Is IRF9640 the best choice for a high-side switching topology?

[kt315]

P-channel vs N-channel - better is N-channel,

in any case. i had answered. we see P-channel in Bara circut just that is a mod, PNP transistor is changed on it.
also early White's PI was done on PNP, that was changed, like *Eric Foster mod*, on P-channel MOSFET.
so no reason to use P one if you can use N one.

[Vadim18]

1.Find better MOSFET
IRFI9634G 250V 1 Ohm

2.Use MOSFET combination

2x in parallel IRF9310. As result you get 400V and 3.5 Ohm

3.Change TX circuitry
use isolated power source for gate driver via transformer, replace IRF9640 with IRF740 and use optocoupler to drive it

OR (not a good option)

with little adaptation for MOSFET, drive it directly true transformer like all flyback transistor drivers do in old CRTs
BUT maximum pulse width will be limited by your transformer

[Emersonpaz]

I'm using Toshiba's K2545 in all my RDSon 0.9ohms 600V projects

[CrizzyD]

1.Find better MOSFET
IRFI9634G 250V 1 Ohm

2.Use MOSFET combination

2x in parallel IRF9310. As result you get 400V and 3.5 Ohm

3.Change TX circuitry
use isolated power source for gate driver via transformer, replace IRF9640 with IRF740 and use optocoupler to drive it

OR (not a good option)

with little adaptation for MOSFET, drive it directly true transformer like all flyback transistor drivers do in old CRTs
BUT maximum pulse width will be limited by your transformer

1. This MOSFET looks like a good option. Better VDS but slightly lower avalanche energy ratings and double the RDSon (No idea how much an ohm or two would affect the thermals)
2. Googling the part number gave https://www.infineon.com/dgdl/Infine...5356110a7d1d95 which doesn't match what you're stating at all?
3. I intend to do this for the next revision in the distant future. The current version is powered from a 24VDC SMPS and the recovery curve is digitised by an ADC of 0V -> +3.3V so everything was kept with positive voltages. Next version I'll create some negative voltage rails and go from there for op amps and the transmit voltage. Presently though, I'm stuck with the design I have. My only option is to change the part.

[CrizzyD]

I'm using Toshiba's K2545 in all my RDSon 0.9ohms 600V projects

This appears to be an N type? Not sure how suitable it would be for a high-side configuration without a lot of extra work. Plus, it doesn't look all too common either.

[Altra]

A good p channel is the IXTA10P50P. -500v, 10A, RDSon less than 1 ohm.

[CrizzyD]

A good p channel is the IXTA10P50P. -500v, 10A, RDSon less than 1 ohm.

Since starting this thread, the IXTP10P50P is one that I've discovered and seems to work fine but the thermals aren't as greatly reduced as I'd expect. It looks like the back-EMF generated stops at about -800V so it's still avalanching (you can tell because on the scope it still clips rather than peaks), just not at the same level as the IRF9640

[Vadim18]

CrizzyD, sorry for mistake - not IRF9310 but IRFR9310
You have ambitios project will be nice to see how it will groove
and about IRF9630, in my Surfmaster it work better than 9640 wich have less Ron

[Vadim18]

Why 9630 can be better, let's compare
IRF9630.....................................IRF964 0

Ron - 0.8Ohm............................Ron - 0.5 Ohm
Cin - 550pf.................................Cin - 1200 pf
Cout - 170pf...............................Cout - 370 pf
Crev - 50pf.................................Crev - 81 pf

At least IRF9630 turns off two times faster than 9640, becouse of input capacitance
But yeah i noticed that advantage in my first Surfmaster build in 2017, and i remembered one reason why it work poorly
i find coil inductance was to big - more than 400uH becose was made from cheap Ethernet twist pair and had R about 10 Ohm
And than i decided to try another TX transistor i put scrapped from crt monitor 9630 and get a little sensitivity boost

In my case with poor coil both MOSFETS are not reached avalanche point and fast turn off makes flyback pulse bigger wich helps in case with 9630

[CrizzyD]

That's a very good point about the switch off timing that I hadn't mentioned in this thread. The IXTP10P50P has:
Css 2840pF
Iss 275pF
Cres 42pF

Which seems superior to the IRF9640 but actually still falls kind of short of the 9630. I'll still have the issue of thermals however as it's likely the repetitive avalanching that's the issue.

Conversely my transmit coil has an inductance of only 80uH but a receive coil of 2000uH.

[Carl-NC]

Options for dealing with this:

1. Decrease the peak current I_p at turn-off. The flyback peak is



so V_p is proportional to I_p. Lower I_p = less depth.

2. Increase the coil capacitance (see above eq). This will slow down the decay and increase the minimum pulse delay.

3. Try a cascoded PMOS:



I've never tried this, but with equal R it should split the flyback voltage equally. R will need to be low enough to keep the speed up, maybe 1k. It also creates a DC current and a DC offset at the preamp input so that will have to be dealt with.

4. Use NMOS

[Vadim18]

l may be wrong, but my idea to use IRF740 as aditional switch at the ground side


I guess this TX will switch-off faster but IRF740 output capacitance must be lower than 9630 better twice time lower
MOSFET at low side will get half off flyback voltage if it output capacitance be equal with 9630

[pito]

IRFZ44 is giving 2x times stronger signal transmitted by coil then IRF740. IRFZ44 it is not high voltage MOSFET but didn't die during testings.

[moodz]

...I dont think the search for the Perfect high voltage P switch is going to pay off despite any amount of CE ( Catalogue Engineering )

I would suggest this circuit ... a simple diode pump driven by a high amp mosfet driver ( single or parrallel ) ... or a custom pulse generator provides a pulsed supply to source of a NFET switch.
Everything works of pos supply.

Put a diode in series with the nfet drain for capacitive isolation on switch off if you want.

Choose a low ESR cap(s) for C1 and size proportional to the pulse length.

eezy peezy.

moodz.

[CrizzyD]

Not quite sure how this would work. VGS would be -12V so tying the gate to ground results in a 12V difference and so it conducts, but what would the purpose of the voltage multiplier be (I think that's what you're referring to when you say diode pump?).