Author Topic: Affordable <200MHz PDN analysis / impedance measurement hardware?  (Read 7573 times)

0 Members and 2 Guests are viewing this topic.

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3518
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #50 on: November 13, 2024, 12:46:38 am »
I’ve reflected on this a good bit and would like to urge you as strongly as possible to attack the problem from first principles, a 4 channel DSO and a clean 50-50 square wave source.  And all that annoying algebra.  I got the crap beat out of me doing a 1st order approximation of a thermistor a few days ago.  Trivial algebra, but if you don’t practice, all skills atrophy and die.  Boy, did I ever get handed my head on a platter.  I finally sorted it, but it took hours to do what would have taken 5 minutes in HS.

It’s a significant effort, but your understanding of network analysis will expand greatly.  Based on my experience, the value of the education exceeds the price of a 4395A.

I have dozens of small bits of test gear I built when I was a grad student.  Don’t even remember what they are or how to hook them up.  Some succeeded, some failed and some succeeded but I and an EE at work failed to test it correctly and it was 30+ years before I realized I’d done it!  At that point I had a crazy high end lab and had learned a bit about impedance matching and what a *really* bad mismatch looks like.


Failure is your best friend.  It tells you what doesn’t work.

The directional bridge is the sum of a pair of sinusoids measured at different points.  If you are looking for impedance mismatch it’s handy.  As I understood the question it was directed at locating power supply impedance mismatches.  I may well have misunderstood the question.

Have Fun!
Reg

 

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 4120
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #51 on: November 13, 2024, 12:51:44 am »
..... I wish someone would make and standardise a miniature “kelvin” RF connector for this

You may find the Split-Kelvin Technique shown here useful for SMD components, certainly works well in the milli-Ω impedance levels with our LCR Bench Meters.

https://www.eevblog.com/forum/projects/different-type-lcr-smd-fixture/

Also, thanks for bringing up the AD2, we've been playing around with it this afternoon and it's quite good at resolving low Z components, especially if you pay attention to the DUT connections (here's where Kelvin Techniques really pay off). With the buffered output higher current drive signal, and the  additional differential low noise gain we can begin to understand the excellent results you've achieved with the AD2 low Z measurements :-+

As an interesting side note, some of the new "entry level" DSOs like the superb Siglent SDS800X HD with 500uV/Div analog scaling and true 12 bit ADCs almost beg for something along the these lines for "Impedance Measurement" augmentation with the built-in Bode Function. They have all the key ingredients already built-in for this to happen ::)

Best
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 4120
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #52 on: November 13, 2024, 01:08:59 am »
I’ve reflected on this a good bit and would like to urge you as strongly as possible to attack the problem from first principles, a 4 channel DSO and a clean 50-50 square wave source.  And all that annoying algebra.  I got the crap beat out of me doing a 1st order approximation of a thermistor a few days ago.  Trivial algebra, but if you don’t practice, all skills atrophy and die.  Boy, did I ever get handed my head on a platter.  I finally sorted it, but it took hours to do what would have taken 5 minutes in HS.

It’s a significant effort, but your understanding of network analysis will expand greatly.  Based on my experience, the value of the education exceeds the price of a 4395A.

I have dozens of small bits of test gear I built when I was a grad student.  Don’t even remember what they are or how to hook them up.  Some succeeded, some failed and some succeeded but I and an EE at work failed to test it correctly and it was 30+ years before I realized I’d done it!  At that point I had a crazy high end lab and had learned a bit about impedance matching and what a *really* bad mismatch looks like.


Failure is your best friend.  It tells you what doesn’t work.

The directional bridge is the sum of a pair of sinusoids measured at different points.  If you are looking for impedance mismatch it’s handy.  As I understood the question it was directed at locating power supply impedance mismatches.  I may well have misunderstood the question.

Have Fun!
Reg

Yes us older folks (I'm even older   :-\  ) learned to appreciate fundamentals early on, and still find it fascinating how electronic things work. From that perspective it's also fun to learn how old & modern instruments work and how to adapt their use.

When working where time is $, investing significant time may be unwise when one can just buy a specific instrument for the task. However, making custom adapters or fixtures is a great means of learning and putting fundamentals at work to help the measurement tasks at hand, and can usually fit within a retiree's budget :-+

Anyway, vaguely remember when a Grad Student or Post Doc would do anything for a brew, seems it's come full circle now ???

Best
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #53 on: November 13, 2024, 02:14:02 am »
Don’t underestimate a DSO with memory and a decent sig gen.   Suck the data into Octave for analysis.  The really hard part is a low frequency directional bridge.
Why would you want to use a directional bridge? For PDN you use a 2 port measurement AFAIK. Push AC current into the DUT from the transmitter port (generator) and measure the resulting AC voltage across the DUT.


I’ve reflected on this a good bit and would like to urge you as strongly as possible to attack the problem from first principles, a 4 channel DSO and a clean 50-50 square wave source.  And all that annoying algebra.  I got the crap beat out of me doing a 1st order approximation of a thermistor a few days ago.  Trivial algebra, but if you don’t practice, all skills atrophy and die.  Boy, did I ever get handed my head on a platter.  I finally sorted it, but it took hours to do what would have taken 5 minutes in HS.

It’s a significant effort, but your understanding of network analysis will expand greatly.  Based on my experience, the value of the education exceeds the price of a 4395A.

I have dozens of small bits of test gear I built when I was a grad student.  Don’t even remember what they are or how to hook them up.  Some succeeded, some failed and some succeeded but I and an EE at work failed to test it correctly and it was 30+ years before I realized I’d done it!  At that point I had a crazy high end lab and had learned a bit about impedance matching and what a *really* bad mismatch looks like.

Failure is your best friend.  It tells you what doesn’t work.

The directional bridge is the sum of a pair of sinusoids measured at different points.  If you are looking for impedance mismatch it’s handy.  As I understood the question it was directed at locating power supply impedance mismatches.  I may well have misunderstood the question.

Have Fun!
Reg

He seems to be asking why you need the bridge to measure S21.  I took it that you feel you can make these measurements with S11, possible not understanding the errors that go along with it for this particular application.

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #54 on: November 13, 2024, 02:34:53 am »
Ah, you want to measure it while on, not like with the VNAs.

?
You seem to infer that when using a VNA to measure a PDN, you would never power it up.  Obviously this is not correct. 

Attached plots are using the original NanoVNA to look at a PDN with the populated board both unpowered and powered.  Also shown is the same bare board. 

Of course, there are risks.  The blocking caps I made for these measurements will support these low frequencies.  They could pack a punch to the VNA.  I have not had damaged my original NanoVNA when making these measurements but it certainly could happen.    I did however unknowingly damage my H4 with what appeared to be an ESD event last time I used it.  I sacrificed my original NanoVNA to bring it back to life.  These VNAs do not have a lot of protection.   

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 8218
  • Country: nl
  • Current job: ATEX product design
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #55 on: November 13, 2024, 10:50:17 am »
Ah, you want to measure it while on, not like with the VNAs.

?
You seem to infer that when using a VNA to measure a PDN, you would never power it up.  Obviously this is not correct. 

Attached plots are using the original NanoVNA to look at a PDN with the populated board both unpowered and powered.  Also shown is the same bare board. 

Of course, there are risks.  The blocking caps I made for these measurements will support these low frequencies.  They could pack a punch to the VNA.  I have not had damaged my original NanoVNA when making these measurements but it certainly could happen.    I did however unknowingly damage my H4 with what appeared to be an ESD event last time I used it.  I sacrificed my original NanoVNA to bring it back to life.  These VNAs do not have a lot of protection.
VNAs can measure PDN with it powered up. It's a measurement that I never did myself, and not sure what's the setup to do. I can also say, that with the typical low cost VNAs we have access to, it's not useful to make the measurements powered up. The loop response of a power supply is going to be in the KHz region, where these VNAs are not measuring, or the measurements are not accurate enough to measure mOhm.
I quickly looked at the available info about PDN measurements.
https://www.keysight.com/zz/en/assets/7018-08474/application-notes/5989-5935.pdf

(might need registration to download)
In this document they publish ON, OFF and modelled impedance-frequency graphs for practical PDNs. For example Figure 28. The ON and OFF state measurements only differ in any meaningful way up to 5kHz.
Additionally, for the nano/liteVNA any noise present on the power supply would enter the ADC. From what I recall, mixers have frequency response that passes low frequency noise, and these VNAs don't have any bandpass filters on port 2.
So yeah, I wasn't very precise when suggesting VNAs only test OFF.
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #56 on: November 13, 2024, 11:11:05 am »
The directional bridge is the sum of a pair of sinusoids measured at different points.  If you are looking for impedance mismatch it’s handy.  As I understood the question it was directed at locating power supply impedance mismatches.  I may well have misunderstood the question.
PDN measurement is not for impedance matching but for measuring the impedance of a power supply to chips (like high speed processors, FPGA and memory) over a wide frequency range. From DC to tens or even hundreds of MHz. High speed logic chips typically have a specification in the design manuals for the required impedance of the power supply versus frequency. So this is something you'll want to verify.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #57 on: November 13, 2024, 12:33:14 pm »
The directional bridge is the sum of a pair of sinusoids measured at different points.  If you are looking for impedance mismatch it’s handy.  As I understood the question it was directed at locating power supply impedance mismatches.  I may well have misunderstood the question.
PDN measurement is not for impedance matching but for measuring the impedance of a power supply to chips (like high speed processors, FPGA and memory) over a wide frequency range. From DC to tens or even hundreds of MHz. High speed logic chips typically have a specification in the design manuals for the required impedance of the power supply versus frequency. So this is something you'll want to verify.

Impedance Matching Is Critical,  Steve Sandler

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #58 on: November 13, 2024, 01:00:17 pm »
VNAs can measure PDN with it powered up. It's a measurement that I never did myself, and not sure what's the setup to do. I can also say, that with the typical low cost VNAs we have access to, it's not useful to make the measurements powered up. The loop response of a power supply is going to be in the KHz region, where these VNAs are not measuring, or the measurements are not accurate enough to measure mOhm.

Ok, I think these two statements help me understand.   It seems like you have one particular problem in mind, or a narrow view of what problems  to look for.   There are some very good presentations on-line about proper design.

I quickly looked at the available info about PDN measurements.
https://www.keysight.com/zz/en/assets/7018-08474/application-notes/5989-5935.pdf

(might need registration to download)
In this document they publish ON, OFF and modelled impedance-frequency graphs for practical PDNs. For example Figure 28. The ON and OFF state measurements only differ in any meaningful way up to 5kHz.
Additionally, for the nano/liteVNA any noise present on the power supply would enter the ADC. From what I recall, mixers have frequency response that passes low frequency noise, and these VNAs don't have any bandpass filters on port 2.
So yeah, I wasn't very precise when suggesting VNAs only test OFF.

We are interested in impedance far beyond 5kHz.   

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 8218
  • Country: nl
  • Current job: ATEX product design
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #59 on: November 13, 2024, 01:04:20 pm »
VNAs can measure PDN with it powered up. It's a measurement that I never did myself, and not sure what's the setup to do. I can also say, that with the typical low cost VNAs we have access to, it's not useful to make the measurements powered up. The loop response of a power supply is going to be in the KHz region, where these VNAs are not measuring, or the measurements are not accurate enough to measure mOhm.

Ok, I think these two statements help me understand.   It seems like you have one particular problem in mind, or a narrow view of what problems  to look for.   There are some very good presentations on-line about proper design.

I quickly looked at the available info about PDN measurements.
https://www.keysight.com/zz/en/assets/7018-08474/application-notes/5989-5935.pdf

(might need registration to download)
In this document they publish ON, OFF and modelled impedance-frequency graphs for practical PDNs. For example Figure 28. The ON and OFF state measurements only differ in any meaningful way up to 5kHz.
Additionally, for the nano/liteVNA any noise present on the power supply would enter the ADC. From what I recall, mixers have frequency response that passes low frequency noise, and these VNAs don't have any bandpass filters on port 2.
So yeah, I wasn't very precise when suggesting VNAs only test OFF.

We are interested in impedance far beyond 5kHz.
I don't think you understand (with respect). Measuring a PDN with a VNA, it's going to produce the same graph whether it's ON or OFF above ~5KHz. You cannot measure with a nano/liteVNA below 5KHz. So why would you measure it when it's ON, it's just extra risk for your VNA, plus noise, without any benefits.

mod: Note to others, read below, this is not correct.
« Last Edit: November 13, 2024, 03:55:12 pm by tszaboo »
 
The following users thanked this post: nctnico

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #60 on: November 13, 2024, 01:09:19 pm »
The directional bridge is the sum of a pair of sinusoids measured at different points.  If you are looking for impedance mismatch it’s handy.  As I understood the question it was directed at locating power supply impedance mismatches.  I may well have misunderstood the question.
PDN measurement is not for impedance matching but for measuring the impedance of a power supply to chips (like high speed processors, FPGA and memory) over a wide frequency range. From DC to tens or even hundreds of MHz. High speed logic chips typically have a specification in the design manuals for the required impedance of the power supply versus frequency. So this is something you'll want to verify.

Impedance Matching Is Critical,  Steve Sandler
Still, using a directional bridge is not the right tool for the job as the impedances are far too low. For those interested, I think this page explains the goals of PDN design pretty well: https://www.signalintegrityjournal.com/blogs/12-fundamentals/post/2108-power-integrity-fundamentals-impedance-vs-frequency In real designs you won't get the curve perfectly flat. And it doesn't have to be; just as long as the impedance stays below the maximum for the circuit to work correctly.
« Last Edit: November 13, 2024, 01:44:38 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 8218
  • Country: nl
  • Current job: ATEX product design
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #61 on: November 13, 2024, 01:16:03 pm »
Let's see, this should explain it.
« Last Edit: November 13, 2024, 01:17:49 pm by tszaboo »
 

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #62 on: November 13, 2024, 03:21:00 pm »
VNAs can measure PDN with it powered up. It's a measurement that I never did myself, and not sure what's the setup to do. I can also say, that with the typical low cost VNAs we have access to, it's not useful to make the measurements powered up. The loop response of a power supply is going to be in the KHz region, where these VNAs are not measuring, or the measurements are not accurate enough to measure mOhm.

Ok, I think these two statements help me understand.   It seems like you have one particular problem in mind, or a narrow view of what problems  to look for.   There are some very good presentations on-line about proper design.

I quickly looked at the available info about PDN measurements.
https://www.keysight.com/zz/en/assets/7018-08474/application-notes/5989-5935.pdf

(might need registration to download)
In this document they publish ON, OFF and modelled impedance-frequency graphs for practical PDNs. For example Figure 28. The ON and OFF state measurements only differ in any meaningful way up to 5kHz.
Additionally, for the nano/liteVNA any noise present on the power supply would enter the ADC. From what I recall, mixers have frequency response that passes low frequency noise, and these VNAs don't have any bandpass filters on port 2.
So yeah, I wasn't very precise when suggesting VNAs only test OFF.

We are interested in impedance far beyond 5kHz.
I don't think you understand (with respect). Measuring a PDN with a VNA, it's going to produce the same graph whether it's ON or OFF above ~5KHz. You cannot measure with a nano/liteVNA below 5KHz. So why would you measure it when it's ON, it's just extra risk for your VNA, plus noise, without any benefits.

I refer you to my previous data where the on/off states converge at 400kHz.   
https://www.eevblog.com/forum/testgear/affordable-lt200mhz-pdn-analysis-hardware/msg5712085/#msg5712085

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 8218
  • Country: nl
  • Current job: ATEX product design
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #63 on: November 13, 2024, 03:53:13 pm »
I refer you to my previous data where the on/off states converge at 400kHz.   
https://www.eevblog.com/forum/testgear/affordable-lt200mhz-pdn-analysis-hardware/msg5712085/#msg5712085
Interesting. You are right. I totally missed that.
Do you know what is the DC-DC converter used on that board? It would make sense that Keysight's generic DC-DC module would have a slow control loop, and an FPGA board would have a fast transient response converter.
 

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #64 on: November 13, 2024, 03:55:35 pm »
The directional bridge is the sum of a pair of sinusoids measured at different points.  If you are looking for impedance mismatch it’s handy.  As I understood the question it was directed at locating power supply impedance mismatches.  I may well have misunderstood the question.
PDN measurement is not for impedance matching but for measuring the impedance of a power supply to chips (like high speed processors, FPGA and memory) over a wide frequency range. From DC to tens or even hundreds of MHz. High speed logic chips typically have a specification in the design manuals for the required impedance of the power supply versus frequency. So this is something you'll want to verify.

Impedance Matching Is Critical,  Steve Sandler
Still, using a directional bridge is not the right tool for the job as the impedances are far too low. For those interested, I think this page explains the goals of PDN design pretty well: https://www.signalintegrityjournal.com/blogs/12-fundamentals/post/2108-power-integrity-fundamentals-impedance-vs-frequency In real designs you won't get the curve perfectly flat. And it doesn't have to be; just as long as the impedance stays below the maximum for the circuit to work correctly.

Yes,  S21 is the better choice.   Having the lowest impedance may not be the best choice.  Discontinuities can cause unwanted radiation.   I had posted a few video links on PDN design and testing back when I was working on the software to measure them with the low cost VNAs.  You can find that here:
 
https://www.eevblog.com/forum/rf-microwave/impedance-measurement-with-vna-using-series-shuntseries-through-methods/msg3556953/#msg3556953

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #65 on: November 14, 2024, 02:34:50 am »
I don't think you understand (with respect). Measuring a PDN with a VNA, it's going to produce the same graph whether it's ON or OFF above ~5KHz. You cannot measure with a nano/liteVNA below 5KHz. So why would you measure it when it's ON, it's just extra risk for your VNA, plus noise, without any benefits.

I wanted to get some data before responding to this comment.   Shown is my NanoVNA H4 Rev 4.2.  Again, it was damaged and repaired.  During that time, I had added 5 capacitors which greatly improved the high frequency performance but not the low.  I suspect it would be on par with another virgin H4.   

Data was taken with the current released firmware which supports a lower limit of 1.6kHz.  VNA was swept from 1.6k to 10kHz, 401 points, 100Hz IFBW.     

Looking at S21, after normalize and terminated,  H4_1 showing dynamic range.  It's at least 70dB below 5kHz (not K or Kelvin BTW). 

I purchased a cheap 40dB attenuator for another test.  With a 70dB dynamic range we should be able to measure it without any problem.  Results can be seen in H4_2.   

Also showing the setup. 

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #66 on: November 14, 2024, 02:03:46 pm »
nano/liteVNA

Over the years, I have attempted to explain that below 300MHz, the original NanoVNA can often out perform the V2Plus/4 and LiteVNA/64.  See my previous post:
https://www.eevblog.com/forum/testgear/affordable-lt200mhz-pdn-analysis-hardware/msg5670833/#msg5670833

I have seen people post about their disappointment after they purchased one of these higher frequency VNAs and attempt to use them for narrow band measurements.  Even if you crank down the IFBW, take a day per sweep rather than seconds, it will still produce artifacts that we can't address.  I've always said the fix, get the original NanoVNA.     

OP was using an H4.  The video I linked used my original NanoVNA.  These are both much better suited for PDN measurements.  You bring up the Lite though.    Using the latest 3.2 hardware and released firmware, 100Hz IFBW, 801 points, normalized.  Swept from 1.6kHz to 50kHz with ports terminated.  Note the poor dynamic range compared with my H4.  All of my LiteVNA64s are also very non-linear once you get below about 25kHz.   I never use them this low so it's not been a problem for me.   

Hopefully this clears up some of the misconceptions about these low cost VNAs. 



Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 4120
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #67 on: November 15, 2024, 02:45:17 pm »
I ended up designing a little adapter board for the Analog Discovery series!

Here's the board measuring a low impedance capacitor array:

(Attachment Link)

(Attachment Link)

Here's a bunch of verification resistors (with varying inductances):

(Attachment Link)

I realized I didn't need to measure higher than ~25MHz - once I see the inductance slope at the end I'll known enough about what happens afterwards. The board has a 200mA output current buffer, and two fully differential opamps. This gives me both more current and enough resolution for all the low stuff. The bottom limit is about 100uΩ. :)

The flat tinned bars are spaceholders for tantalum capacitors, to be able to measure with a DC Bias as well. Works great so far.

Thanks for all the tips along the way!

Took a better look at your schematic and have a question.

The buffer amp U1 is powered from +-5V and the sense amps  U2 and U3 are from +-2.5V. How does the current sense differential amp U3 deal with the voltage across current sense resistor R2 which is directly driven by the output from U1?

Seems this might exceed the Common Mode range for the inputs to U3 which the Data Sheet shows +Vs -1.2V and -Vs -0.2V which would indicate an input range of +1.3 to -2.7V with +- 2.5V supplies?


Best
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3518
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #68 on: November 15, 2024, 06:13:10 pm »
He seems to be asking why you need the bridge to measure S21.  I took it that you feel you can make these measurements with S11, possible not understanding the errors that go along with it for this particular application.

I was only responding to "my VNA won't go to 10 Hz and I can't afford a 4395A". 

I got super lucky as shown by the screen shot. The seller didn't know what it was and wanted it gone. They were in the business computer business.   So far as I can tell it works flawlessly.  And I've used it a lot as a scalar network analyzer evaluating combinations of EMI filters.
2437269-0

VNAs have directional bridges in them. I was simply pointing out that you don't need a VNA. A signal generator, DSO and understanding the problem will solve it. 

To the frequency limits of the DSO,  analog log amplifier probes and some software are capable of providing full VNA capability from  DC to the limit of the DSO you currently have.  Any that will write data to an SD card will do.

I bought a cheap 12 bit 100 MHz Owon because it has very deep memory, 20 MPt IIRC.  Sucks as a scope but is great for data capture.  A perfect square wave with fast rise times will generate harmonics of very high order.  Choose the fundamental based on the sample interval in frequency and use a sub ns rise time 7400 series divide by two output.  Record it open circuit and then measure at various points.  The processing understanding is a one time effort and valuable skill to boot.

That said, my favorite tool for impedance discontinuity tests is a Tek 11801 and SD-24, but that is rather exotic and  hard to come by.  But it does show the precise (<0.1 mm) relative location of *every* impedance change.  Just not well suited to a use case where an active device  is at the other end of the line.

As should be clear from the various links, this is NOT a simple problem to solve even with a large sack of money to apply as a hammer. 

After rereading the OP a few times I should start with a DSO and a very small DIY  H field probe to isolate which segments of an actual PDN have issues as a first step.  Small loop of the smallest coax you can find.  Preferably semi-rigid.   You'll be able to directly detect what parts of the PCB have problems.

Then use an FFT to get a frequency domain view to give a sense of scale to the parasitics of concern.  Build the fixtures in the Keysight App note and measure them.

A differential DC log amp probe with a high input impedance should overcome the DSO dynamic range limitation.  In principle, one should be able to probe two nearby points on a trace to measure the current flow changes.  We all know that there is a voltage change as the current varies, but whether that's actually practical as a measurement technique for this use case is another story. I suspect that careful construction would make it viable as a punch through the solder mask probe.  But someone will need to try that out to know for sure.  Almost anything else has ground loop issues to be addressed.

Have Fun!
Reg
 

Offline inevitableavoidanceTopic starter

  • Contributor
  • Posts: 33
  • Country: nl
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #69 on: November 16, 2024, 10:07:06 am »
I ended up designing a little adapter board for the Analog Discovery series!

Here's the board measuring a low impedance capacitor array:

(Attachment Link)

(Attachment Link)

Here's a bunch of verification resistors (with varying inductances):

(Attachment Link)

I realized I didn't need to measure higher than ~25MHz - once I see the inductance slope at the end I'll known enough about what happens afterwards. The board has a 200mA output current buffer, and two fully differential opamps. This gives me both more current and enough resolution for all the low stuff. The bottom limit is about 100uΩ. :)

The flat tinned bars are spaceholders for tantalum capacitors, to be able to measure with a DC Bias as well. Works great so far.

Thanks for all the tips along the way!

Took a better look at your schematic and have a question.

The buffer amp U1 is powered from +-5V and the sense amps  U2 and U3 are from +-2.5V. How does the current sense differential amp U3 deal with the voltage across current sense resistor R2 which is directly driven by the output from U1?

Seems this might exceed the Common Mode range for the inputs to U3 which the Data Sheet shows +Vs -1.2V and -Vs -0.2V which would indicate an input range of +1.3 to -2.7V with +- 2.5V supplies?


Best

The board is developed around a maximum excitation voltage of 100mV. Besides the 10x gain on the PCB I have modified my AD2 for an additional 5x less attenuation internally, so am limited to an input voltage of ±140mV either way. :)
 
The following users thanked this post: egonotto, mawyatt

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #70 on: November 16, 2024, 05:04:11 pm »
I just bought a NanoVNA-H4, with a dynamic range of 70dB from 50kHz-300MHz, specified with a minimum frequency of 10khz.

After my friend flipper saw the low frequency data from my repaired/modified H4,  they sent me a brand new H4.   It is revision 4.3_MS with the MS synthesizer where mine has the Si.  The plan was to modify his H4 with the same capacitors and TVSs that I used on mine.  He also wants to use that VNA for measuring PDNs.  However, when I did a baseline check the low frequency performance is really poor.

Please do me a favor.  Provide the revision of your H4 and synthesizer type.  Set it to 401 points, 1kHz IFBW, with a freq range of 1.6k to 10kHz.  Short port 1&2 and normalize.  Open the ports and terminate.  Measure S21.  I want to know if you get anywhere near that 75-80dB I measure with mine.   

I mentioned I doubted the caps I added had anything to do with the low frequency performance, but I never took a baseline this low.   I wonder if that synthesizer is going to be a problem.   

They have their original NanoVNA which was purchased from the same source, same order as mine.   I am wondering if I will need to swap the that old Si part to get the H4 to perform down this low.   

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 4120
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #71 on: November 16, 2024, 06:02:01 pm »
The board is developed around a maximum excitation voltage of 100mV. Besides the 10x gain on the PCB I have modified my AD2 for an additional 5x less attenuation internally, so am limited to an input voltage of ±140mV either way. :)

Keeping the excitation at 100mV with a 1/2 Ω current sense places the DUT current at 200mA seems reasonable for the direct USB Powered AD2 and the OPA891 Buffer Amp limitation. With an external Power Supply the AD2 specs indicates a max of 700ma, but you would need a higher current capable buffer and the later voltage limit with the 10X gain, both of which increase complexity.

Reducing the AD2 attenuation by shorting the input resistor seems a good idea, but exposes the ADG612 to external signals without any series current limit resistor. Maybe a 10K parallel resistor might be better and provide some level of protection and could be compensated with the Scale Value.

Anyway, are the Gerber files available? We are considered building this adapter since our AD2 doesn't get utilized for anything now.

Best
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline inevitableavoidanceTopic starter

  • Contributor
  • Posts: 33
  • Country: nl
Re: Affordable <200MHz PDN analysis / impedance measurement hardware? [Solved!]
« Reply #72 on: November 17, 2024, 11:02:40 am »
Keeping the excitation at 100mV with a 1/2 Ω current sense places the DUT current at 200mA seems reasonable for the direct USB Powered AD2 and the OPA891 Buffer Amp limitation. With an external Power Supply the AD2 specs indicates a max of 700ma, but you would need a higher current capable buffer and the later voltage limit with the 10X gain, both of which increase complexity.

Reducing the AD2 attenuation by shorting the input resistor seems a good idea, but exposes the ADG612 to external signals without any series current limit resistor. Maybe a 10K parallel resistor might be better and provide some level of protection and could be compensated with the Scale Value.

Anyway, are the Gerber files available? We are considered building this adapter since our AD2 doesn't get utilized for anything now.

Best

The gerber files, BOM, and pick an place files are in the output folder in the github repository: https://github.com/remidavidkoster/Redako-PDNAnalyzer/tree/master/Project%20Outputs%20for%20PDN%20Analyzer

I'm not sure what DC bias caps are in the BOM - I ended up assembling 5 different versions of those with the best I could find for each voltage range. It doesn't show on the pictures, but I paralleled / stacked two capacitors together for the excite signal current path.

I agree some current limiting with indeed be best, and simple soldering another resistor over the current 820k ones is easy enough.

Please keep me posted if you end up making some - happy to see someone else get a use of the design. :)
 

Offline inevitableavoidanceTopic starter

  • Contributor
  • Posts: 33
  • Country: nl
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #73 on: November 17, 2024, 11:22:40 am »
After my friend flipper saw the low frequency data from my repaired/modified H4,  they sent me a brand new H4.   It is revision 4.3_MS with the MS synthesizer where mine has the Si.  The plan was to modify his H4 with the same capacitors and TVSs that I used on mine.  He also wants to use that VNA for measuring PDNs.  However, when I did a baseline check the low frequency performance is really poor.

Please do me a favor.  Provide the revision of your H4 and synthesizer type.  Set it to 401 points, 1kHz IFBW, with a freq range of 1.6k to 10kHz.  Short port 1&2 and normalize.  Open the ports and terminate.  Measure S21.  I want to know if you get anywhere near that 75-80dB I measure with mine.   

I mentioned I doubted the caps I added had anything to do with the low frequency performance, but I never took a baseline this low.   I wonder if that synthesizer is going to be a problem.   

They have their original NanoVNA which was purchased from the same source, same order as mine.   I am wondering if I will need to swap the that old Si part to get the H4 to perform down this low.

I'm getting nowhere near that figure - 10dB per division:



Where do I find the revision and synthesizer?
« Last Edit: November 17, 2024, 11:24:26 am by inevitableavoidance »
 

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 12280
  • Country: us
Re: Affordable <200MHz PDN analysis / impedance measurement hardware?
« Reply #74 on: November 17, 2024, 02:59:50 pm »
I'm getting nowhere near that figure - 10dB per division:
...
Where do I find the revision and synthesizer?

The back of the unit should have the hardware revision and possibly an _MS.   You can also select Config, Expert Settings, More.  At the top is should show the synthesizer selected (MS5351 or Si5351).   

You can try to set the IFBW to 30Hz which should at least remove the pattern you are seeing. 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf