What Mosfets are Good for 1 Mhz, 100W Power Stage?

Hello Johny,

Thanks for your interest in TI FETs. Am I correct in assuming the FETs will be in a half-bridge configuration? For a 24V DC max input, at least a 30V FET would be required although you might consider a 40V device to make sure there is enough margin for switching spikes and transients. How are the FETs being driven? Are you using a gate driver IC and if so, what is the gate drive voltage and drive capability (current or resistance)? TI has a number of Excel-based FET selection tools but nothing specific to your application. I'm thinking the FOC (sinusoidal) motor drive control tool might give a reasonable estimate on power loss. I'd recommend using FETs in a 5x6mm SON package as that gives us the most options for 30V or 40V FETs. In a 30V device, I'd start with the CSD17579Q5A and in 40V, I'd go with the CSD18514Q5A. Those are the most cost effective devices at those breakdown voltages and there are upgrades to higher performance FETs in the same package if the power loss is unacceptable. I'm including a link below to the FOC motor drive FET selection tool. Please let me know if I can be of further assistance.

https://www.ti.com/lit/zip/splr005

Best Regards,

John Wallace

TI FET Applications

Hello Johny,

Following up to see if this has resolved your issue. Please let me know.

Thanks,

John

John Wallace1 said:

Thanks for your interest in TI FETs. Am I correct in assuming the FETs will be in a half-bridge configuration? For a 24V DC max input, at least a 30V FET would be required although you might consider a 40V device to make sure there is enough margin for switching spikes and transients. How are the FETs being driven? Are you using a gate driver IC and if so, what is the gate drive voltage and drive capability (current or resistance)? TI has a number of Excel-based FET selection tools but nothing specific to your application. I'm thinking the FOC (sinusoidal) motor drive control tool might give a reasonable estimate on power loss. I'd recommend using FETs in a 5x6mm SON package as that gives us the most options for 30V or 40V FETs. In a 30V device, I'd start with the CSD17579Q5A and in 40V, I'd go with the CSD18514Q5A. Those are the most cost effective devices at those breakdown voltages and there are upgrades to higher performance FETs in the same package if the power loss is unacceptable. I'm including a link below to the FOC motor drive FET selection tool. Please let me know if I can be of further assistance.

Hi John, Thxnks for your response.

My question isn't resolved yet. 

• I assume half-bridge, or possibly quarter-bridge. 
• The FETs will be driven with whatever is best. I assume a gate driver. It will be whatever gate-driver is the best match for the chosen FET. 
• Thanks for your package recommendation. 
• A motor-drive selector might help, tho' i wonder if such a tool might not include high-speed devices. 
• Any thoughts on clock speed?

Thx!

Hi Johny,

Thanks again for your interest in TI FETs. TI has a large selection of gate drivers but I am only responsible for our discrete and power block MOSFETs. I don't understand enough about the application and therefore I don't have a recommendation on the clock frequency (PWM/switching frequency of the FETs). My guess would be 10x the sinusoidal input to the PWM generator should be adequate. The FETs I recommended in my previous response can be hard switched up to ~1MHz. At higher switching frequencies, switching losses become dominant and overall power loss may exceed the package capability. Also, power MOSFETs generate high frequency noise when hard switched. I'm not sure how to estimate the power losses and the closest tool I have is the sinusoidal motor drive tool. The tool has a "limit" of 100kHz max PWM frequency but it still works if you put in 1000kHz - the cell just turns red.

There is a more general synchronous rectifier FET selection tool that just uses rms current and switching frequency to estimate the FET losses. It's available at the link below. Let me know what else I can do to help you out.

Thanks,

John

https://www.ti.com/tool/SYNC-RECT-FET-LOSS-CALC

John Wallace1 said:

I don't understand enough about the application and therefore I don't have a recommendation on the clock frequency

This application is a power supply, feeding an inductive load, up to 5V @ 5A. 

The load requires pure sine power. The sine wave will be 100 kHz. It would pulse-width modulate a higher freq square wave (or more correctly, be compared to higher freq triangle to generate the PWM). 

I assume that the PWM clock speeds depends on the filters. What's the lowest clock speed that we can easily filter it out, to produce a clean sine? I assume that depends on practical filter design. 

Re hard switching, why can't we soft-switch?

At what freq will those FET's NOT have excessive losses?

Are there FET's that won't have so much losses at 1 Mhz?

Maybe we can lower the PWM freq to 5x or 2x the sine freq. That would require lower-freq filtering. 

Thx for recommending the synchronous rectifier FET selection tool. You suggested that just for FET selection, and not because you're recommending an SR for my application, correct?

Thx!

Hi Johny,

I have forwarded your system related question onto TI's power design services team to get their inputs. I'll update you as soon as I receive their response. Soft switching is possible but adds components, cost and complexity to the circuit. I'm including a link below to an article that gives a quick overview of hard switching vs. soft switching. Switching loss is proportional to switching frequency and is a function of the FET parasitic capacitance/charge and external gate drive circuit. Reducing switching frequency results in a proportional reduction in switching loss. Yes, 300kHz or 500kHz would be better for switching loss. Conduction loss and switching loss are not independent of each other. For a given voltage rating and technology, a lower on resistance FET uses a larger die size which is higher capacitance/charge (and cost) and results in more switching loss. The FETs I recommended are higher resistance/lower charge and relatively low cost compared to some of our lower resistance devices. Yes, the SR tool I pointed you to is just to give you a way to compare FET losses if you know the drain voltage, rms current, switching frequency and gate drive voltage and is not dependent on a particular circuit topology.

https://www.powerelectronicsnews.com/power-supply-design-notes-hard-switching-and-soft-switching-to-reduce-losses/

Thanks,

John

Hi Johny,

I am going to reassign this thread to TI's Power Design Services team to address your system questions.

Thanks,

John