EEVblog #1104 - Omicron Labs Bode 100 Teardown

[precaud]

So if you look up the AP Instruments Model 102B Manual (I found one here: http://www.apinstruments.com/files/102Bman.pdf),  Page 54 (PDF numbering) / 50 (document numbering) shows your typical FRA setup used. The VNA calculates the frequency response from the ratio of its inputs Channel A and Channel B. So the Source impedance and transformer phase / amplitude cancels out from the equation. No calibration required, just a good matching between Channel A and B (which one expects from a decent VNA).

Within limits of course, since one needs a minimum amplitude of source signal coupled through the transformer to get a useful signal above the noise floor.

OK, I see, for that topology the xfmr is literally doing nothing more than "injecting" the signal. So the xfmr requirements are relaxed, compared to what they are when using it for impedance measurement.

[rx8pilot]

How did you calibrate?

Huh? You want me to explain the 102B's cal procedure?

I know the question sounds fundamental - and it is. This is a topic that I am new to and I have very little confidence in the results I am getting. I have been trying to better understand the network analyzer (in a general sense, not model specific), inductor measurements and analysis, and frequency response analysis of SMPS control loops. One of the things I am trying to figure out is essentially a 'sanity check' to make sure the results I am getting are reasonably accurate.

The part that I worry about is getting results that are close - but skewed enough that I end up chasing my tail. I don't have a solid intuition of how sensitive the measurements can be, where I can get false/skewed measurements and such. I think back to when I was learning to use an oscilloscope to measure delicate signals and the 10x passive probe had a huge ground clip connected directly to the power supply terminal. Everything looked broken, of course, and I spent a long time trying to add capacitors to reduce phantom noise. At some point - I learned how to use the instrument properly and of course, realized that there was never a problem - the measurements themselves were the problem.

Perhaps gathering the manuals for the AP VNA, Bode100, E5061-LF, etc would help me gain some confidence in how I am going about the process. Repeating some of the measurements in this thread should also help since I have the Jensen ISO-MAX VB-1BB that has been measured by a couple of people (comparison)

[capt bullshot]

rx8pilot:
For building up some confidence in the cal and measurements procedures, I'd recommend to rebuild the HP35676A test set:
http://hparchive.com/Manuals/HP-35676-SCHEMATIC.pdf

No need for all the fancy metalwork, and don't worry about the unknown capacitor, it still works built on a piece of SMT protoboard using some MiniMelf resistors (some of the series connected to get near the original value).

Once calibrated using the full cal procedure (again, no need for a fancy cal standards set, a wire and a 50 Ohm resistor works, you'll get some ripple at the 200MHz end), then switching the 3577A to impedance display (choose "F4" as the input to display , it's written somewhere in the manual), you can get impedance sweeps of known components (start with resistors in the 10 Ohm to k Ohm range) and get familiar with the way it works. Measure some inductors and capacitors, read the imaginary part at some frequencies and do the math to verify your reading. Inductors are quite fancy regarding their impedance over frequency, many of them are inductors at rather low frequencies only.

There's also somewhere a manual on the net (for this or the HP35677 test set) that has a better description of how to measure impedance using the 3577A than the 3577A manual.

Edit: Found it, here it is: https://cb.wunderkis.de/wk-pub/Keysight%2035676A%20Data%20Sheet.pdf

Edit 2: Don't know what you mean by "skewed", the manual has some tips regarding the RBW setting: Reverse the sweep direction and watch if the graph shifts. If it does, your sweep speed is to fast for the chosen RBW.

Edit 3: "F4" is impedance for a 50 Ohm system and calibration, "F5" is for 75 Ohm

[DualTriode]

Hello,

This is looking for the Sweet Spot.

Most of these injection transformers are set up for only a few ma of current before they saturate.
This transformer is wound on a VAC 250F core. This core has permeability that is much reduced from the VAC 500F core used in the B-WIT-100 transformer and will tolerate much more current before the onset of saturation.

The tradeoff is reduced band width, mostly trimmed off the LF end of the scale.

My impression is if you want to focus in on the LF use a different transformer. If you want to put the secondary of this transformer in series with the power supply to your audio amplifier to test PSRR this just might be the transformer to use.

This transformer is wound on a VACUUMSCHMELZE T60006-L2040-W964-02 250F core. There are 40 turns of CAT 6 plenum solid twisted pair.
 
@rx8pilot,

Thanks for sharing your approach to learning this stuff. You can look at all the various equipment user’s manuals, they don’t take much space to explain the interworking of stuff. The best single source I can recommend today is “Power Integrity”, Measuring, Optimizing, and Troubleshooting Power Related Parameters in Electronics Systems by Steven M. Sandler. https://www.amazon.com/Power-Integrity-Optimizing-Troubleshooting-Electronics/dp/0071830995/ref=sr_1_1?s=books&ie=UTF8&qid=1534877929&sr=1-1&keywords=power+integrity . Cheep at$16.00

DT 

[capt bullshot]

From the frequency / phase response, I suppose you've used a rather low termination resistor for this transformer? This core has a quite low A_L (2.3uH @ 10kHz), so I'd expect a higher low frequency cutoff if it was terminated to 50 Ohm.
For comparability, you should specify its value.

[DualTriode]

Thank you.

For this plot the resistor is the 50Ohm resistor internal to the Bode 100.
Same as Dave used in the video.

DT

[capt bullshot]

Okay, thanks, still wondering. Doesn't match my experience with my home rolled transformers and the 3577A. I vaguely remember the Omicron having a "low impedance mode" for its output. What was the used source impedance? My 3577A has 50 Ohm output impedance, in case the Bode 100 was switched to something else, this might explain the difference.

Edit: did you measure "source" to "transformer output" ratio (including the source impedance into the measurement) or "transformer input" to "transformer output" (using two inputs of the VNA, excluding the source impedance)? I believe this is the reason for our different results, as I've done all the measurements the first way (using only one input).

BTW: didn't watch the whole video (sorry Dave), I do prefer reading stuff over watching videos

[DualTriode]

Oops!

capt bullshot,

You were correct, I did not have the 50 Ohm output switched on. I switched on the 50 Ohm output and ran the plot again, this time down to 1 Hz.

The LF response was not noticeably changed. The HF response was slightly improved.

I will lug the 4395A VNA to the bench and run the plots on that VNA and post here.

Thanks DT

[rx8pilot]

@rx8pilot,

Thanks for sharing your approach to learning this stuff. You can look at all the various equipment user’s manuals, they don’t take much space to explain the interworking of stuff. The best single source I can recommend today is “Power Integrity”, Measuring, Optimizing, and Troubleshooting Power Related Parameters in Electronics Systems by Steven M. Sandler.

I actually met him at a trade show and purchased the book right there. It has been an excellent resource! There are, however, plenty of details that are hard for me to absorb without sitting at the bench and doing the work. Now that I have the gear, I have run out of excuses and just need to dive into the deep end. Sandlers book and all the other resources I have gathered are just now starting to make sense.

rx8pilot:
For building up some confidence in the cal and measurements procedures, I'd recommend to rebuild the HP35676A test set:
http://hparchive.com/Manuals/HP-35676-SCHEMATIC.pdf

No need for all the fancy metalwork, and don't worry about the unknown capacitor, it still works built on a piece of SMT protoboard using some MiniMelf resistors (some of the series connected to get near the original value).

Once calibrated using the full cal procedure (again, no need for a fancy cal standards set, a wire and a 50 Ohm resistor works, you'll get some ripple at the 200MHz end), then switching the 3577A to impedance display (choose "F4" as the input to display , it's written somewhere in the manual), you can get impedance sweeps of known components (start with resistors in the 10 Ohm to k Ohm range) and get familiar with the way it works. Measure some inductors and capacitors, read the imaginary part at some frequencies and do the math to verify your reading. Inductors are quite fancy regarding their impedance over frequency, many of them are inductors at rather low frequencies only.

There's also somewhere a manual on the net (for this or the HP35677 test set) that has a better description of how to measure impedance using the 3577A than the 3577A manual.

Edit: Found it, here it is: https://cb.wunderkis.de/wk-pub/Keysight%2035676A%20Data%20Sheet.pdf

Edit 2: Don't know what you mean by "skewed", the manual has some tips regarding the RBW setting: Reverse the sweep direction and watch if the graph shifts. If it does, your sweep speed is to fast for the chosen RBW.

Edit 3: "F4" is impedance for a 50 Ohm system and calibration, "F5" is for 75 Ohm

Thank you....excellent information!

[DualTriode]

rx8pilot,

It is easy to make a mistake. I suggest that you check all the connections with a volt meter to confirm that the voltages are safe for your expensive test equipment. It is easy to vent the magic smoke.

Thanks DT

[rx8pilot]

rx8pilot,

It is easy to make a mistake. I suggest that you check all the connections with a volt meter to confirm that the voltages are safe for your expensive test equipment. It is easy to vent the magic smoke.

Thanks DT

The low-frequency operation makes AC coupling a challenge - I hope I can avoid blowing up my box. The HP 3577A has fairly decent overvoltage protection, but obviously, it has limits.

Measuring passive components is pretty safe, but doing FRA introduces a considerable risk of smoke. Not sure what the Bode 100 has in terms of input protection other than user selectable attenuators.

[precaud]

I know the question sounds fundamental - and it is. This is a topic that I am new to and I have very little confidence in the results I am getting. I have been trying to better understand the network analyzer (in a general sense, not model specific), inductor measurements and analysis, and frequency response analysis of SMPS control loops. One of the things I am trying to figure out is essentially a 'sanity check' to make sure the results I am getting are reasonably accurate.

Ah, I totally appreciate that. Previous advice given by others is very good. My uses tend more toward FRA than VNA, so I'll only add, I find it extremely useful to have a set of components (R's, C's, L's) that cover a wide range of values, which I have measured on other devices to compare my results to. "Confidence by Consensus".

Definitely put together a set of DC blocking cap fixtures for your 3577A. It's not a chance worth taking.

Perhaps gathering the manuals for the AP VNA, Bode100, E5061-LF, etc would help me gain some confidence in how I am going about the process. Repeating some of the measurements in this thread should also help since I have the Jensen ISO-MAX VB-1BB that has been measured by a couple of people (comparison)

I'd skip the stuff in AP's manuals about impedance measurement, it's pretty unsophisticated. They only cover the "Series-R" topology and do nothing to measure and remove residuals, which is necessary for good results.

[rx8pilot]

What kind of LF response do you target with your DC blocks?

Short and misplld from my mobile......

[precaud]

The 3577A low freq limit is 5Hz, so something below that. 100nF into 1MOhm gives -3dB a tad over 3 Hz. Scale to taste...

[rx8pilot]

Oh, of course! I was fixated on 50 Ohm which is a totally different thing, lol.

For FRA, the receivers will (likely) always be 1M. That is when the system is in the danger zone..... connected to all kinds of power supplies where DC blocks are needed.

Does anyone know if the Bode 100 has AC coupled inputs? It did not look like it after scanning the brochure.

Short and misplld from my mobile......

[precaud]

Does anyone know if the Bode 100 has AC coupled inputs? It did not look like it after scanning the brochure.

I'm pretty sure it does. They demo the "shunt-thru" impedance technique, which requires 50 Ohm term on both channels, without adding an external term.

[precaud]

Perhaps gathering the manuals for the AP VNA, Bode100, E5061-LF, etc would help me gain some confidence in how I am going about the process. Repeating some of the measurements in this thread should also help since I have the Jensen ISO-MAX VB-1BB that has been measured by a couple of people (comparison)

This app note from AP is more useful than their manual:

http://u.dianyuan.com/upload/space/2010/10/29/1288318638-361233.pdf

The AP200 shown is the same as my 102B except it interfaces to the computer via a parallel port.

[Wolfgang]

I wonder why Dave was so impressed by this transformer.

It's a heck of a lot better than some DIY ones I've used.

Hi Dave,

I was wondering what is so special with the Omicron B-WIT100 from your teardown transformer, and I was able to duplicate the transfer characteristics using two stacked cores from Vacuumschmelze (VAC), PN T60006-L2030-W514. The result are almost identical (all was measured on a Keysight E5061B-3L5 I now have on a test loan).

The big difference is the cost: My transformer is about 60 to 70€ including the box, the Omicron is about 500€ incl. VAT.

All the nittigritti is documented here (at the bottom):

https://electronicprojectsforfun.wordpress.com/injection-transformers/

[rx8pilot]

I was wondering what is so special with the Omicron B-WIT100 from your teardown transformer, and I was able to duplicate the transfer characteristics using two stacked cores from Vacuumschmelze (VA), PN T60006-L2030-W514. The result are almost identical (all was measured on a Keysight E5061B-3L5 in now have on a test loan).

The big difference is the cost: My transformer is about 60 to 70€ including the box, the Omicron is about 500€ incl. VAT.

All the nittigritti is documented here (at the bottom):

https://electronicprojectsforfun.wordpress.com/injection-transformers/

I am VERY thankful for this write up @Wolfgang!  .
It so happens that I have the exact same core sitting on my bench and it is about to become my new injection transformer.
Not sure why this is so interesting and fun.....but it is.

Off to do some winding now.

[Wolfgang]

Do you want the STL for the holder ?

[precaud]

All the nittigritti is documented here (at the bottom):

https://electronicprojectsforfun.wordpress.com/injection-transformers/

One heckuva job, Wolfgang!   

[rx8pilot]

Do you want the STL for the holder ?

For sure, I don't have a 3D printer but I can CNC machine it from Delrin.

I will probably machine a case for it and give it a fun name: NVT-1

Nude Virgin Transformer - 1

[Wolfgang]

... for all interested: I uploaded the STL file of the holder in a ZIP and the webpage has the link in it now.

Have fun !
  Wolfgang

[rx8pilot]

Thank you so much, this is a great project.

Short and misplld from my mobile......

[Wolfgang]

Do you want the STL for the holder ?

For sure, I don't have a 3D printer but I can CNC machine it from Delrin.

I will probably machine a case for it and give it a fun name: NVT-1

Nude Virgin Transformer - 1

NVT1 - A Good name, why not. I'll make a label for it tomorrow.

Delrin is the same as POM? I like it too but the electrostatic chips when milling drive me nuts. There is no place where they dont stick to.