N-channel vs P-channel MOSFETs

[kt315]

with IRF840 500V voltage max rating plus diode STTH1210 1000V peak reverse voltage this gives us 1500V max of delta pulse value max. you may easy encrease the (TX power) PULSE WIDTH to 500 microseconds and maybe more. that was been released in Delta Pulse on the new board O_o.

[VinceC]

3 point. a point where to connect the shield of schematic and a coil. with P-channel the point is a ground of the schematic. with N-channel the point is +E of the schematic, the PLUS of batteries.

Thanks! This is a very good point about the shield that I neglected completely.

[Davor]

A diode in series with drain fixes both drain capacitance, and leakage current. Most spice models do not use all these effects, so you'll not see any leakage current, and you'll see only gate capacitances. Hopefully, with drain diode, say MUR460, you'll not experience any of these.

For example, a simple model for IRF740 is:
.model IRF740 VDMOS( Vto=3V Ron=0.55 Kp=5.8 Vds=400 Cgs=1.4n Is=1p Rb=20m Qg=63n mfg=IR)

While a much more accurate one is:
.MODEL IRF740 VDMOS(KP=3.1089 RS=0.0048 RD=0.4166 RG=0.91 VTO=4.5 LAMBDA=0.001 CGDMAX=1218p CGDMIN=15p CGS=1300p TT=533n IS=2.41E-09 N=1.401 RB=0.013053 m=0.452 Vj=0.36 Cjo=1424.39pF mfg=STmicro Qg=35n Ron=0.48 Vds=400)

Even the better one is too optimistic regarding leakage current.

The former one will not show any improvement using a diode in a circuit, while the latter will show you exactly how diode effectively cuts off MOSFET from the rest of circuitry after flyback.

[Davor]

with IRF840 500V voltage max rating plus diode STTH1210 1000V peak reverse voltage this gives us 1500V max of delta pulse value max.

No it doesn't. What it does is something a bit different, but IMHO much more useful.
There are two advantages:
1) lower capacitance
When a MOSFET goes to flyback, and voltage rises to a few hundred volts, so does the Coss charge. At high voltages it is relatively small ~200pF or so, but at small voltages it rises to over 2nF. Because this capacitance is in parallel with a coil, we don't want even that 200pF. A diode in series with drain gets reverse polarised after flyback, thus dimensioning it for voltages much higher than the MOSFET avalanche makes little sense. Diodes at high reverse voltages have significantly lower capacitance than Coss. MUR460 has less than 40pF, and STTH1210 less than 20pF at high voltages
2) elimination of leakage current
The above-mentioned Coss holds ~40nC of charge at 200V and after flyback it discharges either through a coil with no diode, or through a diode and MOSFET leakage currents. With reasonable pulse repetition rate, this discharge stops well above 50V on a drain before a next charging pulse, a diode remains shut, and thus a coil is well protected from leakage and the accompanied noise. Both STTH1210 and MUR460 are rated at below 10uA reverse current, while IRF740 leakage current is 25uA or more. Lower breakdown diodes tend to have even lower leakage, so again, going for kV diodes make little sense. My excuse for using MUR460 in Minipulse is that I have a few already.

Two flies in a single blow.

But not peak voltage increase.

[VinceC]

Davor,

Could you please show a simple schematic of what you suggested? Thanks!

[green]

Some real scope pictures of coil turn off. With no diode the circuit resonance drops from 1.64 Mhz to 395 Khz. Would need a lower value damping resistor, longer delay before first sample. Depending on the targets maybe not a problem.

[VinceC]

Some real scope pictures of coil turn off. With no diode the circuit resonance drops from 1.64 Mhz to 395 Khz. Would need a lower value damping resistor, longer delay before first sample. Depending on the targets maybe not a problem.

That is a very instructional picture. Thanks!

I now appears that the only real negative for n-channel MOSFETs is what kt315 mentioned about the ability to ground the shield. Am I correct?

[Chet]

Some real scope pictures of coil turn off. With no diode the circuit resonance drops from 1.64 Mhz to 395 Khz. Would need a lower value damping resistor, longer delay before first sample. Depending on the targets maybe not a problem.

Thank you for sharing this it is a great demonstration and very informative.

[Carl-NC]

Some prior answers were correct, so I'll combine them in a comprehensive answer.

The original Surf PI used a PNP switch which got converted to a PMOS switch. A P-side switch allows the coil to be "ground-referenced," whereas with an N-side switch the coil is "supply referenced." This has an impact on how the analog power supplies are designed, but it is almost entirely a psychological difference; some people struggle with the concept of the (+) battery terminal being called "ground" as it often is in an NMOS design. In the PMOS design, you would create the other analog rail with a voltage inverter, in the NMOS design you would instead use a voltage doubler. It turns out the two approaches are almost identical.

In an average performance PI detector like the SMPI, PMOS works as well as NMOS. NMOS is the better choice for higher performance designs.

[Davor]

Davor,

Could you please show a simple schematic of what you suggested? Thanks!

I referred to a latest delta pulse implementation, but green stripped the unnecessary parts and provided some perfect scope diagrams. You may also see that there is no extra voltage in a flyback, but obviously the coil becomes faster.

Fun part is that most amateur builds may benefit from this simple add-on, and it is less expensive than peanuts.

[VinceC]

Thank you kt315, Davor, green, and Carl-NC for all of the helpful information. I learned a lot in this thread!

[VinceC]

Fun part is that most amateur builds may benefit from this simple add-on, and it is less expensive than peanuts.

I assume that you mean the fast recovery diode.

[kt315]

Some real scope pictures of coil turn off. With no diode the circuit resonance drops from 1.64 Mhz to 395 Khz. Would need a lower value damping resistor, longer delay before first sample. Depending on the targets maybe not a problem.

you forgot inner capacitance of a o-scope probe influence. connect it AT THE COIL DIRECTLY and you will get something another.

[Davor]

I assume that you mean the fast recovery diode.

Yes. It also has to be capable of handling the peak current through a coil.
At the end, you'll not find many candidates that fit well with a PI machine, and are also available at your local shop. Point is that even with a modest fast recovery diode you'll be able to reduce sample delay.

A downside of fitting such a diode on a PCB is that the leads are awfully thick. I had to grind mine to fit it into a Minipulse.

[green]

you forgot inner capacitance of a o-scope probe influence. connect it AT THE COIL DIRECTLY and you will get something another.

The scope probes are connected to the coil and IRF740 drain for the decay pictures, but not the resonance pictures. The decay curve with the diode shorted looks like it takes an extra 250 nsec to decay. With the change in R damping that's required to prevent oscillation the delay period before the first sample might have to increase by 3 times. With 100 volts/division it's hard to see when the coil has damped close to zero volts. It's more than the 1.5 usec from the trace.