What is the bandwidth of conventional current transformers?

[asrock70]

1. Can conventional current transformers, such as this one, measure the current of a switched converter, with a switching frequency above 100kHz?
2 Is it better to measure the current on the primary (mains voltage and lower current) or on the secondary, in front of the diodes (lower voltage about 10 times higher current?

 

[KlausST]

Hi,

1) I doubt that it is very good at 100kHz. Good CTs come with a detailed datasheet. With specified frequency response.

2) it depends on what information you are after. Is it input current, ouput current, converter switch current...
One can not replace the other.

Klaus

 

[Warpspeed]

There are many different kinds of current transformers designed for very different purposes, they fall very roughly into two categories.

The first are high precision transformers used for instrumentation at 50/60Hz type frequencies.
These usually have a very high inductance, and turns count, a core which can be lossy at high frequencies, and must be used well below any self resonance which makes them unsuitable at the higher switching power supply frequencies.

The second class of current transformers are specifically designed for high frequency operation with faster rise times, are smaller and use cores that produce lower inductance and lower high frequency losses. These are not precision devices but designed more for applications such as very fast current limiting and shutdown. They simply do not have enough inductance to work down at 50/60 Hz.

As Klaus says, the data sheet should give enough details to indicate the intended application and suitability.

If its physically fairly large, its almost certainly the low frequency type because a larger volume core is required to get enough inductance at the specified turns.
If it mounts on a circuit board and is really small, and the specifications mention inductance or rise times, its likely the high frequency type.

 

[Easy peasy]

At 100kHz it is better to have a ferrite core - as the losses in iron will be significant even at low B.

 

[Warpspeed]

You can get away with an iron core in a choke to surprisingly high frequencies, provided the ac ripple component is kept low, compared to the dc component.

In transformer that only sees ac, core losses can become hugely significant, definitely ferrite at 100 Khz.
If the pulses are rectangular there may be components up into the Mhz range. Serious stuff.

 

[KlausST]

Hi,

This is confusing.

provided the ac ripple component is kept low, compared to the dc component.

This sounds as if the transformer can be used for DC.
I'd say it can't be used for DC at all, and if combined with AC ... the DC causes the core to saturate, so then even the AC transmission is prohibited.

Where am I wrong?

Klaus

 

[Warpspeed]

You can get away with an iron core in a choke to surprisingly high frequencies, provided the ac ripple component is kept low, compared to the dc component.

I said CHOKE not a transformer.

In transformer that only sees ac, core losses can become hugely significant,

A current transformer is a TRANSFORMER its all ac, no dc, so you need a good core.

Where is the confusion ?

 

[KlausST]

Hi,

Because the OP explicitely asks about transformer and not choke.

And even for a choke, when DC is high compared to AC, I guess the core becomes saturated and thus it's AC performance becomes reduced.
Your statement:

You can get away with an iron core in a choke to surprisingly high frequencies, provided the ac ripple component is kept low, compared to the dc component.

...sounds as if a choke needs high DC current to get improved AC performance.
"AC low compared to DC" means: I_DC > I_AC

Klaus

 

[Warpspeed]

Klaus, you are confusing everything.

Transformers operate only with ac, so the ac flux swing is all there is. The core losses increase with both frequency and flux swing, and if you have a lot of both a low loss core is a necessity.

A choke is a different beast entirely, it must have some dc flowing or it would be called an inductor.
If its totally 100% dc then there is no ac component and there will be zero core loss.
If the ac component is kept small core loss might be small too, and you can get away with running a lossy core.

Steel cores and powdered iron cores can work well in choke applications up to much higher frequencies than would be possible in a transformer because the ac induction is far less.

Dc and ac design of a choke are completely separate and need to be treated individually. The dc aspect has nothing at all to do with core loss, only with saturation and copper loss.

The ac aspect involves all kinds of things, such as eddy current and hysteresis loss in the core, possible resonance and also skin effect in the wire.
If the ripple current is kept very low, often the ac losses can be negligible in a choke.

 

[FvM]

Enough already! I think we should stop the discussion about chokes as it is mostly unrelated to the original question.

It has been stated by all contributors that dedicated high frequency CT uses ferrite rather than iron core. The original question doesn't exactly clarify what "conventional current transformer" means. If it means "standard 50/60 Hz industry product without frequency range specification", I won't expect any guaranted response in the high frequency range. We can however buy CT with a 20 Hz to 200 kHz frequency specification.

But I don't see any use of this transformers for high-frequency-only applications.

 

[asrock70]

Take a look at Mouser.com for available current transformers for PCB 100A.
Usually the frequency is completely silent for one stated usability up to 100kHz.
So. What I need ?

Measure the current on the primary or secondary of the switched power supply (flyback 120kHz) output 0-100A, resolution of current measurement 0,1A.

First option. Shunt resitor for example this on secondary side in ground.
Disadventage
- 7.5W loss at 100A
- mechanically complicated
- for cuurent 100mA is voltge on shunt only 75uV ie. necessary to use a high quality and expensive amplifier

Second option. Use current transformer on primar or seconder (in front of diodes).
- Here's the question of where to take a suitable, cheap and small CT.
- What will be the accuracy?
In welding inverters, measuring CT in feedback are commonly used and trained at frequencies above 60kHz.

I already thought of making a CT, but my initial consideration was the toroid core T130-2, ie iron powder. What core material do you consider most suitable?

When it comes to chokes, i found this link here on the forum.
CT for measure 0-100A on 42kHz Tr3 an use EMI suppression choke including the original winding, ie 75 turns.

 

[Warpspeed]

O/k the turns will be fixed by the required current ratio, and we need a core material with an acceptably low loss. And we need to arrange our secondary turns to have a self resonant frequency well above the operating frequency. It will all become slightly easier if we keep our turns and current ratio reasonably low.

If for example, the secondary turns are 50, the secondary must be able to carry 2 amps without excessive heating.
That sets the current rating for our current transformer. We also need to ensure the peak secondary voltage never becomes high enough to saturate the core.

But we might not be of the woods yet. If we were measuring symmetrical ac current, all that is then needed is to calculate a suitable burden resistor for the secondary.

Flyback circuits only flow current in one direction, both in the primary and the secondary, and that can potentially upset things mightily for our current transformer. But provided the current flow has less than 50% duty cycle through the current transformer it should be workable.

 

[Easy peasy]

any DC component in a CT will severely affect its operation ...!

 

[carpenter]

any DC component in a CT will severely affect its operation ...!

Yes, but the flyback converter has a maximum allowed duty cycle of 50% and so it should not have any DC component. Main pulse transoformer is without air gap. Where would the DC component come from?

For interest, here is a CT design that has a bandwidth of 67kHz to 67MHz. The question is, how far would he handle 100A?
Respectively increase the number of threads on the secondary

 

[FvM]

Yes, but the flyback converter has a maximum allowed duty cycle of 50% and so it should not have any DC component. Main pulse transoformer is without air gap. Where would the DC component come from?

Although the transformer flux is balanced, primary and secondary transformer current have DC component. The CT circuit must allow flux reset during off-time.

 

[Warpspeed]

The core will reset naturally if the duty cycle is less than 50%.

The danger is when measuring flyback transformerr secondary current, if the output is overloaded, the current in the secondary might take a very long time to ramp down, even if current limit pulls back the flyback transformer primary duty cycle. That could be very bad for the current transformer.
And an overloaded output is exactly where you really need the current transformer to work !

There are ways around this of course.
The usual "fix" is to feed the burden resistor through a series diode, so that the diode conducts during the measurement part of the cycle. A low forward drop Shottky diode is recommended for that.

To enable the core to reset during the non measurement period, the current transformer secondary voltage can be allowed to rise to a higher voltage. That allows the stored volt microseconds in the current transformer to dissipate faster, and core reset can occur more rapidly. Another diode (connected with opposite polarity) and a second burden resistor of higher resistance will accomplish a faster core reset, without having any effect on measurement.

 

[FvM]

The core will reset naturally if the duty cycle is less than 50%.

There's no specific 50 % limit for core reset. The capability of "natural" flux reset depends on duty cycle, CT secondary load, rectifier diode, shunt, voltage limiting devices.

 

[Warpspeed]

The current transformer secondary just has the burden resistor across it in the most basic configuration.
T=L/R holds for both forward and recovery. There may however, still be a problem with remnant flux in the core, and that can be quite high with ungapped high permeability material.

You are right though, when you add an asymmetric load that has diodes, voltage clamps or whatever, the situation changes, and that can be used to advantage.

 

[Easy peasy]

The core will reset naturally if the duty cycle is less than 50%.

this is not quite true - because it very heavily depends on the circuitry attached to the CT output.

The paper shown above is for a 5:1 CT - it can have a high BW because the capacitance is relatively low

For a 500:1 CT, say 150A rms @ 70kHz to 300mA, 10 ohm burden, other issues arise including total stray capacitance of the wdg, core properties , and connected circuit.

--- Updated Mar 8, 2022 ---

For a true "DC" component to affect a CT it must be there all the time, hence DC, pulsed current where the core can reset are not constant DC ( true DC ) but rather pulsed, the fourier "DC" component does not apply except to calc the average B in the core.

 

[asrock70]

I will return to the beginning and try to describe the situation in more detail.
The inverter is ma x výkon 100forward with two transistor, 230V, fé over 100kHz (maybe 120)
I calculate the maximum power of 100A for the duty cycle 0,35 never exceeds 50%
Diodes on the secondary, I do not know yet, but something in ISOTOP, and an example TTH200L06TV

Assume a current measurement on the secondary
Why, higher range and therefore higher accuracy at low currents, but maybe I'm wrong and it's better to put it on the primary



Toroid 36X25X15 CF=195A AL=6700 Initial Permeability μiac 5000
1 turn prim 100A Cu row
20 turn sec 5A maybe Cu tape or 4x0,35mm wire(bifilar)
L1 = 6,70uH
L2 = 2,68mH

What do you think about it?
Ferrite suitability, number of turns, etc.?

Reset circuit?
The first question is whether it is necessary?
If so, here are three options to choose from?

a) only one schottky diode
b) adds a second diode and a reset voltage when inactive
c) same as b, only an additional active load less resistance