Bandwidth of Hall sensor?

[treez]

Hello,
Cold a Hall sensor be used to reliably sense the FET current waveform in a flyback converter? Would it distort the waveform too much?. The attached shows the waveform of the FET current as a voltage across a sense resistor, but how would a Hall sensor compete with this?

 

[KlausST]

Hi

The hall sensors should be fast enough.
Read datasheet to get information on specific devices.

Klaus

 

[dick_freebird]

A bare Hall sensor is fast, but put it inside an op amp loop
to get decent amplitude on low currents and you're stuck
with op amp type slew rates / bandwidths.

There's a variety out there and you might find some that
have both sensitivity and speed, adequate. At least for
average current. Pulse by pulse, following real current to
better than 10% accuracy in-the-moment, I kind of doubt.

I think I'd put up any that seemed likely, against a CT-1
and check the dynamic response.

You might be able to put together your own Hall element
and your own amplifier, and do better for your particular
use than what's on the shelf. For example if you wanted
100X gain you could cascade two 10X amps but the folks
selling packaged Hall+amp probably only think to use one
at 100X, and 10X worse bandwidth as a result. Maybe
you would trade DC accuracy for a really fast amp, in
your values-system, since the voltage loop ought to be
where DC accuracy resides, not the current loop. That
kind of thing.

 

[treez]

thanks, we're doing a 7kw battery charger and current sense resistors are going to be too dissipative, so we look into Hall. Current sense transformers also may struggle with the sheer amount of primary current we're putting through them.
And current sense resistors, the 4 terminal ones , are not readily parallelable. -Neither are current sense transformers parallelable in real terms.

 

[Warpspeed]

It depends on your priorities.

A straight Hall sensor amplified by a wide bandwidth amplifier can be made very fast indeed, but it may have fairly disappointing accuracy or repeatability.
But for a fast current limit sensing application, that may not be an issue.

The accurate ones purposely designed for measurement have a solenoid coil and an error nulling amplifier which vastly improves everything, but the response to a step change may typically be a microsecond or two.

For flyback, you will know the rate of current rise, and a microsecond or two may not be such a big deal. You may even be able to put the Hall sensor directly in the airgap.

 

[mtwieg]

Depends on the sensor. I've seen some with bandwidths of a few kHz, others in the MHz. I assume you are referring to using a closed loop hall sensor like a LEM. When used in the closed loop configuration the overall bandwidth will typically be lower than the hall sensor's open loop bandwidth. Usually this is find so long as you're measuring continuous current waveforms, but with a flyback that is not possible, unless you were to wind both the primary and secondary through the sensor (never seen this done before). Because of that I would recommend against a hall sensor and just use a shunt resistor.

 

[schmitt trigger]

Like everything, there are tradeoffs.
Hall elements by themselves have significant amounts of offset. Therefore, for precision sensors, sometimes an internal chopper stabilized amplifier is used. Chopper amps are inherently slow.
But...you'll have to check the particular sensor architecture itself, and read the datasheet.

 

[treez]

Thanks, we wish to use Hall because the power is too great for sense resistors, and there are problems paralleling high power, 4 terminal sense resistors.

 

[D.A.(Tony)Stewart]

Thanks, we wish to use Hall because the power is too great for sense resistors, and there are problems paralleling high power, 4 terminal sense resistors.

Linear Hall Effect Sensors use ADC and DAC with sample rates around 7kHz. Otherwise they are non-linear.

Watts your problem with a 75mV shunt? SNR? power? accuracy?
resolution? You may be doing something wrong. What current range?

 

[FvM]

Linear Hall Effect Sensors use ADC and DAC with sample rates around 7kHz. Otherwise they are non-linear.

Why, particularly? There's no obvious relation between sample rate and the bandwidth of a linear component.


The popular Allegro hall current sensors have e.g. 120 kHz bandwidth, others even more. Although they implement a chopper alike "dynamic offset cancellation" method. The main reason why hall sensors are designed with limited bandwidth is sensor and amplifier noise, there's are also DC current probes with MHz bandwidth and respective higher noise level.

Like SunnySkyguy, I don't believe that you already exhausted the capabilities of current shunts. There's, by the way, no problem to use four-terminal shunt in a parallel circuit, if necessary at all. Simple resistor networks can average there output.

 

[treez]

if one parallels four terminal resistors , then you can get problems with the power current unwontedly flowing along the paralleling tracks. As you know, the paralleling tracks are only wanted for signal.

 

[FvM]

You probably overlooked that I suggested resistor networks to connect the sense terminals. I'm sure you can figure it out.

 

[memarian]

Hi,

I don't know about your particular application , but for pulse to pulse current monitoring we used CT Transformers and/or Shunt Resistors easily in our inverter welding machines.(Power : 20KW Max)

We used shunt resistor in secondary (@ 500A Current) . These are typical 75mv/500A shunt resistors available widely and can be purchased easily.

We used CT Transformers in primary (@ 80A Current) without any problem.These are also available widely and can be purchased easily.

BTW : Our Maximum Switching Frequency is ~40KHz.

You can purchase these components easily even from chinese manufacturers. You should try looking for these parts with welding keyword.

 

[D.A.(Tony)Stewart]

I meant there are Hall Sensors with ADC/DAC in the chip that run at 7kHz to linearize them.

Allegro also linearize but without compromising bandwidth and have the current wire inside. e.g.ACS709

Changes are your peak shunt voltage is far more than 75mV and thus your peak power demands parallel shunts. This is somewhat of a standard voltage but demands low offset circuits or high side current sense IC.

 

[treez]

Thanks, we looked into Allegro...
We are going to use the ACS758LCB-050B Bidirectional (datasheet below) Hall sensor to monitor the current in series with the resonant inductor of our 7.4kW Full Bridge LLC Converter. This is to shut the LLC controller down in case of overcurrent.
Do you know why our competitors are not doing this?, it is, after all, so easy.
We miss sensing the negative half of the sinusoidal current, but that is not going to matter.
The LLC converter is resonant at 60KHz, and we operate it from 45kHz to 200kHz.
Presumably nothing untoward will happen when we go up to 150khz?

ACS758LCB-050B Hall sensor datasheet:
https://www.farnell.com/datasheets/1927450.pdf

 

[memarian]

Thanks, we looked into Allegro...
We are going to use the ACS758LCB-050B Bidirectional (datasheet below) Hall sensor to monitor the current in series with the resonant inductor of our 7.4kW Full Bridge LLC Converter. This is to shut the LLC controller down in case of overcurrent.
Do you know why our competitors are not doing this?, it is, after all, so easy.
We miss sensing the negative half of the sinusoidal current, but that is not going to matter.
The LLC converter is resonant at 60KHz, and we operate it from 45kHz to 200kHz.
Presumably nothing untoward will happen when we go up to 150khz?

ACS758LCB-050B Hall sensor datasheet:
https://www.farnell.com/datasheets/1927450.pdf

Hi,

according to the hall sensor datasheet, Sensor response time is around 4us, and @ 200KHz , your half cycle is 2.5us , or @ 150KHz your half cycle is 3.3us .

if you want to use this sensor for sensing pulse current, you can't. it isn't fast enough.depend on your frequency you will read a nonlinear value which can be incorrect too.

we use hall sensors in our LLC inverter welding machine @ output , for sensing average output current. For sensing pulse current , i think the CT transformer is the best solution.

 

[treez]

Thanks, I see what you mean, but that propagation delay looks like it just applies to when the current is just starting to be non zero.

The Hall sensor says it has a bandwidth of 120KHz…the half period there, is 4.17us. If it has indeed got a bandwidth of 120khz, then it must be able to depict the 4.17us “half sines” of a 120khz waveform.
Therefore, it must have a real delay time of much less than 4us.

Also, this is for overcurrent sensing only.

 

[FvM]

We are going to use the ACS758LCB-050B Bidirectional (datasheet below) Hall sensor to monitor the current in series with the resonant inductor of our 7.4kW Full Bridge LLC Converter. This is to shut the LLC controller down in case of overcurrent.

My first choice for an AC current measurement would be a current transformer or coreless dI/dt (Rogowski) sensor. The Allegro sensors aren't bad if the isolation is sufficient for your application.

The datasheet doesn't specify a large signal bandwidth or linearity for AC current measurements. It would be necessary to determine this parameters empirically. At least for the high current range (150 or 200 A), you should also calculate a skin effect related current derating in high frequent AC applications.

 

[treez]

Actually I see your point, the Allegro Hall sensor may not be able to "keep up" with the changing AC, and may just signal a low output, (even in case of overcurrents) since the average input is zero (because of AC), so the sensor output will not be able to cumulatively build up a value incrementally....yes, I agree, because its AC, we need a CST. This leads to our next problem, that we cannot find CST torroid cores which have low enough permeability so that their primary inductance is low enough (with 1:100 turns) not to add considerably to the value of resonant inductor seen by our LLC stage. We need the one turn primary to have an inductance of less than 250nH.
The current in the LLC bridge is 25 Amps rms.

We went through the entire mag-inc range and not one gave us low enough primary inductance when wound 1:100.

 

[FvM]

Simply don't understand your current transformer problem. High frequency CT can be made with single digit nH leakage inductance.