AC Constant Current Source (2Apk-pk, 10~20kHz) - Possible?

[onemilimeter]

I wish to have an AC constant current source, which should be capable of supplying a constant 2Apk-pk AC current in the frequency range of 10~20kHz.

The load is an inductive load with impedance (0.5 + 2*pi*f*L) ohm, where L is 4mH.

Do you it's possible to achieve such current source?

Kindly advise if you know any useful online resources.

Thanks.

[Zero999]

It makes sense to me, it's an AC current source which always gives a peak current of 2A regardless of the load impedance.

I don't understand what the impedance is though, the inductance is 4mH but what's the resistance? 0.5 Ohm if I've interpreted what you've written correctly but why didn't you just say in the first place?

The impedance of a 4mH inductor at 20kHz is 503 Ohm

The voltage would need to be 1006kV peak or 711V RMS.

This is not an easy task.

I think the most sensible way of doing it would be to build an audio amplifier and use a transformer to boost the voltage. The feedback could be taken via a hall effect sensor or a current sense transformer.

[NiHaoMike]

If only one frequency is needed, make a series resonant circuit driven by a H bridge. Regulate the current by varying the frequency.

[Zero999]

Yes that will work but you'll need some feedback to keep the current constant if the load impedance changes.

The frequency is variable though.

If you use a 28nF capacitor and tune it to resonance at 15Hz. Now the impedance should always be <126Ohm so the voltage can be reduced to 252V peak, 178VRMS.

[onemilimeter]

If only one frequency is needed, make a series resonant circuit driven by a H bridge. Regulate the current by varying the frequency.

"Series Resonant Circuit driven by a H bridge" is something new to me. In my application, I need to vary the frequency from 10Hz to 20kHz.

[onemilimeter]

Yes that will work but you'll need some feedback to keep the current constant if the load impedance changes.

The frequency is variable though.

If you use a 28nF capacitor and tune it to resonance at 15Hz. Now the impedance should always be <126Ohm so the voltage can be reduced to 252V peak, 178VRMS.

Would you please advise how did you get "<126Ohm"? By the way, why did you select 15Hz?

Thanks.

[Zero999]

Oh, I thought you said 10kHz to 20kHz, you need to go down to 10Hz which makes it more difficult.

I selected 15kHz (15Hz was a typo) because I thought it was half way between the two frequencies you need.

126 Ohm was calculated as follows:

First I worked out the impedance of the inductor at 15kHz:

X_L = 2pi*FL = 2*3.142*15000*0.004 = 377 Ohm

For resonance at 15kHz the capacitor connected in series with it needs to also be 377 Ohm

C = 1/(2pi*FX_C) = 1/(3.142*15000*337) = 0.0028 = 28nF

The impedance at 20kHz will be:
Z = X_L - X_C
X_L = 2pi*20000*0.004 = 503Ohm

C_C = 1/(2pi*20000*0.0028) = 284 Ohm

503 - 284 = 219 Ohm

The impedance at 10kHz will be:

Well we don't need to bother doing the calculations again, the impedance of the capacitor will be doubled and the impedance of the inductor will be halved:
X_C = 2*284 = 568 Ohm
X_L =  503/2 = 251.5 Ohm

Z = 568 - 251.5 = 316.5

Oh well looks like I was wrong LOL, the maximum impedance is 317 Ohm I obviously didn't think it through, never mind you've chosen 10Hz to 20kHz so power factor correction won't work. You need an amplifier capable of putting 2A into a 503 Ohm load which won't be easy to make.

[onemilimeter]

Thanks Hero999...

I've decided to reduce the frequency to 10kHz and the pk-pk current to 1.0A.

From calculation, the impedance at 10kHz is [0.5ohm + 2*pi*(10kHz)*(4mH)] = 251.83ohm.

The required pk-pk voltage = 251.83 * 1.0 = 251.83V. Or, RMS voltage = 251.83 * 0.3536 = 89.05Vrms.

Do think it is easy to construct this "reduced" spec variable frequency constant pk-pk AC current source? Using a transformer?

Thanks.

[scrat]

Is the frequency to be varied continuously? I mean, could you divide the range into subranges, and move from one to another with some discontinuity, for example moving from one transformer ratio to another?

In the hypothesis of changing the ratio at some steps, I think it would be feasible, otherwise I think it would be quite difficult to control a current by means of an actuator voltage that goes across 3 orders of magnitude...

[onemilimeter]

Is the frequency to be varied continuously? I mean, could you divide the range into subranges, and move from one to another with some discontinuity, for example moving from one transformer ratio to another?

In the hypothesis of changing the ratio at some steps, I think it would be feasible, otherwise I think it would be quite difficult to control a current by means of an actuator voltage that goes across 3 orders of magnitude...

I wish the frequency of the sinusoidal current to be varied continuously, or in step of 10Hz.

I'm not sure if the following works:

Audio Power Amplifier (e.g. LM3886) ==> Step-Up Transformer ==> Load

With +/-35V supplies, the LM3886 is capable of supplying 50W at 8ohm (i.e. Iout=2.5Arms and Vout=20Vrms).

The transformer is designed to have a step-up turn-ratio of 5. Thus, the output voltage of the secondary winding will be 20Vrms*5=100Vrms, which is higher than the required 89.05Vrms. For a secondary output current (i.e. load current) of 0.3536Arms (i.e. 1.0Apkpk), the primary current will be 0.3536*5=1.768Arms, which is still within the output current capability of the LM3886.

For frequencies lower than 10kHz, the output Vrms of the LM3886 is decreased to maintain the load current of 1.0Apkpk.

The impedance of the transformer should be made as small as possible. The transformer should not be saturated when the secondary current is 1.0Apk-pk (10kHz). What do you think about using toroidal core to build the transformer?

Do you think the above idea will work?

Please kindly advise if you know a better power amplifier with high supply voltages and high output current.

Thanks.

[NiHaoMike]

A MOSFET H bridge using current feedback PWM at 100kHz or so should work nicely for direct driving the load. A 10Hz transformer is not going to be very practical. And unless you need really low noise, PWM techniques will result in a cheaper circuit due to less power dissipation. (If low noise rules out PWM, vary the supply voltage with the peak of the output voltage to reduce dissipation.)

Note that most modern motor drives are PWM AC current sources.

[scrat]

If steps are allowed, and precision is needed, I'd go for a multiple output transformer, switching between the various outputs depending on frequency (= depending on output voltage). At least for the linear solution.

A MOSFET H bridge using current feedback PWM at 100kHz or so should work nicely for direct driving the load. A 10Hz transformer is not going to be very practical. And unless you need really low noise, PWM techniques will result in a cheaper circuit due to less power dissipation. (If low noise rules out PWM, vary the supply voltage with the peak of the output voltage to reduce dissipation.)

Note that most modern motor drives are PWM AC current sources.

A current control on (~100kHz) PWM modulation could be a solution, with a good filtering. Maybe, even here, choosing between multiple outputs of a tranformer would be better for noise reduction and lower voltage components.
Otherwise, you could change DC bus voltage according to frequency.

As Mike says, with PWM the transformer can be smaller.

[Zad]

To be honest, it just sounds like a Class B audio power amplifier. If you need less distortion then bias is into Class AB. If you need no distortion at all, and don't mind heating the house then go full Class A.

If you need the ultimate efficiency then use a Class D ampl like this http://focus.ti.com/docs/prod/folders/print/tas5630.html It will drive two independent 300W loads at 50V (600W in push/pull) and up to 80kHz. Full overload protection and reporting.

[scrat]

To be honest, it just sounds like a Class B audio power amplifier. If you need less distortion then bias is into Class AB. If you need no distortion at all, and don't mind heating the house then go full Class A.

If you need the ultimate efficiency then use a Class D ampl like this http://focus.ti.com/docs/prod/folders/print/tas5630.html It will drive two independent 300W loads at 50V (600W in push/pull) and up to 80kHz. Full overload protection and reporting.

The only (!) difference I see is the fact you need to control current, not voltage, with an impedance which grows one thousand times.

[NiHaoMike]

The only (!) difference I see is the fact you need to control current, not voltage, with an impedance which grows one thousand times.

Just like a brushless DC servo motor drive, except at about 50 times higher maximum frequency. (Assuming the same output torque, brushless DC motors have an effective impedance that increases with speed/frequency due to back EMF.) A current feedback PWM should work nicely. It might not even have to be a proper PWM - a Schmitt trigger comparator in a current feedback loop might be good enough.

To be honest, it just sounds like a Class B audio power amplifier. If you need less distortion then bias is into Class AB. If you need no distortion at all, and don't mind heating the house then go full Class A.

At those frequencies and power levels, a PWM design is highly preferred unless there's a good reason not to.

A friend of mine has wrote about this before.
http://www.diyaudio.com/forums/chip-amps/172570-want-build-300-watt-stereo-power-amp-3.html#post2286906

Above a certain power level, the overall initial cost is lower for a switching design. It used to be 100W, it's now more like 10W. (That doesn't even count the lower operating costs, which would push the point even lower.)
...
FWIW, a power electronics professor (Dr. Ehsani) at Texas A&M thinks of operating a power transistor in "Tiffany Yep" mode as like going full throttle in a car while riding the brakes to control speed. It's just not a good way to do it!

In other words, PWM makes more efficient use of your transistors.

[scrat]

The only (!) difference I see is the fact you need to control current, not voltage, with an impedance which grows one thousand times.

Just like a brushless DC servo motor drive, except at about 50 times higher maximum frequency. (Assuming the same output torque, brushless DC motors have an effective impedance that increases with speed/frequency due to back EMF.) A current feedback PWM should work nicely. It might not even have to be a proper PWM - a Schmitt trigger comparator in a current feedback loop might be good enough.

Excuse me...As I'm involved into electric drives, I can't help specifying... Increasing impedance in every motor is due to inductance, not to back-EMF, but you're right about the fact that voltage increases mainly because of flux linkage.
I would be careful considering an electric drive inverter as a good example of current source: it can control quite good the average current, not the instant value. By using PWM (either from a modulator or an hysteris comparator) you are switching a high voltage even at low frequencies (since you need it for going at the high end of frequency range). This will likely produce much noise, which COULD be unacceptable if, for example, the application is signal generation or something related to a measure. This is the reason why some instrumentation generators are still working in linear mode.

What I'm trying to say is... IMHO, PWM can be better or worse depending on the application, and the same for solutions with or without different DC bus voltage values or transformer ratios for different frequency ranges.

[NiHaoMike]

Varying the supply voltage depending on output peak voltage can give you a greater dynamic range. Some of the higher powered TI Hybrid Digital amplifiers vary their supply voltage depending on volume setting. It is also done a lot in motor drives where there already is a voltage converter before the inverter.
http://techno-fandom.org/~hobbit/cars/ginv/VH.html

Interesting things happen in the 30 - 40 MPH region, especially in EV mode.
I know from some other observations that the inverter drive signals for MG2
move from a multiple-kilohertz PWM regime into simply switching 3-phase square
waves at the motor's native electrical rotation speed, because it's more
efficient and the motor is turning fast enough to smooth out any torque ripple
that would produce.  But overall applied motor current can still be regulated
smoothly!  How?  By using a variable boost voltage.  In this speed range I see
VH rising and falling corresponding to my go-pedal demand, with its lowest
baseline creeping up a bit as I head toward 40 MPH and MG2's own peak output
rises sufficiently.  It's almost like having the switching behave like a
brush commutator, simply leading the electrical rotation angle by 60 or more
degrees, and regulating motor speed via applied voltage like it was a big ol'
toy-train rheostat.

[onemilimeter]

A MOSFET H bridge using current feedback PWM at 100kHz or so should work nicely for direct driving the load. A 10Hz transformer is not going to be very practical. And unless you need really low noise, PWM techniques will result in a cheaper circuit due to less power dissipation. (If low noise rules out PWM, vary the supply voltage with the peak of the output voltage to reduce dissipation.)

Note that most modern motor drives are PWM AC current sources.

Thanks for your suggestion. The PWM switching noise may affect my application. However, if the PWM switching frequency is 100kHz or higher, then it should be OK as the frequency range of interest for my application is 0~20kHz.

If you read my previous posts, you may find out that in my application, at 10kHz, the impedance of the load will be [0.5ohm + 2*pi*(10kHz)*(4mH)] = 251.83ohm. Thus, to push a 1.0Apk-pk sinusoidal current through the load, the circuit should be able to supply 251.83Vpk-pk or higher. Do you think we can achieve this with the "MOSFET H bridge" you suggested?

Cheers

[onemilimeter]

A current control on (~100kHz) PWM modulation could be a solution, with a good filtering. Maybe, even here, choosing between multiple outputs of a tranformer would be better for noise reduction and lower voltage components.
Otherwise, you could change DC bus voltage according to frequency.

As Mike says, with PWM the transformer can be smaller.

Thanks. Is it difficult to design the filtering stage?

100kHz MOSFET PWM H-Bridge --> Filtering --> Step-up Transformer --> Load


In my application, the impedance changes with supply current frequency:
10Hz: 0.75 ohm
100Hz: 3.01 ohm
1000Hz: 25.63 ohm
10000Hz: 251.82 ohm
20000Hz: 503.15 ohm

Since the load, by nature, is lower at low frequency range, the filtered output of the PWM H-Bridge can be directly connected to the load. For higher frequency range, a transformer is used to step-up the filtered output voltage of the PWM H-Bridge. What do you reckon?

[onemilimeter]

It is also done a lot in motor drives where there already is a voltage converter before the inverter.
http://techno-fandom.org/~hobbit/cars/ginv/VH.html

Thanks for this link. Since the current in my application is relatively low compared to those HEV drives, it's possible to switch the power MOSFET at higher frequency, e.g. 100kHz.

[Zero999]

Anyway why do you want to do this? It seems like a very odd thing to do.

Have you looked into class D power amplifier ICs?

[qno]

Looks like you need a IGBT module.

Check out this one

http://www.fujielectric.com/device/semi/products/powerdevices/2pack.html