EEVblog #1104 - Omicron Labs Bode 100 Teardown

[EEVblog]

It looks like they take +/-40deg as the "frequency range" of the unit (i.e. 7Hz to 5MHz).
The 1Hz to 10MHz is the "usable range"

[EEVblog]

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.

[Hydron]

Dave, could you quickly check the magnetising inductance of the isolation transformer? Would help with understanding what the low frequency performance is actually like.

If you wanted to go to town and play with the new toy then feel free to do a full set of measurements of the parasitics - the VNA (aka FRA) is great for this. Definitely agree that an instrument like this is very handy to have around - you'll keep finding uses for it.

Edit: Just found the SPICE model on their website - no need to measure it unless you want to compare specs to reality.

[bitwelder]

For 5K I'd hoped Omicron' virgins would have taken care to align those relays a little better. 

[Wolfgang]

@Dave

Why didnt you use a matching network to straighten things out ?

BTW: In about 2 months Im getting my hands on a Keysight E5061B-3L5 (5Hz to 3GHz) VNA with impedance measurement option.
Then I can repeat my measurements, and I will publish them as usual.

regards
  Wolfgang

[Wolfgang]

Omicrons nude virgins definitely work in the software department

[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.

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

The pattern of deformation is exactly what PTFE looks like when its heated. Soldering that center pin takes a lot of heat...

The core isn't anything special either. I'd be willing to bet it's a bog standard iron powder T184-2. Just a hunch.

[Floyo]

It cant be that T184-2 core, the AL is way too low to get the specified inductance. My bet goes out to a Vacuumschmelze W424, the colour matches, the size matches, as do the specs if you run the numbers. I knew I had one of those laying around somewhere, so I dug it up and whipped up another test. This time 50 turns (I had some wire leftover after 40 ) on the W424 core, the picture speaks for itself.

Once again I measured with the analog discovery. The "Cal" image shows  the wavegen output connected to both CH1 and CH2.
CH2 is displayed in blue and is in relative mode, so gain and phase relative to channel one. This removes any gain variations due to wavegen output loading etc (its a 50 ohm output).

"Abs" shows the measurement as Omicron does it, I think, Here the amplitude is absolute compared to the wavegen (and thus -6db because of the 50Ohm termination), and the phase is relative to CH1.
"Rel" is the same measurement, but now in relative mode, this shows the transfer function of just the transformer.
"100mV" is the previous absolute measurement but now at 100mV instead of 1V input amplitude, this lessens the loading on the wavegen output stage.

Pri -sec capacitance is 130pF
Magnetizing inductance is 274mH
And leakage inductance is 3uH

[rx8pilot]

What influence does the twist pitch of the wire have on performance?

[Wolfgang]

It changes the characteristic impedance of the transmission line. Depending on twists/m, the impedance will be around 100Ohms, with more twists making the impedance lower.
Its not extra simple, because wire diameter and dielectric used also play a role.

[Floyo]

How about using coax, it has already been tuned to be a good 50 ohm transmission line. And how about impedance matching from primary to secondary, and/or the transmission line impedance.

For practical measurements on power supplies etc 1-2 Mhz should be plenty of bandwidth. I have never seen anything that comes close to those bandwidths, but you never know

[GigaJoe]

it seems close , Vitroperm 500 F , Vacuumschmelze
https://www.mouser.com/pdfdocs/VACChokesandCoresDatasheet.pdf

[Hydron]

Damn, just went to post my DIY effort and found I'd been beaten to it!

I too used a nanocrystaline core, this one scavenged from a Wurth 3-ph CM choke:
https://katalog.we-online.com/pbs/datasheet/744839047160.pdf
Looks to be a little bigger than the one linked earlier, and encapsulated inside a plastic shell (I removed the protrusions in the pic before winding it).

Made a twisted pair with a drill and some random hookup wire and put 40 turns on it, as this coincidentally both matched the original number and also gave the same magnetising inductance as the SPICE model of the original indicated (actually 39 would have been the correct number in restrospect).

Took maybe an hour all up to build (not including measurements) and gave:
Xm = 222mH
Xl (per winding) = 1.25uH
Cpri-sec = 140pF

Unfortunately I couldn't do the whole sweep from 1Hz-10MHz in one hit, so there are four graphs here (corresponding to Floyo's "abs" and "rel" measurements - filenames should be descriptive enough of what's being measured).

While the low-freq performance of mine is slightly better, it only makes it to ~7MHz before falling apart. Possibly because I wound it while clothed, and don't fulfil the other two criteria either?

[JS]

Damn, just went to post my DIY effort and found I'd been beaten to it!

I too used a nanocrystaline core, this one scavenged from a Wurth 3-ph CM choke:
https://katalog.we-online.com/pbs/datasheet/744839047160.pdf
Looks to be a little bigger than the one linked earlier, and encapsulated inside a plastic shell (I removed the protrusions in the pic before winding it).

Made a twisted pair with a drill and some random hookup wire and put 40 turns on it, as this coincidentally both matched the original number and also gave the same magnetising inductance as the SPICE model of the original indicated (actually 39 would have been the correct number in restrospect).

Took maybe an hour all up to build (not including measurements) and gave:
Xm = 222mH
Xl (per winding) = 1.25uH
Cpri-sec = 140pF

Unfortunately I couldn't do the whole sweep from 1Hz-10MHz in one hit, so there are four graphs here (corresponding to Floyo's "abs" and "rel" measurements - filenames should be descriptive enough of what's being measured).

While the low-freq performance of mine is slightly better, it only makes it to ~7MHz before falling apart. Possibly because I wound it while clothed, and don't fulfil the other two criteria either?

Nice attempt, I like the response, being bigger might give you the good LF response. The plastic shell leaves the wire further away from the core, making a worse coupling thus higher leakage inductance which could affect the HF inductance. Also, the non PTFE wire would affect HF performance, I guess one or both of those are giving you the differences. You could try twisting the wires a bit more or building a thiner shell for the core (like some tape to keep it together and not damage the wire insulator) if you are after a better HF response.

I'd love to have some cores to play around with!

JS

[chris_leyson]

Big thanks to Tim for the heads up on Vacuumschmelze nanocrystaline material. T60006-L2040-W424 looks like the core they use, it's got the right Al value. Thanks to Floyo and Hydron for winding some test transformers and providing test results.
I wonder, did Omicron try 2:1 and 3:1 transformer ratios if they are injecting voltage into a 10 Ohm load. Why would you pick 10 Ohms and not 50 Ohms for a load anyway, it doesn't make any sense. Twisted trifilar or quadrifilar wound would give you the impedance match either side of 10 Ohms and also drop the LF corner. Not sure how that would effect the HF corner, will have to try it out. Nice 1:1 wideband transformer, no magic voodo involved and overpriced for what it is. More marketing than practical engineering.
Anyways, thanks Tim for the heads up on the material, it's not cheap and won't compete with ferrite in terms of price but for wideband transformers it has potential, all depends on the HF losses.

[T3sl4co1l]

Cheers.

Note that you can use ferrite just fine, but you'll lose a couple octaves at the low end.  You can find materials with mu_r up to about 20k, with >8k being typical for hi-mu types.

The mid-band (i.e., magnetizing) impedance also isn't as high, but this isn't relevant at these impedances.  It's quite useful for CMCs, hence the application of nanocrystalline material there as well.

Tim

[Hydron]

I took some 50 ohm load measurements at the same time if they are of interest - see attachments.

As for PTFE vs PVC wire, I didn't have any of the former, though I could butcher an Ethernet cable for a pair of something which I assume is not very lossy. Would it really make a major difference at <10MHz? Throwing my measured lumped parameters into SPICE gives reasonable correlation to the measurements (biggest outlier is that the resonance point in SPICE is a little higher - more like 8.5MHz than ~7MHz).

There are also some other things I could do to improve matters if I ever needed to use this in anger up near the top of it's range - a few less turns, shorter lead-in wires etc etc. No real need though, this was mainly done for the challenge (wasn't much of one given the right core material) and I already have an isolated sig-gen for my Cleverscope which lets you do the PSU loop measurements from <1Hz up to 65MHz with no magnetics required.

[capt bullshot]

Building a transmission line transformer like this appears quite a challenge ...

Anyway, here's my results (using random cores from the junk box, didn't record the µH/n² values):

The ugly, a rather bulky ferrite core and thin litz wire (about 5m wire length)


Works (-3dB) from 81Hz to 3.4MHz


Next try, a core used as an EMC suppressor (just a multi-strand cable fed through it), and Wire-Wrap (ca. 3m)


-3dB from 117Hz to 11Mhz


A core scavenged from an EMC common mode choke (VAC / Sekels branded, rather small), used less than 2m twisted pair from a CAT5e ethernet cable


-3dB from 48hz to 14Mhz


Now I'm searching for a somewhat larger core of the last kind (probably micro-crystalline, as it is from VAC/Sekels) to give it another try and reach lower frequencies. The length of the twisted pair clearly defines the upper limit, simply verified by measuring a length of twisted pair before winding it onto the core.

[Kleinstein]

There are 2 effects limiting the usefulness at low frequency:
1) core saturation, which is amplitude (voltage) dependent
2) the inductance and thus impedance, which would depend also on the source / load impedance

With the high Al ferrite cores, it can be important to keep the mechanical stress from the windings low (e.g. add so foam). Stress can lower the AL value quite a lot.

The nano-crystalline material, has low magnetostriction and is thus less sensitive to stress.

Magnetostriction also causes another effect (ferro-resonance):
 there can be a rather narrow mechanical resonances (high Q, maybe 1000) that can also act back on the electric parameters. Due to the small width it could be easily overlooked. For high quality it might help to have some soft (e.g. wax) potting of the transformer / winding to dampen the resonances.

[halexa]

Why is there no ground fill ont that main PCB?

[rx8pilot]

Building a transmission line transformer like this appears quite a challenge ...

Anyway, here's my results (using random cores from the junk box, didn't record the µH/n² values):
-3dB from 48hz to 14Mhz

I would be rather happy with that one in many applications.

Is that a 3577A you used for the test?

[thm_w]

Why is there no ground fill ont that main PCB?

You'd have to ask the designer. But there would be at least a few inner planes (4+ layer PCB), used for power and ground.
So strictly there is no need for a top layer ground fill, but could reduce EMI slightly if it were used.

[capt bullshot]

Anyway, here's my results (using random cores from the junk box, didn't record the µH/n² values):
-3dB from 48hz to 14Mhz

I would be rather happy with that one in many applications.

It's easy to build, I'll try to find the original part number of the core / EMC choke, and I'm going to look for more cores / chokes when I have some spare time, visiting the various junk boxes (including the new  stock, that isn't junk at all).

Is that a 3577A you used for the test?

Yes, it is mine. Located in my home lab, we don't have such fancy stuff at work, but quite a variety of EMC chokes ...

[capt bullshot]

Just for the record, the core I've used for the third transformer is most probably the Vacuumschmelze T60006-L2030-W423, this is the nearest match. The common mode choke that I scavenged that core from is a T60405-R6166-X018-80. Didn't find any more interesting cores for more experiments. Can't beat the mentioned T60006-L2040-W424 with any of the cores that I've found.

[rx8pilot]

Is that a 3577A you used for the test?

Yes, it is mine. Located in my home lab, we don't have such fancy stuff at work, but quite a variety of EMC chokes ...

For this thread - I was setting up to do some tests with the same 3577A network analyzer unit. The PSU has some issues and it finally would not power up at all. I believe I have located the problem and parts are on the way. Hopefully, if the failed PSU did not fry the rest of the system, I can do a few tests as well.

I have 3 DC-DC converters on the design table and would really like to improve my ability to analyze loop stability.