3
\$\begingroup\$

Because I was worried about the gate inrush current of a p-mosfet, because of a lacking gate-resistor, I decided to measure the switching times with an oscilloscope. I also considered another transistor with less input capacitance, which also may help to limit the inrush current.
However, I do not understand the measurements. I would expect that under similar conditions, the mosfet with less input capacitance would switch faster. This is not the case. Can an expert comment on this?

The setup:
- source of p-mosfet: 11.2 Volt
- no gate resistor, switching with 11.2 Volt from the output of a level shifter CD4504BPWR
- load between drain and ground is a 30 ohm resistor
- there is a strong pull-up resistor (680 ohm) from the gate to 11.2V; could be less... I also tried 10k with similar results.

schematic

simulate this circuit – Schematic created using CircuitLab

The two p-mosfets considered:
- red: Onsemi NTJD4152PT1G
- yellow: Infineon BSD223P

Figure 1: The switching speed/capacitance characteristics of 'red':

The switching speed/capacitance of 'red'

Figure 2: The switching speed/capacitance characteristics of 'yellow':

The switching speed/capacitance of 'yellow'

Finally, measurements with scope:

Figure 3: Overview of total switching:

enter image description here

==

Figure 4: Zoomin for rise-time:

enter image description here

==

Figure 5: Zoomin for fall-time:

enter image description here

So, why is red faster than yellow?
I would expect the opposite.
The specs in the datasheet are in favour of yellow...

Additions 16 nov 2019

Multiple people suggested to measure the gate vs the drain.
I have only a two channel scope, so I can not combine this in one measurement.

Figure 6: Gate (yellow) vs Drain (red) for switch-on of the NTJD4152PT1G:

enter image description here

==

Figure 7: Gate (yellow) vs Drain (red) for switch-on of the BSD223P:

enter image description here

==

Figure 8: Gate (yellow) vs Drain (red) for switch-off of the NTJD4152PT1G:

enter image description here

==

Figure 9: Gate (yellow) vs Drain (red) for switch-off of the BSD223P:

enter image description here

To be honest, these additional measurements raises more questions for me, instead of clarifying things.....

  • It looks that the threshold voltage of both mosfets is similar
  • I do not understand why the voltage in the Miller plateau raises for the BSD223P when it switches on.

Note that the rise timed of both mosfets are quite similar for the switch-on; the BSD223P starts later.

\$\endgroup\$
  • 1
    \$\begingroup\$ 1) please include a schematic ! 2) you appear to be driving the MOSFETs using a CD4504, this is a very old IC and it uses very old CMOS technology which means the inverters in it cannot deliver much current. In the order of less than 5mA per inverter output. That means inrush current is no issue at all the CD4505 cannot deliver enough current. Regarding the scope traces: it is unclear what is shown, include the schematic and show what you probed. \$\endgroup\$ – Bimpelrekkie Nov 13 '19 at 13:01
  • \$\begingroup\$ 1) done :) 2) yes, the CD4504 is an old device, but I could not find a better device that could level shift from 3.3 to 12 Volt. Because the CD4504 can handle only a limited output current, I am wondering whether charging the mosfet gate may destroy the CD4504 device, because the charging the mosfet gate may draw too much current. That is basically why I started measuring. \$\endgroup\$ – Billy Nov 13 '19 at 13:40
  • \$\begingroup\$ I am wondering whether charging the mosfet gate may destroy the CD4504 device No need to worry about that, the outputs of the CD4505 are so weak that they're basically short-circuit proof. You can connect the outputs to VDD or GND and no damage is done. Only a few mA will flow and that's not enough to do any damage. \$\endgroup\$ – Bimpelrekkie Nov 13 '19 at 13:47
  • \$\begingroup\$ Nice to know! I destroyed e.g. Beaglebone Blacks by drawing too much current from their output pins. Maybe old technology is also less sensitive / more robust. \$\endgroup\$ – Billy Nov 13 '19 at 15:29
  • \$\begingroup\$ Why not scope the signal at the MOSFET Gate? This might give you a better idea of what is happening. \$\endgroup\$ – Bruce Abbott Nov 13 '19 at 18:13
2
\$\begingroup\$

If you look at the data sheet for the CD4504 you will see that it has a high to low fall time of circa 400 ns. From this you could make a first order approximation that the output changes from 11 volts to 0 volts at a rate of about 36 ns/volt.

If you look at the data sheet for the NTJD4152P you'll see that it starts to "activate" (turn on) at a gate source threshold voltage of circa -1.5 volts. The BSD223's equivalent threshold voltage (for the same drain current) is about -2.2 volts (reading a bit between the lines).

Because your drive voltage is so sluggish, the input capacitances are hardly playing any role and it comes down to (largely) which device has the lower threshold voltage and clearly, the NTJD4152P will win that battle and activate or deactivate more quickly.

If you'd have plotted the drive voltage alongside the MOSFET drain voltages you would probably be in a better position to see the bigger picture.

\$\endgroup\$
  • \$\begingroup\$ Thanks for the insight. The gate threshold voltage ranges look quite similar in both datasheets, but the graphs do not.... Minor remark: The CD4504 high to low fall time of 400 ns is the propagation time. I would expect that in this comparison the transition time of 50 ns is the relevant parameter, which translates to about 5 ns/volt. This does of course not change the point made to look more carefully at the gate threshold voltage graphs. \$\endgroup\$ – Billy Nov 13 '19 at 15:26
  • \$\begingroup\$ @Billy yes, my mistake but why not clear this up and plot the drive waveform alongside the output waveform for each mosfet? \$\endgroup\$ – Andy aka Nov 13 '19 at 15:33
  • \$\begingroup\$ Smaller turn-off voltage should cause it to turn off slower, not faster, as the drive voltage has to get closer to the supply rail. Output voltage transition occurs during the Gate charge plateau, which is similar for both devices. But the BDS223 has much lower Gate charge so it should turn off faster. Yet we are told that it is slower. If this is true then something else is going on. \$\endgroup\$ – Bruce Abbott Nov 13 '19 at 19:08
  • \$\begingroup\$ @BruceAbbott it’s a good point. The lower threshold device certainly turns on in much less time but turn off for both is very much closer so, maybe something else is afoot. \$\endgroup\$ – Andy aka Nov 13 '19 at 22:05
  • \$\begingroup\$ @Andyaka Just added the suggested waveforms. It seems that the voltage thresholds when the mosfets start switching are about the same. So, I think it must be something different. \$\endgroup\$ – Billy Nov 16 '19 at 15:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.