.     # dB, on R

def db(a): return math.pow(10., a/10.)

Z=s.Symbol('Z', complex=True)

def zpar(a,b): return (a*b)/(a+b)

def zdiv(a,b): return b/(a+b)

# T relative to R:
# T=Tr*Ta; we read A/R so measure T/Tr = Ta
T=zdiv(Rdiv, zpar(Z, Ra))

val = db(float(sys.argv[1]) - Pad)

print s.solve(s.Eq(T, val), Z)

The "pad" is on R, to permit use of higher power levels, but that didn't seem to make any difference and I stopped doing it (just made cal harder).  The impact of R_A is minimal.  But, unfortunately, since it's a divider it's not possible to simply measure a know quantity like a 1ohmish resistor and then scale relative to it.  (Well it is, it's just the error grows as Z gets closer to R_div.)  So I short cal'ed it to A=R by shorting out R_div.  (Again, the first trace above is before all this, when I was just happy to see the tub shape. )

python calcz.py -23.1
[0.0492131962124952]

I also experimented with biasing, but clearly the supply I used (HP 66312) doesn't have sufficient bandwidth; not to mention the risk of blowing an input (though my HP 3577A claims on the front that its trip protection is safe to 25V DC and I was only biasing the - terminal by -0.2V for the -5dBm signal level, but still...).  If I make a board I might stick a biggish CLC pi filter on it for biasing with an external supply.  Maybe.  I'm still looking to convince myself there is no dielectric polarization effect skewing the response.

The phase response in general though (missing from the first image) nicely shows X_C=X_L and hence a purely resistive residual impedance.



Not sure I should post here since I'm confident I got lots wrong and some nutter is going to accuse me of being a Trump-sized liar.   But, regardless, it was a fun little Sunday afternoon project and the measurements don't look totally off!

[RoGeorge]

calculated the ESR of a brand new 100uF 35V Nichicon PW to ~50mohm

It would be interesting to measure the same capacitor, then compare the measured ESR with the calculated one.

[bson]

calculated the ESR of a brand new 100uF 35V Nichicon PW to ~50mohm

It would be interesting to measure the same capacitor, then compare the measured ESR with the calculated one.

My Keysight U1733C gives me ~120mohm at 10kHz - it's unable to obtain a stable reading at 100kHz.  The cap datasheet promises max 100mohm @ 100kHz for the 35V size.
So, looks plausible... 

[precaud]

Should be fine at least to a few MHz, then it rolls off instead of climbs indefinitely.  I might spin an actual board for this, but for electrolytics it seems to work fine as is.

Looks good for a Q&D fixture. Accuracy vs frequency will be a function of the impedance. Unlikely to hold up into the "several MHz" range, except above 100mOhm. At 100kHz you're fine at 50mOhm. Definitely a good idea to short your terminals to see what the baseline is. If you can store that trace data in Real/Imag format and subtract it from new measurements, you'll have an effective Short compensation. That will extend the usable BW and improve accuracy.

The "pad" is on R, to permit use of higher power levels, but that didn't seem to make any difference and I stopped doing it (just made cal harder).

Yes, it helps when your analyzer has good dynamic range

Not sure I should post here since I'm confident I got lots wrong and some nutter is going to accuse me of being a Trump-sized liar.   But, regardless, it was a fun little Sunday afternoon project and the measurements don't look totally off!

No accusers here. Keep sharing your results. You now have one of the largest ESR meters available!   

[precaud]

I wanted to see what could be done with the Shunt-Thru technique using an FFT approach. Cleverscope gets good results doing so up to 1MHz with their 10-bit 100MSa/sec ADC. It so happens I have a Lecroy 9430 laying about (2-channel 10-bit 100MSa/sec DSO). So I decided to write a driver to control it, gather the data, and merge it into the software I already have working. It took a few days but it's working fine. That is, the data is transferring accurately. Question is: is the data itself accurate?

The source signal is a "Chirp sweep", the same signal used by the HP 3562A DSA for its transfer function measurements. I have HP software that creates these chirp sweeps and can load them into an arb generator, in this case a Wavetek 75A. The general idea is to make the period of the chirp exactly equal to the time record length, so that it is a "periodic" waveform and plays nice with FFT algorithms without using any windowing. 100 time domain averages were taken to improve signal to noise. This resulted in 12- to 13-bit data from the 10-bit ADC. The gain of the test channel was raised (by 100X) to give good SNR with the low test channel's signal levels. (This is one advantage over most VNA's which use the same gain on both inputs). Combined, these two should put us into the same 100dB dynamic range territory that is typical of VNA's in this freq range. Open/Short compensation is part of my software and was used.

The first plot below is of the same two 25 mOhm currrent sense resistors that I used earlier with the MS420K; one SMD, one axial. The plot is linear freq scale, 0 to 1MHz, 2kHz per point. We know that the axial resistor should start to show inductive behavior around 100kHz. It doesn't. And the [hase rising well before that. It doesn't. The Z should be rising with increasing frequency all the way to 1MHz and beyond. Above 500kHz, it rolls off the other way. The trace is a bit noisy too.

I have included the sweep of the same two R's in the Anritsu: same software, same fixtures, etc. (The relevant curves are the yellow and green ones.) The different is the measurement technique; filtered swept sine vs FFT'ed sweep.

An FFT'ed swpt sine approach might fare better. But it would be very slow; this measurement took less than 10 seconds, including averaging.

Needless to say, these results do not inspire confidence in the broadband FFT approach.

[precaud]

Now this is interesting.  With the kind of error seen, I would expect to see poor dynamic linearity at low levels. So I tested the transmission characteristics (using an HP 355D attenuator) from 0dB to -90dB. Basically straight lines down to -60dB. Noise, and ground loop error, starts to show at -80dB, and becomes really strong at -90dB. But the response is still pretty flat. I'm not seeing anything even at -80dB commensurate with the HF rolloff seen in the previous post's measurement. I'm thinking that perhaps high input capacitance on the 9430's inputs is the culprit.

Nevertheless, not bad for a 20-year-old 10-bit 100MSa/sec ADC, eh?   

[Pitrsek]

Do you still have the raw data from measurements?If you plot VNA and lecroy in same graph, and flip the lecroy upwards, how much off it is?
I would be very wary about FFT without any windowing, are you certain that everything is sample perfect synchronized?
I'm no expert on signal processing, but I did get fair share of FFT rubbish from wrongly aligned data(lt spice).

[bson]

The rising phase clearly is inductive reactance.  I can't think of any reason for |Z| to fall off, other than if there's a calculation error...

Here's my math...

The DUT impedance is Z and its admittance Y
The input port (L=load) admittance Y_L = 1/50 - jwC_L.
The divider resistor is R and its conductance G = 1/R

Vout/Vin = T = G/(G + Y + Y_L)

=> Y = G/T - G - Y_L

=> Z = 1/(G/T - G - Y_L) = 1/(1/(RT) - 1/R - 1/50 + jwC)

|jwC| disappears with 15pF at 1MHz (0.94uS), which gives:

Z = 1/(1/(RT) - 1/R - 1/50)

The 1/50 term impacts perhaps 0.1% and is probably not significant given all the other sources of uncertainty... 
T of course is the transfer function, the difference of the complex FFTs for acquisitions A and B or obtained with VNA phase detecting receivers (tracking the source).
Or it could be the real transfer function off trace data, |T|=|A|/|B|, but then only |Z| can be determined.

The falling |Z| given the rising phase really smells like a sign error...

[precaud]

Do you still have the raw data from measurements?If you plot VNA and lecroy in same graph, and flip the lecroy upwards, how much off it is?

You and BSON are thinking the same. It looks like a sign error. That was my first thought too and I checked the math subroutines. They are unchanged. I have been using these same subroutines for 35 years. I have pretty high confidence in them  These are the same routines used with the Anritsu MS420 results posted earlier. In fact I started with the same program file. The only significant changes are the data input routines. The MS420 outputs T/R in polar form; the 9430 FFT gives R and T Real/Imag results in rectangular coordinates. Once that is converted and divided, the math is all the same.

I don't have the raw data, I typically don't integrate load/save routines until I get the measurement happening correctly and the data structure defined.

I would be very wary about FFT without any windowing, are you certain that everything is sample perfect synchronized?

Understood. This signal is specifically designed to be used without windowing functions. Windowing would act like a bandpass filter. I can change the arb generator readout clock and watch the spectrum changes. The technique is quite forgiving of sample sync, as long as the sequence completes within the window and is consistent for both channels. The routine that generates it even allows you to specify a number of zero fill at the end. IIRC, I didn't use any here.

The rising phase clearly is inductive reactance.  I can't think of any reason for |Z| to fall off, other than if there's a calculation error...
The falling |Z| given the rising phase really smells like a sign error...

The interesting thing is, though; both are WAY off. The axial R should have +45º phase shift at about 200kHz. That's the second division from the left. It's nowhere close.

Even the short measurement (not shown) has only a very small inductive rise, which is clearly in error.

I'll double-check the math subroutines, but they're pretty straighforward. For example: in HTBasic, the complex division is:
: Zref and Zdut are complex arrays
:  MAT Mag=ABS(Zdut)           (MAT operates on the entire array. Mag and Phase hold the Test signal)
:  MAT Phase=ARG(Zdut)        (ABS and ARG convert rectangular to polar)
:  MAT Mag2=ABS(Zref)
:  MAT Phase2=ARG(Zref)    (Mag2 & Phase2 hold the Ref signal)
:  scan Mag2 (the real part of the denominator) for zero values and replace them with something really small (I use 1.E-12, equivalent to 1 picoOhm), division by 0 is a fatal error...
:  MAT Mag=Mag/Mag2           (with data in polar, divide the magnitudes)
:  MAT Phase=Phase-Phase2   (and subtract the angles)
:  then convert back to rectangular and stuff it back into the Zdut array

I could just use Zdut=Zdut/Zref but I have found that the polar routine generates less "garbage", i.e. handles phase wrap better. There are fewer division operations.

That's it. Open and Thru measurements use this same routine.

[Wolfgang]

Isn't Z(f) unbiased in your fixture

Yes

and isn't this problematic with a polarized device like an electrolytic cap?

Not in my experience. I experimented with this in the mid-90's using an HP 4274A, and saw no change in impedance of a 'lytic between biased and unbiased states. That was up to 100kHz. Is it different at, say, 1MHz and above? That I can't answer. But I rather doubt it.

Hi,

I am sure it *is* different. When I made some blocking cap from 1000uf/50 electrolytics, ESR and ESL caused a significant rise from ca. 2MHz to several Ohms. At lower frequencies, the impedance was just the ERS of less than 100mOhms.
Measured on a Keysight E5061B-3L5 VNA/Impedance Analyzer.

[precaud]

I am sure *is* different. When I made some blocking cap from 1000uf/50 electrolytics, ESR and ESL caused a significant rise from ca. 2MHz to several Ohms. At lower frequencies, the impedance was just the ERS of less than 100mOhms.

Interesting. I have a good HP 4275A with DC Bias option now, so I'll run some comparisons and see if I get the same results as you do.

BTW, I *did* solve the issue I was asking about in the thread, I just forgot to come back and write about it.   

Now its been so long, I can't remember what it was...

[precaud]

I just measured an Elna 1000uF 50V lytic from 10kHz to 10MHz in the 4275A. The differences in C and ESR with and without +5V DC bias are insignificant. This is typical of past results.

[Wolfgang]

Maybe I try with another brand.

[precaud]

Maybe also we should remind ourselves, a 1000uF 'lytic is not a capacitor at 2MHz. Typical phase zero crossing (and most meaningful ESR value) is in the 100kHz range. So if the calculated ESR is dropping at 2MHz, it only means that the cap is close to being purely inductive. I saw that in the Elna cap. By 10MHz the phase is +87.5º and ESR drops to 1/3 its value at 100kHz, suggesting that the ESR and ESL is resonating with some parallel C up there...

[Wolfgang]

Maybe also we should remind ourselves, a 1000uF 'lytic is not a capacitor at 2MHz. Typical phase zero crossing (and most meaningful ESR value) is in the 100kHz range. So if the calculated ESR is dropping at 2MHz, it only means that the cap is close to being purely inductive. I saw that in the Elna cap. By 10MHz the phase is +87.5º and ESR drops to 1/3 its value at 100kHz, suggesting that the ESR and ESL is resonating with some parallel C up there...

Hi,

I tried around a bit and found the following facts:

- The influence of DC *is* cap brand dependent, as is the SRF. I dont know what causes this. The effect is about 10% of ESR and SRF shift.
- SRFs of electrolytic caps 1000uF/50V depend a *lot* on make. Some large ones from Panasonic were very good, some no-name had a SRF of just a few 100kHz.

https://electronicprojectsforfun.wordpress.com/rf-measurement-techniques/making-power-supply-measurements-with-a-vector-network-analyzer/

has the setup for a BNC environment and normal lab power supplies.

Some news:

What it could be is that one sort of caps sat at the shelf for quite a while and drew some forming current. The other was a none-name brand ordered by a large electronics reseller.

[precaud]

- The influence of DC *is* cap brand dependent, as is the SRF. I dont know what causes this. The effect is about 10% of ESR and SRF shift.
- SRFs of electrolytic caps 1000uF/50V depend a *lot* on make. Some large ones from Panasonic were very good, some no-name had a SRF of just a few 100kHz.

Could be. I can't say it doesn't happen; only that I've never seen it (in 30+ years of looking...)

[Wolfgang]

I will let them sit charged for a while and repeat the measurement. Then I'll know, I will post curves when done.

[bson]

Interesting. I have a good HP 4275A with DC Bias option now, so I'll run some comparisons and see if I get the same results as you do.

BTW, I *did* solve the issue I was asking about in the thread, I just forgot to come back and write about it.   

Now its been so long, I can't remember what it was...

One possible experiment is to simply reverse the cap.  If it has different impedance depending on polarization, well then it stands to reason that biasing it is guaranteed to affect the impedance as well.  I've been meaning to try this sometime but just never got around to it...

[precaud]

One possible experiment is to simply reverse the cap.  If it has different impedance depending on polarization, well then it stands to reason that biasing it is guaranteed to affect the impedance as well.  I've been meaning to try this sometime but just never got around to it...

We await your results...

[Wolfgang]

Picotest injectors - yes, they are pricey. If you go for a bundle it gets a bit more reasonable. There is one "feature" of the injector that i did not manage to wrap my head around. The current sense port is "feedforward", ie. the signal is not about what is there, but about what should be there. Current sense output works even with disconnected load. I have on idea why did they designed it this way.  On the other side, it's nice for calibration

10Hz and 1mOhm - it's quite tough. Depending on dynamic range of your VNA, you might probably need to build a preamp. Ideally with differential input, to get rid of the braid error(If you feel like building something like that - let me know, I'd be glad to help - I could use it too, same goes for the current load). If I may ask, why do you need to go so low?  There is usually not much happening...

Apart from two sma(really nice if you can accommodate the space, I use a lot of edge mounted ones), I used diy probes from semirigid coax and some needles(as recommended in "right the first time" book) - it sucked.  I have a new version with spring loaded gold plated tip in works, I'll post pictures when it's done.

I've attached pics of my setup - short semi rigid line +  common mode transformer. Of course it could be improved, ie. directly from coax to bnc(no adapters) and semi flexible coax for the CM transforem(+box). PN of the core is W518-03. Measured are 4x0.1R 0603 in parallel. I'll try to get something smaller, to show you the braid error with this setup. Actually this is quite a bit better than old setup with WE emc ferite core.

Got the same problem. Below a certain terminal voltage, the monitor output is still there, but no actual current is flowing.
There is not a line in the specs about this.