For class-D hard switching I experimented with some MOSFETs and measured some rise and fall times of the waveform. Generally the lower parasitic capacitance of the FETs the faster the flanks become. No surprise.
For fun I took a pair of BSS127 that have very small capacitance and tho not power houses I was expecting rather fast switching when not loaded. The pcb has some C but likely less than 5pF in the FETs proximity.
The rise and fall times were pretty slow.
My reference being BSZ22DS20 slews 100V in ca 3-4ns. The BSS127 > 20ns.
Loading the switching pair w 100pF indicates BSS127 can source and sink about 100mA.
So what other than parasitic capacitance lìmits the slew? The low transconductance?
Thnx
For fun I took a pair of BSS127 that have very small capacitance and tho not power houses I was expecting rather fast switching when not loaded. The pcb has some C but likely less than 5pF in the FETs proximity.
The rise and fall times were pretty slow.
My reference being BSZ22DS20 slews 100V in ca 3-4ns. The BSS127 > 20ns.
Loading the switching pair w 100pF indicates BSS127 can source and sink about 100mA.
So what other than parasitic capacitance lìmits the slew? The low transconductance?
Thnx
Useful doc here:
https://www.ti.com/lit/ml/slua618a/slua618a.pdf?ts=1611006108625
With further reading links within
https://www.ti.com/lit/ml/slua618a/slua618a.pdf?ts=1611006108625
With further reading links within
The long fall time is influenced by the threshold voltage. devices with a low threshold take ages to turn off. you need to pull the charge out of a gate when the gate drive is zero volts, but charge in the gate is at 1.5V for example. 4V threshold gates have less turn-off delay. another option is to drive to gate to a negative voltage to reduce the turn-off delay.
Ok thnx. I just find it odd that such a small MOSFET is that much slower. How can it inhibit more charge than a beefy power FET? I would think the turn on and off be quick as hell and then only the rdson limits the slew of the rising and falling flanks.
When selecting MOSFET for high speed switching I usually select a low C-parasite part. But I think gm and rdson needs to be a certain amount before the C-parasite dominates.
When selecting MOSFET for high speed switching I usually select a low C-parasite part. But I think gm and rdson needs to be a certain amount before the C-parasite dominates.
Even with the low cap, if you don't make the gate negative, the charge can't get out. You need to actively pull down the gate. As explained above.
Jan
Jan
Hmmm. Lost in translation? I am using a very fast gate driver. The exact same as on the big power MOSFETs. The circuit and components are identical. I only swap various MOSFETs in the circuit. To test each MOSFET and compare.
just run the sims and change the negative level of the gate drive. I'm working with 13.56Mhz wireless charging, there the switch-off delay really hits hard.
the reverse capacitance takes it's tall on the driver current capability, so selecting a fet with a 20x lower reverse than gate is fine, like the BSZ22. the BSS127 fall time is large as we only have 10mA of drain current to drain the output capacitance.
So how to id a slow FET on datasheet when capacitance is low? What other factors to look for? When gm is very small, as w BSS127, does that indicate a weak and thus slow device?
I see on the datasheet the BSS127 has really slow and long t-off and t-on. But I still thought when finally on and off the MOSFET would slew fast.
I guess it makes sense we need some power to slew fast and small signal MOSFETs just dont have the muscle. Basic physics.
I mostly wonder if there is another parameter to select MOSFETs other than rdson and C-parasites when highest speed is important.
I see on the datasheet the BSS127 has really slow and long t-off and t-on. But I still thought when finally on and off the MOSFET would slew fast.
I guess it makes sense we need some power to slew fast and small signal MOSFETs just dont have the muscle. Basic physics.
I mostly wonder if there is another parameter to select MOSFETs other than rdson and C-parasites when highest speed is important.
the other factor is gate resistance, how fast can you drain the charge out of the gate or into it. I found it to my peril that Gan fet have a very low gate resistance, or a high Q gate capacitance. just a minimum of gatedriver trace length can already resonate with the fet gate capacitance, creating a lot of EMC during switching. while the gate driver source resistance is usually sufficient for mos fets not to resonate, the GAN fet always needed a series resistance to stop the resonance. GAN fets are so much faster in switching..