EEVblog #1104 - Omicron Labs Bode 100 Teardown

[DutchGert]

Hmm, am I the only one that is not impressed at all by the PCB layout job done?

The DC-DC convertors are done sloppy (big loops, small traces etc).
TH electrolitic caps everywhere, what about SMT MLCC's or Polymer jobbies?
Power is routed with thick traces, what about a nice solid power plane(s)?

[dardosordi]

Im sure they regret not having potted the transformer now...

[PA4TIM]

I just thinking, the output and input of the instrument are single ended and both connected to the same ground.  The transformer is balanced. I have no experience with this kind a measurements but if you use it to isolate the DUT from the analyser then you need two of them for 2 port measurements. One on the input and one on the output.
 I think you must connect it between two baluns to measure how it behaves for real isolation.

[rx8pilot]

Im sure they regret not having potted the transformer now...

I doubt it....their customer base is probably very unlikely to bother making their own transformer. As much as I like to save money, spending a few days and at least a couple hundred $$ on failed attempts is a terrible way to 'save' $500.

For those looking for a bargain price, they were never going to be a viable customer anyway.

[rx8pilot]

Im sure they regret not having potted the transformer now...

I doubt it....their customer base is probably very unlikely to bother making their own transformer. As much as I like to save money, spending a few days and at least a couple hundred $$ on failed attempts is a terrible way to 'save' $500.

Those looking for a bargain price were never going to be a viable customer anyway.

[Wolfram]

That's not ferrite and it's definitely not powdered iron, that's got to be nanocrystalline and nothing else.   Most likely Vacuumschmeltze, probably https://www.mouser.com/ProductDetail/Vacuumschmelze/T60006-L2040-W453?qs=sGAEpiMZZMs2JV%252bnT%2fvX8PvC43ppqs%252bksq4V5kp6Ay4%3d or similar.

  Oh s***t, it's blue! http://allegro.pl/rdzen-magnetec-m-083-rtn-40x25x15-nanoperm-i7430998721.html I don't know, I can't find the actual one, this looks like the closest, but still could work. I couldn't find a way to source the cores, always hard here in Argentina, if someone can source them would be cool to see some tests.

Never seen any in blue, I wonder if they got them as special order, or if there's another mfg I don't know about.  IIRC, it was just VAC and HMG (former Metglas) doing rapid-quench materials, but maybe there's new ones from China I don't know about?

Checking, I see very little on Ali Express, so probably not.

At PCIM this year, Tamura was presenting nanocrystalline CMCs, so they might also have the capability.

That's not ferrite and it's definitely not powdered iron, that's got to be nanocrystalline and nothing else.   Most likely Vacuumschmeltze, probably

That's for sure, the banner picture on the ones I posted had blue and brown cores, the ones I found for sale with the part numbers I was looking shown blue pictures but that's the only reference I got and I can't even read the web in that language!

I can confirm that the Vacuumschmelze nanocrystalline cores are the same color as the core in the EEVBlog video.

[AndersJ]

When the transformer secondary is connected to the DUT,
it will most likely cause a short circuit.

Why is there no coupling cap in the box?

[Wolfram]

When the transformer secondary is connected to the DUT,
it will most likely cause a short circuit.

Why is there no coupling cap in the box?

An injection transformer like this is not connected to the output of a power supply, it is inserted in series with the feedback loop (across a resistor). See the drawing in the user manual for clarification: https://www.omicron-lab.com/fileadmin/assets/Bode_100/Accessories/B-WIT_100/B-WIT-B-LFT-User-Manual-V1.1.pdf

[Wolfgang]

It goes into the cut open feedback loop, not parallel to the load. What you are thinking of is a current injector. The Omicron people also have those.

[T3sl4co1l]

I just thinking, the output and input of the instrument are single ended and both connected to the same ground.  The transformer is balanced. I have no experience with this kind a measurements but if you use it to isolate the DUT from the analyser then you need two of them for 2 port measurements. One on the input and one on the output.
 I think you must connect it between two baluns to measure how it behaves for real isolation.

Quite right, you only get good CMRR when the transformer is used in balanced operation, and the source and load are understood to have perfect CMRR.

Unbalanced, you get full capacitance (at medium frequencies, until it becomes obviously transmission line based in the HF limit).

The trick is to put a CMC on both sides of the transformer, to restore CMRR in the MF to HF limit.  Note that the CMC inductance resonates with the isolation capacitance, so it should be chosen (lossy material, or R+C in parallel) such that the resonance is well damped.  CMCs can be stacked (given the same consideration) to extend the CMRR arbitrarily.

Tim

[Wolfgang]

Theoretically, this is all correct. In practice, I personally have never seen any documentation or video using two isolation transformers or common mode chokes.
Do you have a working example of this somewhere on the web ?

[JS]

We have all we need to reverse engineer that transformer! Construction method, number of turns, impedance analysis, freq response... It's just matter of finding the right core material, with the impedance and number of turns permeability could be estimated... Would someone sell cores out of that material?

If you could sell a $100 unit with the same performance, you'd sell a truck load of them.
Would be interesting to try and duplicate their design and what performance it has with just some random core.

Forgot to answe this, it will result in a poor LF reaponse, depending on the core permeability, bassicaly the prinary inductance and the 50Ω makes a HPF which limits that, but even over that freq there is some distortion due to magnetization hysteresis which tapers off to the transformers higher freqs.

Please dave, when you find yourself on the lab with a screw driver and some length measurement device can you take a look at the core? To aim for a similar freq range that's kind of a big deal and I can't start making prototypes to shoot for one, sourcing those cores in Argentina is kind of tricky, as almost any component other than a 2n3904...

JS

[T3sl4co1l]

Theoretically, this is all correct. In practice, I personally have never seen any documentation or video using two isolation transformers or common mode chokes.
Do you have a working example of this somewhere on the web ?

In fact it is out there in the wild -- you're probably using it right now, though you may not have seen or noticed it!  Example: https://www.mouser.com/datasheet/2/336/H329-1199189.pdf

Tim

[envisionelec]

I reverse engineered the entire design and ran a cost analysis on a 10 piece basis.

The BOM cost is 110.84 including nominal labor/machining fees.

That's PTFE coated wire - 1USD a foot and there is about 24USD worth of wire on that core. I'll build one and characterize it.

[rx8pilot]

I reverse engineered the entire design and ran a cost analysis on a 10 piece basis.

The BOM cost is 110.84 including nominal labor/machining fees.

That's PTFE coated wire - 1USD a foot and there is about 24USD worth of wire on that core. I'll build one and characterize it.

If that is correct.....the $500 price tag is a bargain for such a specialty/low volume piece of gear.

[Wolfgang]

Well, I never saw them in a *measurement* application like Bode plotting or LF VNA. Any use there I missed ?

[JS]

I reverse engineered the entire design and ran a cost analysis on a 10 piece basis.

The BOM cost is 110.84 including nominal labor/machining fees.

That's PTFE coated wire - 1USD a foot and there is about 24USD worth of wire on that core. I'll build one and characterize it.

What tells you it's PTFE? I see some deformation of the insulatoe on the BNC solder, that shouldn't happen with PTFE...

For running small batches I could machine those my self and all labor so all that would come to my pocket but I'm trying to find the customs fees for importing the cores and then exporting the devices, that could ruin my budget to sell them at any reasonable price...

I intend to order a few cores to test and see what happens, worst case I end with an injection transforner for myself and a few nice cores for other applications, I probably will find them useful in audio, but before they arrive I could make some with any cheap core and look at the HF response where permeability doesn't play a big role. LF response will be anything but nice, I know that.

JS

[Wolfgang]

Of course you could try to reproduce the omicron injection transformer 1:1.

An alternative could be to use a stock ISDN or current transformer and to add a damping/compensation network to flatten out amplitude and phase response. Most swept generators have by far enough output for this approach, and the injected stimulus voltages are normally very small in order not to overdrive the DUT. When I tried this and if you apply the same criteria (5% phase error), results are not so vastly different from what omicron offers (e.g. 100Hz to a few 100kHz). The omicron claim of 1Hz to 10MHz is pure marketing; at band corners they are 15dB down and have phase errors of 70°. If you apply 1dB amplitude and 5°C phase error criteria, the range for them is ca. 50Hz to a few 100kHz (information from the datasheet of the B-WIT100). The big difference is that you have a 40dB power attenuation for the homebrew approach, which does not hurt at all in most cases.

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

[rx8pilot]

The omicron claim of 1Hz to 10MHz is pure marketing; at band corners they are 15dB down and have phase errors of 70°.

They compensate in the box/software but print the spec on the box - making it look like is wider than anyone else would claim. Since it is a closed system, it's only sorta cheating. 

[Wolfgang]

I see it the same way, its marketing. Of course you can calibrate this out, but you could do the same with a homebrew probe. The homebrew has further advantages (4kV isolation voltage plus less capacitance). I wonder why Dave was so impressed by this transformer.

[JS]

Of course you could try to reproduce the omicron injection transformer 1:1.

An alternative could be to use a stock ISDN or current transformer and to add a damping/compensation network to flatten out amplitude and phase response. Most swept generators have by far enough output for this approach, and the injected stimulus voltages are normally very small in order not to overdrive the DUT. When I tried this and if you apply the same criteria (5% phase error), results are not so vastly different from what omicron offers (e.g. 100Hz to a few 100kHz). The omicron claim of 1Hz to 10MHz is pure marketing; at band corners they are 15dB down and have phase errors of 70°. If you apply 1dB amplitude and 5°C phase error criteria

I like my phases in ºK, as ºC aren't absolute...

the range for them is ca. 50Hz to a few 100kHz (information from the datasheet of the B-WIT100). The big difference is that you have a 40dB power attenuation for the homebrew approach, which does not hurt at all in most cases.

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

One could also make a current amplifier and build a current transformer for the same freq range which might be easier, LF in current mode is much more forgiving than voltage mode, then terminate the transformer with the proper resistor and go from there. All that seems fine for the average lab, but this transformers exists for a reason, also, 1:1 transformer have better

This is from a transformer I've built some years ago for audio applications, source impedance was low, like 10Ω at most. The wiggle in the top end is due to the sound card I used, not coming from the transformer, I wander how it keeps going to a higher frequency, I only got the sound card at the time, so not much over 20kHz to work with at the time but I still have some of those at home, I'd only expect to get to a few tens kHz, is a big chunk of iron with a lot of turns, but I could wind a smaller one with less turns and better insulated wire to try.

I see it the same way, its marketing. Of course you can calibrate this out, but you could do the same with a homebrew probe. The homebrew has further advantages (4kV isolation voltage plus less capacitance). I wonder why Dave was so impressed by this transformer.

  It's specs aren't 1Hz to 10MHz, at least not in Dave's measurement, 5º was happening around 1.5MHz but at 1Hz was only 3º out. Test setup can influence on the measurement but his were at least as clean as you will get in a measurement setup. You don't need more than that, you need smooth transfer function without too much attenuation, as you can pick the reference from the output of the transformer, maybe a 3rd coil could be useful for that as it will reflect what the output is putting out better than the input coil and you wouldn't need a differential probe. In any case I wouldn't trust that what the generator is sending is the same as what the device is getting.

JS

[chris_leyson]

The core in the injection transformer is probably some nanocrystalline material as T3sl4co1l pointed out. It's got to be to get the high permiability required to get high magnetising inductance at low frequency. As it's a wideband transformer the low frequency limit is defined by the allowable peak flux density. From the data sheet fig. 5-2 they recommend a maximum drive level of 10dBm at 10Hz assuming a 10ohm injection resistor.
So, 10dBm is 0.707V RMS into 50 ohms or 1.414V RMS open circuit voltage at the generator output. Assume load resistance is 10 Ohms so that gives you 1.414*10/(50+10) or 235mV RMS across the primary. Peak flux density from the transformer equation is Bpeak(T) = Erms/(4.44 * f * N * Ae). N = 40, f = 10Hz, lets be generous and say Ae is 3cm² = 3x10^-4 m² putting the numbers in gives about 440mT peak at 10Hz. Also in the data sheet they give peak flux density as a volt time product, 3.5E-3 Vs, assuming 3cm² that gives a peak flux density of around 300mT. In calculating the RMS primary voltage I left out the primary magnetizing inductance and that would reduce the primary RMS voltage a little bit. It seems like quite a high working flux density, compared to ferrite anyway. It's only just about useable at 10Hz and I would say "usable at 1Hz" is pushing it a bit

Bifilar literally means "two filiaments" and it's two parallel wires bonded together, I've got a reel of it somewhere for winding coupling transformers on tiny binocular cores. Scientific wire company sell it https://www.scientificwire.com/acatalog/bifilar.html The winding on the toroidal core is twisted to probably reduce the electric coupling between adjacent turns, MiniCircuits for example wind coupling and matching transformers using twisted wire.

Where are the mixers ? They're probably using analog switches, cheap, wide dynamic range and poor noise figure.

[Kleinstein]

The advantage of the nano-crystalline core material is that is can be used to higher flux levels than ferrites. So 400 mT would be no problem. When used in mains transformers some 800-1000 mT should be OK, saturation is around 1.1-1.2 T.

At low frequency the maximum power gets lower and the impedance gets lower. Still injecting 0.1 V might still be enough.

From the other tread (EEV-blog 1103):  Dave showed a picture with an AD834 that is likely used as the mixer.

[JS]

So switches aren't used for the mixer, autocal, signal routing and filter managment I guess, there where a lot of them, I should take a look for the high res pictures!

JS

[Wolfgang]

Of course you could try to reproduce the omicron injection transformer 1:1.

An alternative could be to use a stock ISDN or current transformer and to add a damping/compensation network to flatten out amplitude and phase response. Most swept generators have by far enough output for this approach, and the injected stimulus voltages are normally very small in order not to overdrive the DUT. When I tried this and if you apply the same criteria (5% phase error), results are not so vastly different from what omicron offers (e.g. 100Hz to a few 100kHz). The omicron claim of 1Hz to 10MHz is pure marketing; at band corners they are 15dB down and have phase errors of 70°. If you apply 1dB amplitude and 5°C phase error criteria

I like my phases in ºK, as ºC aren't absolute...

I agree

the range for them is ca. 50Hz to a few 100kHz (information from the datasheet of the B-WIT100). The big difference is that you have a 40dB power attenuation for the homebrew approach, which does not hurt at all in most cases.

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

One could also make a current amplifier and build a current transformer for the same freq range which might be easier, LF in current mode is much more forgiving than voltage mode, then terminate the transformer with the proper resistor and go from there. All that seems fine for the average lab, but this transformers exists for a reason, also, 1:1 transformer have better

 

This is from a transformer I've built some years ago for audio applications, source impedance was low, like 10Ω at most. The wiggle in the top end is due to the sound card I used, not coming from the transformer, I wander how it keeps going to a higher frequency, I only got the sound card at the time, so not much over 20kHz to work with at the time but I still have some of those at home, I'd only expect to get to a few tens kHz, is a big chunk of iron with a lot of turns, but I could wind a smaller one with less turns and better insulated wire to try.

For stability analysis I dont see a case where a 1:1 transformer is needed, because injected signal level is always very low in order to remain in the linear domain. I would be curious about a practical example of a 1:1 case

I see it the same way, its marketing. Of course you can calibrate this out, but you could do the same with a homebrew probe. The homebrew has further advantages (4kV isolation voltage plus less capacitance). I wonder why Dave was so impressed by this transformer.

  It's specs aren't 1Hz to 10MHz, at least not in Dave's measurement, 5º was happening around 1.5MHz but at 1Hz was only 3º out. Test setup can influence on the measurement but his were at least as clean as you will get in a measurement setup. You don't need more than that, you need smooth transfer function without too much attenuation, as you can pick the reference from the output of the transformer, maybe a 3rd coil could be useful for that as it will reflect what the output is putting out better than the input coil and you wouldn't need a differential probe. In any case I wouldn't trust that what the generator is sending is the same as what the device is getting.

JS

Still a big discrepancy to their own datasheet. I need to research this a little bit. I am afraid I need to get hold of a Keysight E5061B with LF option