What kind of transformer is this ? (EEVBlog #1119)

[Kleinstein]

The normal relays have problems providing the required insulation. So it would need special high voltage relays. For a full insulation with a floating bucket cap  it would need 4 contacts and capacitance could be comparable, maybe even high.

With something like 2-3 W the power requirements are not that low.

[grythumn]

silly question, but, given you only care about CMRR when actually making a measurement, and we know the high speed sampling means the device cannot actually stream continuously, then perhaps a (super) capacitor could be charged during the non measurement time by a simple relay connected DC/DC ??  Given the low power requirement for the front end, you wouldn't need that much "on time" to charge it?

Didn't he say in the video that they were streaming continuously, but decimating the data stream internally before putting it out over USB3?

-R C

[T3sl4co1l]

Most likely a molded part, glass filled liquid crystal resin or the like -- same thing bobbins and connectors are usually made of.  Failing that, GF nylon would be cheaper and probably okay as far as not falling apart during soldering.

FYI, PTFE can't be molded, at least not easily as far as I know.  The next best thing is PFA or something related.

Tim

[coppercone2]

how come it does not have a electrostatic shield?

can you do 2 slotted shield each to a separate ground plane? isolation? Cant you put teflon plate?

[tautech]

how come it does not have a electrostatic shield?

Doesn't need one.
The front ends are in cans where the waveforms are digitized and then passed on via fiber.
After a year of development they'd have used shields if they were needed which arguably they should've even just to help protect their IP.

[coppercone2]

i mean on the transformer that looks kinda crazy

[tautech]

i mean on the transformer that looks kinda crazy

So do I. Watch the vid, just the PSU took a year of development......that includes the toroids for the PSU's.

[coppercone2]

i mean on the transformer that looks kinda crazy

So do I. Watch the vid, just the PSU took a year of development......that includes the toroids for the PSU's.

I watched the video. He just says it took a year to develop. Did I miss him talking about electrostatic transformer shields?

[T3sl4co1l]

Shields would only increase capacitance without affecting EMI (which is minimized thanks to other filtering and presumably a nearly-sinusoidal resonant converter).

Tim

[coppercone2]

how many DB can you get with conventional shielding techniques vs these advanced novel methods?

is his design a theoretical feat? So if we want to approach these performance levels that is the only way thats reasonable?

[BrianHG]

how many DB can you get with conventional shielding techniques vs these advanced novel methods?

is his design a theoretical feat?

If I remember from the video, it was quoted as over 100db isolation at 50Mhz.  You get more crosstalk, less isolation, at 50Mhz on 2 parallel 1 inch long traces, 20 mil spaced apart on a normal PCB.

If you study that black plastic case, you will notice it has guides for the inner core's windings so they are in fixed positions.  The outer windings never run parallel over the inner core windings.  It is beautifully designed to make sure no RF crosses from side A to side B without any shield which would add a capacitive link between the inner winding and the frame-ground where that shield would be tied to.  You cannot do that in this application where the GND connection of the probe may be an active modulating 500v signal.

[tautech]

i mean on the transformer that looks kinda crazy

So do I. Watch the vid, just the PSU took a year of development......that includes the toroids for the PSU's.

I watched the video. He just says it took a year to develop. Did I miss him talking about electrostatic transformer shields?

No you didn't miss anything except arising to the obvious conclusion (for me) that no shields are needed. Why could you think that guys that design stuff like this and take a year to perfect it could possibly overlook shielding if it was needed. They wouldn't, period.

[coppercone2]

i mean on the transformer that looks kinda crazy

So do I. Watch the vid, just the PSU took a year of development......that includes the toroids for the PSU's.

I watched the video. He just says it took a year to develop. Did I miss him talking about electrostatic transformer shields?

No you didn't miss anything except arising to the obvious conclusion (for me) that no shields are needed. Why could you think that guys that design stuff like this and take a year to perfect it could possibly overlook shielding if it was needed. They wouldn't, period.

i wanna know engineering details on its behavior and when this technique should be used and why the other one does not work I am not questioning that it works, and I want to know for sure that its not done that way for cost reasons.

[coppercone2]

how many DB can you get with conventional shielding techniques vs these advanced novel methods?

is his design a theoretical feat?

If I remember from the video, it was quoted as over 100db isolation at 50Mhz.  You get more crosstalk, less isolation, at 50Mhz on 2 parallel 1 inch long traces, 20 mil spaced apart on a normal PCB.

If you study that black plastic case, you will notice it has guides for the inner core's windings so they are in fixed positions.  The outer windings never run parallel over the inner core windings.  It is beautifully designed to make sure no RF crosses from side A to side B.

I wonder if you can extrapolate some kind of theoretical model/equation for shielded torroidal transformer and their capacitance/coupling so you know when your grinding your mind doing something impossible. What the maximum shield limit is.

[BrianHG]

Yes, the outer winding crosses the inner winding only at 2 points at 2 sides on the transformer, and at 90 degrees, not in parallel, traversing at the wrong angle.  For everything else, you need to speak with a specialist in the field & build and test measure with believe it or not, high voltage signal tracers/network analyzers up to 100 Mhz.  Which was also divulged in the video that they did such tests and their test hardware only went up to around 65Mhz where a plot was shown on paper.

[coppercone2]

Well I did 2 estimated size donut shaped shields layers surface area with 2mm spacing and the result is like 40pF for dual shields coupling ground planes to each other if anyone is curious

[BrianHG]

like 40pF for dual shields coupling ground planes to each other if anyone is curious

That's like a 0.07 ohm short at 50Mhz.
You got nothing on this scope's transformer's somewhere less than 0.04pf capacitance from primary to anything else.  That's more than 1000x less capacitance.

[coppercone2]

I guess you just can't use shields at higher frequencies for isolation in a transformer

[BartSchroder]

Hi,
It took us a while to design this transformer, so we will keep it as a puzzle for you to delve into.
The transformer is just for power isolation. We use the optical fibers for communication.
However, the key requirements for the transformer are:
1. It has to have as low capacitance as possible. As several have pointed out, a 500V 8ns edge will go through a capacitor like butter if the capacitance is too high.
2. That it be as balanced as possible - this is so that the common mode output voltage, which could get into the circuit being measured, is minimized.
3. That it be reasonably efficient. We have to put 3W across the gap.
The former is 3D printed, and yes the shape is important, as is the material. The toroid is a fairly standard one.

The driving system is also pretty important, and you can use your imaginations on that 
cheers, Bart

[tautech]

Thanks Bart. I had a look at your wares at the Auckland EMMEX show earlier this year, very impressive capabilities. 
We had a chat too but of course I wasn't waving any flags to reveal any allegiances. 

[BrianHG]

Hi,
It took us a while to design this transformer, so we will keep it as a puzzle for you to delve into.
The transformer is just for power isolation. We use the optical fibers for communication.
However, the key requirements for the transformer are:
1. It has to have as low capacitance as possible. As several have pointed out, a 500V 8ns edge will go through a capacitor like butter if the capacitance is too high.
2. That it be as balanced as possible - this is so that the common mode output voltage, which could get into the circuit being measured, is minimized.
3. That it be reasonably efficient. We have to put 3W across the gap.
The former is 3D printed, and yes the shape is important, as is the material. The toroid is a fairly standard one.

The driving system is also pretty important, and you can use your imaginations on that 
cheers, Bart

I was off by 0.5watts.  The only thing, which I'm sure is a secret, is that coupling capacitance which I'm sure you will keep a secret.  Though, I did mention that it was in house 3D printed... 

And Bart, I am thoroughly impressed the extent you went through on your scope to achieve the degree of performance you achieved.  I am nutty in the audio world and know about going to engineering extremes to achieve a particular spec, or, many times for DAC units with absurdly good enough isolated outputs.

[BravoV]

Hi,
It took us a while to design this transformer, so we will keep it as a puzzle for you to delve into.
The transformer is just for power isolation. We use the optical fibers for communication.
However, the key requirements for the transformer are:
1. It has to have as low capacitance as possible. As several have pointed out, a 500V 8ns edge will go through a capacitor like butter if the capacitance is too high.
2. That it be as balanced as possible - this is so that the common mode output voltage, which could get into the circuit being measured, is minimized.
3. That it be reasonably efficient. We have to put 3W across the gap.
The former is 3D printed, and yes the shape is important, as is the material. The toroid is a fairly standard one.

The driving system is also pretty important, and you can use your imaginations on that 
cheers, Bart

Thank you Bart for sharing few hints, really appreciate it. 

[mzzj]

Besides a transformer and photovoltaic, there would also be the option to use an acoustic transformer: In the extreme case a glass rod with transducers on both sides. Efficiency is expected to be better than the optical way.

And honorable mention for the the old brute solution of motor-generator combination. Actually really good option if you need say 100kW with 4000kV isolation  and low capacitance

[coppercone2]

can you parallel a buncha other transformers to make a non leaky one so long you don't care its big?

[coppercone2]

lol