Can a filter inductance for an inverter be too big?

[donavan]

Hello,

i have a H-bridge voltage inverter that takes a 50V battery and creates a pwm sinus with about 15A, at the end there is a filter inductor between 1-10mH. My question is now, can the output filter inductance be too big? The scenario i have in mind is, when the Mosfet closes and the inductor tries to keep the current flowing and the drain voltage of the Mosfet should go down and exceed the rated drain-source voltage which destroys the Mosfet.

could this become a problem?

[MrAl]

Hello,

i have a H-bridge voltage inverter that takes a 50V battery and creates a pwm sinus with about 15A, at the end there is a filter inductor between 1-10mH. My question is now, can the output filter inductance be too big? The scenario i have in mind is, when the Mosfet closes and the inductor tries to keep the current flowing and the drain voltage of the Mosfet should go down and exceed the rated drain-source voltage which destroys the Mosfet.

could this become a problem?

Hello there,

In a DC to DC converter the inductor helps filter the pulsing that comes from the transistor stage.  It works with the filter capacitance, and the two form a filter that helps smooth out the DC.  The inductance can not be too small or it will allow peak currents that are too high and that causes an RMS current into the cap that can be too high for the rating of the cap which in turn can blow out the cap.  Thus the choice of inductor plays with the choice of cap to some degree.
Because the inductor works in conjunction with the output cap and the ESR of the inductor is always small, that means there is also a natural frequency of oscillation that will occur and that act to prevent perfect control of the output level in a normal feedback arrangement. The inductor energy is proportional to the inductance L/2*i^2 so the larger the inductor the more energy it stores, and this wants to dissipate even in the event that the control circuit wants to reduce the output due to changing environmental conditions.  This means the speed of response is reduce with larger inductor values.  The tradeoff is to get decent response as well as decent filtering action.

An AC converter however is usually of low frequency like below 100Hz, which means speed of control usually isnt as much of a concern.  However, as you noted (although not exactly how it works), when the transistor turns on it has to take the full inductor current at the time it turns on (not quite as much about voltage).  That's not really too much of a problem though because the transistor has been doing that all along, since the converter was first turned on.  The problem comes in if we run into a condition where the output load current increases beyond the normal load current, and then when the transistor turns on it may have to deal with a rising current, and there is no way to turn it off because the inductor keeps pumping out current until the energy is dissipated.  This can be more than normal, but there is no way for the converter to stop this because the control circuit does not do that.  This means that the bigger the inductor the longer it will take the converter to respond to an overload.  Depending on how long this takes the load may be damaged.

So the trade off is filtering action vs speed of response.  The larger the inductor the better the filtering and the less RMS current the output cap has to put up with, but the speed of response is increased with larger inductance.

[Zero999]

Hello,

i have a H-bridge voltage inverter that takes a 50V battery and creates a pwm sinus with about 15A, at the end there is a filter inductor between 1-10mH. My question is now, can the output filter inductance be too big? The scenario i have in mind is, when the Mosfet closes and the inductor tries to keep the current flowing and the drain voltage of the Mosfet should go down and exceed the rated drain-source voltage which destroys the Mosfet.

could this become a problem?

You're talking about an inverter which PWMs the DC input voltage to generate a sine wave? I don't think I've seen such a beast. The output voltage will be able 35VAC. Don't tell me it has a massive transformer to step that up to 120V or 240V?

That's not normally how it's done. A DC-DC converter, consisting of a high frequency oscillator (50kHz to 500kHz), a high frequency transformer and a rectifier boosts the voltage to 170VDC or 340VDC (depending on whether it's a 120V or 240V inverter) first and an H-bridge and filter converts it to mains frequency AC.

Any energy stored in the inductors in the filter is transferred to the output, rather than absorbed in the transistor. It is not the same as switching a relay or the leakage inductance in a transformer, which must absorbed by a free-wheeling diode or a snubber network.

[donavan]

You're talking about an inverter which PWMs the DC input voltage to generate a sine wave? I don't think I've seen such a beast. The output voltage will be able 35VAC. Don't tell me it has a massive transformer to step that up to 120V or 240V?

the output is the 35VAC pwm which is filtered with the inductance, no transformer.

Any energy stored in the inductors in the filter is transferred to the output, rather than absorbed in the transistor. It is not the same as switching a relay or the leakage inductance in a transformer, which must absorbed by a free-wheeling diode or a snubber network.

how does the inductor behave when his voltage source is cut off? doesnt the energy stored in the inductor create a voltage to counter the current decrease?

[Zero999]

That's interesting, what do you use a 35VAC inverter for?

[sibeen]

You're talking about an inverter which PWMs the DC input voltage to generate a sine wave? I don't think I've seen such a beast.

 

Sorry, had to grin with this comment, and I'm not trying to have a go, Hero. Every UPS (wild hand waving generalisation) up until about 15 to 20 years ago did basically that. This new fangled DC/DC converter before the inverter - why that's just magic

[tatus1969]

how does the inductor behave when his voltage source is cut off? doesnt the energy stored in the inductor create a voltage to counter the current decrease?

There's nothing to be worried about. There is always a current path for the inductor, either one of the two H bridge switches, or their (parasitic or deliberate) reverse diodes. The inductor's magnetic energy is dumped into the DC bus capacitance, which must be large enough to keep the voltage increase small enough for the circuit. But in real life, I would expect that most of that energy is delivered to the load anyway.

[T3sl4co1l]

Too big?  Sure -- if you keep increasing inductance, without adjusting anything else, your control loop goes unstable.

Your control loop is regulating inductor current, riiiiiiiight?...

Tim

[MrAl]

Hello,

i have a H-bridge voltage inverter that takes a 50V battery and creates a pwm sinus with about 15A, at the end there is a filter inductor between 1-10mH. My question is now, can the output filter inductance be too big? The scenario i have in mind is, when the Mosfet closes and the inductor tries to keep the current flowing and the drain voltage of the Mosfet should go down and exceed the rated drain-source voltage which destroys the Mosfet.

could this become a problem?

You're talking about an inverter which PWMs the DC input voltage to generate a sine wave? I don't think I've seen such a beast. The output voltage will be able 35VAC. Don't tell me it has a massive transformer to step that up to 120V or 240V?

That's not normally how it's done. A DC-DC converter, consisting of a high frequency oscillator (50kHz to 500kHz), a high frequency transformer and a rectifier boosts the voltage to 170VDC or 340VDC (depending on whether it's a 120V or 240V inverter) first and an H-bridge and filter converts it to mains frequency AC.

Any energy stored in the inductors in the filter is transferred to the output, rather than absorbed in the transistor. It is not the same as switching a relay or the leakage inductance in a transformer, which must absorbed by a free-wheeling diode or a snubber network.

Hi there,

Back in the 80's that's the only way we did it, with PWM straight from the DC buss and possibly large output transformer depending on the power rating of the unit.  Some of these transformers weight in so heavy they were very hard to move around.  For a 30kWatt 3 phase unit there were three of them, a massive cabinet to hold everything.
The solar version worked right off of the million dollar solar array, converting the DC to PWM sine with some filtering.
I had actually suggested doing some of it with a front end DC to DC converter, but the votes came in negative against it with the reason that we only wanted to handle the power once.  That was important for very high efficiency with the transistors that were available back then.

[T3sl4co1l]

Back in the 80's that's the only way we did it, with PWM straight from the DC buss and possibly large output transformer depending on the power rating of the unit.  Some of these transformers weight in so heavy they were very hard to move around.  For a 30kWatt 3 phase unit there were three of them, a massive cabinet to hold everything.
The solar version worked right off of the million dollar solar array, converting the DC to PWM sine with some filtering.
I had actually suggested doing some of it with a front end DC to DC converter, but the votes came in negative against it with the reason that we only wanted to handle the power once.  That was important for very high efficiency with the transistors that were available back then.

There was a brief period around the 70s, maybe including the late 60s or early 80s, I'm not sure -- where BJTs were the preferred output device for power switching applications.  Stinkin' big triple-darlington modules, and 1us switching speed if you were lucky.  But at least they don't stay stuck on like the SCRs do...  (Which are still useful for some applications, but it's largely for very high power only, as IGBTs have taken over even many low power (1kW or less) cases.)

Or if you were using MOSFETs, they would've been relatively high Rds(on) for their price, and size and Qg and all that, I think?  Or maybe early IGBTs too, but the ones where you need to be careful about latchup.

Tim

[MrAl]

Back in the 80's that's the only way we did it, with PWM straight from the DC buss and possibly large output transformer depending on the power rating of the unit.  Some of these transformers weight in so heavy they were very hard to move around.  For a 30kWatt 3 phase unit there were three of them, a massive cabinet to hold everything.
The solar version worked right off of the million dollar solar array, converting the DC to PWM sine with some filtering.
I had actually suggested doing some of it with a front end DC to DC converter, but the votes came in negative against it with the reason that we only wanted to handle the power once.  That was important for very high efficiency with the transistors that were available back then.

There was a brief period around the 70s, maybe including the late 60s or early 80s, I'm not sure -- where BJTs were the preferred output device for power switching applications.  Stinkin' big triple-darlington modules, and 1us switching speed if you were lucky.  But at least they don't stay stuck on like the SCRs do...  (Which are still useful for some applications, but it's largely for very high power only, as IGBTs have taken over even many low power (1kW or less) cases.)

Or if you were using MOSFETs, they would've been relatively high Rds(on) for their price, and size and Qg and all that, I think?  Or maybe early IGBTs too, but the ones where you need to be careful about latchup.

Tim

Hi,

Yeah, back then the big 100 ampere Darlington modules were used extensively, but that actually came later.  Before that we used all metal high current transistors with stud mount, and they could be 300 dollars each back then.  When they blew out it was a real loss.  Much later the MOSFET's came in and they were fairly decent.

[donavan]

Too big?  Sure -- if you keep increasing inductance, without adjusting anything else, your control loop goes unstable.

Your control loop is regulating inductor current, riiiiiiiight?...

Tim

thats true, i should consider this.

yeah it regulates the current

how does the inductor behave when his voltage source is cut off? doesnt the energy stored in the inductor create a voltage to counter the current decrease?

There's nothing to be worried about. There is always a current path for the inductor, either one of the two H bridge switches, or their (parasitic or deliberate) reverse diodes. The inductor's magnetic energy is dumped into the DC bus capacitance, which must be large enough to keep the voltage increase small enough for the circuit. But in real life, I would expect that most of that energy is delivered to the load anyway.

ok thanks, then i guess i have nothing to worry about.
only way i think it could be a problem if the h-bridge mosfet that connects output to battery 50V gets destroyed the mosfet diode cant transfer the voltage spike and the lower mosfet also gets destroyed by it.

i also tried to simulate it in ltspice and i couldnt get the voltage between h-bridge output and inductor to spike, so i guess it´s fine

thanks for all your answers

[Phoenix]

Too big?  Sure -- if you keep increasing inductance, without adjusting anything else, your control loop goes unstable.

Really? I would say the opposite - increasing the inductance will only ever stabilise your control loop in a linear sense. Non-linearly you may hit wind-up problems if your anti-windup is not in check.

I see a few issues with increasing inductance
1. Cost and size
2. Loss - more copper loss, more core loss
3. Frequency dependence of the inductor. Larger inductors tend to roll off inductance and become capacitive at lower frequencies - this may have EMI consequences.
4. Dynamic response - larger inductor means slower transient response (assuming the PI is well tuned and reaches output saturation).

[T3sl4co1l]

Really? I would say the opposite - increasing the inductance will only ever stabilise your control loop in a linear sense. Non-linearly you may hit wind-up problems if your anti-windup is not in check.

I see a few issues with increasing inductance
1. Cost and size
2. Loss - more copper loss, more core loss
3. Frequency dependence of the inductor. Larger inductors tend to roll off inductance and become capacitive at lower frequencies - this may have EMI consequences.
4. Dynamic response - larger inductor means slower transient response (assuming the PI is well tuned and reaches output saturation).

You kind of answered your question, without realizing it: slower transient response means poles closer to the right half plane.   That is, while holding the controller constant.

If you adjust the controller to maintain critical damping, then #4 will be correct.  The transient response (the time constant, specifically) will change proportionally.

Keeping the output capacitance constant (assuming the application is a buck converter, class D amplifier, or the like) also raises the filter impedance, meaning more voltage overshoot on transients, even in the absence of a controller (i.e., fixed PWM).  A "home-grown" illustration of this apparently is here:
https://www.eevblog.com/forum/renewable-energy/mppt-noise-reduction/msg1236325/#msg1236325

Keeping switching frequency and core material constant, core loss will definitely be lower, however.  Less ripple, less reactive power, less delta B.  (Assuming, of course, #3 doesn't happen, which can be the case for an extremely too-big inductor!)

Tim

[Phoenix]

You kind of answered your question, without realizing it: slower transient response means poles closer to the right half plane.   That is, while holding the controller constant.

I respectfully disagree. Unfortunately being at work I don't have the time or tools to put together a more thorough response....

--

But you're right when you say ripple will be lower, my bad.