I am fairly new to working with MOSFETs since I normally use BJT but I need to use a mosfet because I need low on resistance.
Using a P-Channel mosfet with the source connected to +15v, is bringing the gate down to 0v sufficient to turn the mosfet fully on?
Thanks in advance.
Yes but taking it up to 5V is not enough to turn it off.
Grumpy_Mike:
Yes but taking it up to 5V is not enough to turn it off.
I was thinking of using a pullup resistor to +15 and use an 2n2222 driven with a MCU pin (with base resistor of course) to pull it down. Would this work?
No it will fry your arduino's output pin. You need to pull it up to 12V with a resistor and pull that resistor down with a transistor.
Grumpy_Mike:
No it will fry your arduino's output pin. You need to pull it up to 12V with a resistor and pull that resistor down with a transistor.
You must have misunderstood me.
I attached a schematic.
Yes absolutely But 15 V enhancement on a mosfet gate is close to the limit, You could put a 5V Zener from gate to collector in series with the existing resistor, cathode to the gate and pull up for a cheap level shifter a 1/4 w 5V1 zener would work well. The Problem is the diaellectric used fir the gate... It will 'Punch Through' at about 18 to 20 Volts and -10V is more than enough enhancement to fully turn on the gate... IMO but as always check the data sheet for max ratings. The Gate is very sensitive to electrical damage... of any kind.
Doc
Docedison:
Yes absolutely But 15 V enhancement on a mosfet gate is close to the limit, You could put a 5V Zener from gate to collector in series with the existing resistor, cathode to the gate and pull up for a cheap level shifter a 1/4 w 5V1 zener would work well. The Problem is the diaellectric used fir the gate... It will 'Punch Through' at about 18 to 20 Volts and -10V is more than enough enhancement to fully turn on the gate... IMO but as always check the data sheet for max ratings. The Gate is very sensitive to electrical damage... of any kind.Doc
The mosfet I am probably going to use is here Intelligent Power and Sensing Technologies | onsemi
It shows the gate-source voltage as +25 so 15v should be ok. The 15v is going to be regulated with a voltage regulator.
I suppose I could use a zener dropper although it seems a bit redundant but as I said before I do not really understand mosfets.
I may have to put a 20v or so zener across the 15v line so it does not go to high from a spike of driving an inductive load.
Go look up a 1.5KE18, it's a Tranzorb, use it on the +15V source and if you are still concerned put a 18 V 1/4 W zener from gate to ground and you will be reasonably well protected. A silicon diode like a 1N5408 is a good idea as well for neg spike control. From +15V to ground, the anode, cathode to +15V... Otherwise you show a fair understanding of how to use a Mosfet.
Doc
smeezekitty:
The mosfet I am probably going to use is here Intelligent Power and Sensing Technologies | onsemi
It shows the gate-source voltage as +25 so 15v should be ok. The 15v is going to be regulated with a voltage regulator.
Yes, you can drive it as you were intending to, since 15v is well below the Vgs rating of 25V.
If you do want to drop the gate voltage a little, then an easier way than using a zener is to replace the single pullup resistor by a voltage divider, e.g. 470 ohms between mosfet gate and 2n2222 collector, and 1K between mosfet gate and +15v (giving 10v Vgs instead of 15v).
If the load is highly inductive, don't forget to include a flyback diode in parallel with it.
Mine was an old school engineering text book idea, yours is much better, I hadn't even considered it that way. I have always had a couple of drawers of zeners when working, material like that encourages laziness, now I make do with TIL431's. And occasionally very good advice from others.
Doc
Docedison:
Go look up a 1.5KE18, it's a Tranzorb, use it on the +15V source and if you are still concerned put a 18 V 1/4 W zener from gate to ground and you will be reasonably well protected. A silicon diode like a 1N5408 is a good idea as well for neg spike control. From +15V to ground, the anode, cathode to +15V... Otherwise you show a fair understanding of how to use a Mosfet.Doc
If you put the zener between gate and source (cathode to source for p-channel) you'll protect the gate from overvoltage and from going negative (not that it minds going negative). For source/drain overvoltage a TVS diode would be useful, but do we know if that's an issue even?
The load is mostly capacitive but somewhat inductive and resistive as well (nice mish-mash).
Docedison:
Go look up a 1.5KE18, it's a Tranzorb, use it on the +15V source and if you are still concerned put a 18 V 1/4 W zener from gate to ground and you will be reasonably well protected. A silicon diode like a 1N5408 is a good idea as well for neg spike control. From +15V to ground, the anode, cathode to +15V... Otherwise you show a fair understanding of how to use a Mosfet.Doc
I will probably go this route.
Yes, you can drive it as you were intending to, since 15v is well below the Vgs rating of 25V.
If you do want to drop the gate voltage a little, then an easier way than using a zener is to replace the single pullup resistor by a voltage divider, e.g. 470 ohms between mosfet gate and 2n2222 collector, and 1K between mosfet gate and +15v (giving 10v Vgs instead of 15v).
Wouldn't the gate still rise to +15v when the NPN transistor is turned off?
If the load is highly inductive, don't forget to include a flyback diode in parallel with it.
Of course. But this will not help on positive spikes.
If you put the zener between gate and source (cathode to source for p-channel) you'll protect the gate from overvoltage and from going negative (not that it minds going negative).
OK
For source/drain overvoltage a TVS diode would be useful, but do we know if that's an issue even?
I doubt that it will exceed 60v but it can be hard to say.
The load is mostly capacitive but somewhat inductive and resistive as well (nice mish-mash).
No you can't say that.
If you have a capacitive load and an inductive load the two cancel out according to the values of C and L and you end up with either a capacitive or an inductive load. In the case where they both are equal, that is a resonance point.
Grumpy_Mike:
The load is mostly capacitive but somewhat inductive and resistive as well (nice mish-mash).
No you can't say that.
If you have a capacitive load and an inductive load the two cancel out according to the values of C and L and you end up with either a capacitive or an inductive load. In the case where they both are equal, that is a resonance point.
I am still going to use plenty of protection since diodes are only a few cents.
Now off to solving the next hurdle: current sensing.
smeezekitty:
Wouldn't the gate still rise to +15v when the NPN transistor is turned off?
The rating we are talking about is the gate to source voltage and the source is tied to +15v. So having the gate at +15v is no problem because then the gate to source voltage is zero. On the other hand, when the 2n2222 turns on, its collector is close to 0v so Vgs is around -15v. The point of the voltage divider is to increase the gate voltage to e.g. 5v so that Vgs is limited to (5 - 15) volts = -10v instead of -15v.
Resonance is defined as the frequency where XL and XC are equal, at that point they cancel out and the resultant is just the resistance of the circuit. The formula are somewhat difficult for one not familiar with the concept. In your description of combined reactances, inductive and capacitive the values for XL and XC are opposite in sign relative to each other and add algebraically. If for example if the XC was -180 ohms and XL was 90 ohms them the resultant would be -90 ohms. -reactances are capacitive and +reactances are inductive, so if XC =1/(2 X 6.28 X F X C) then 1/C = XC/6.28/F or if positive (inductive) L = XL/6.28/F (C=Farads and L = Henry's). Frequently small value reactances are used to negate other reactance's... one example is power factor correction where typically the load might be inductive in nature (lots of big motors) and provide a mismatched load. Remember that max power transfer occurs where Xload = Xsource (for ac circuits) a capacitor equal to XL (load) is placed in parallel with the load to cancel out the inductive part of the load. The technique used in driving capacitive mosfet gates is to just 'swamp' them out i.e. provide a drive impedance so low as to force them into a minor consideration. Another perhaps better analogy would be parallel resistances. Assume a 10K resistor in circuit and the value might need to be 2K2 ohms and since (1/Rt) = (1/R1) + (1/R2) {Formula for the parallel equivalent resistance of 2 resistors).
We can say that (1/Rpar) = (1/Rreq) - (1/Rcir)... `(1/2200) = .0004545... and 1/10000 = .0001... therefore .0004545... - .0001 - .0003545... = .0003545...Rreq = 1/.0003454... = 2K82 ohms and the proof is 1/Rt = 1/10000 + 1/2820 = 2K2 ohms. The only difference between DC and AC is the sign of the reactance. Long and complicated until you have done it a few times but trivial with a little experience. Thus ends Basic AC theory lesson 1.6.2... or some such number. I didn't write the book and I remember a few pages, fuzzily. In Closing capacitive reactances I.E. Mosfet Gates are usually driven with a generator that is at least 1/10 the impedance of the gate being driven. The math and descriptions would occupy several pages and would need to come from 3 different books, Much too long, I think that if you were able to follow my reasoning so far you have some third year electronics theory education. I also think that If you are reading this sentence you are either glassy eyed or seriously interested in the topic of impedance matching. This subject is very complicated as the available gate drive current is a function of the Rise Time of The Driving Pulse not it's Amplitude since we are driving a capacitor. Remember that Rise time is piece-wise equivalent to frequency and that XC (Gate impedance) is an inverse function of frequency.
Doc
Basically the circuit will be driving a capacitor in parallel with an unknown load which could be capacitive, inductive or resistive or a mixture. I was simply not clear enough.
Docedison:
Resonance is defined as the frequency where XL and XC are equal, at that point they cancel out and the resultant is just the resistance of the circuit. The formula are somewhat difficult for one not familiar with the concept. In your description of combined reactances, inductive and capacitive the values for XL and XC are opposite in sign relative to each other and add algebraically. If for example if the XC was -180 ohms and XL was 90 ohms them the resultant would be -90 ohms. -reactances are capacitive and +reactances are inductive, so if XC =1/(2 X 6.28 X F X C) then 1/C = XC/6.28/F or if positive (inductive) L = XL/6.28/F (C=Farads and L = Henry's). Frequently small value reactances are used to negate other reactance's... one example is power factor correction where typically the load might be inductive in nature (lots of big motors) and provide a mismatched load. Remember that max power transfer occurs where Xload = Xsource (for ac circuits) a capacitor equal to XL (load) is placed in parallel with the load to cancel out the inductive part of the load.
Got that almost 100% The problem is that the circuit is kinda-sorta a type of "power supply" and you do not know if the output will be a 100 Henry inductor or 1 Megaohm resistor or a capacitor. If the inductor is big enough, it may completely negate the effect of the capacitor.
The technique used in driving capacitive mosfet gates is to just 'swamp' them out i.e. provide a drive impedance so low as to force them into a minor consideration.
Wouldn't the 2N2222 driven to saturation provide sufficiently low impedance to have a decent rise time? I only need to run it at around 10KHz maximum.
Another perhaps better analogy would be parallel resistances. Assume a 10K resistor in circuit and the value might need to be 2K2 ohms and since (1/Rt) = (1/R1) + (1/R2) {Formula for the parallel equivalent resistance of 2 resistors).
We can say that (1/Rpar) = (1/Rreq) - (1/Rcir)... `(1/2200) = .0004545... and 1/10000 = .0001... therefore .0004545... - .0001 - .0003545... = .0003545...Rreq = 1/.0003454... = 2K82 ohms and the proof is 1/Rt = 1/10000 + 1/2820 = 2K2 ohms.
Now I am confused. What does "2K2" and "2K82" resistors mean?
The only difference between DC and AC is the sign of the reactance.
So basically inductors and capacitors are opposite from each other and what a capacitor does on AC, an inductor will do on DC and vice-versa.
Long and complicated until you have done it a few times but trivial with a little experience.
Not all that complicated but slightly confusing.
In Closing capacitive reactances I.E. Mosfet Gates are usually driven with a generator that is at least 1/10 the impedance of the gate being driven.
This should not be a problem.
The math and descriptions would occupy several pages and would need to come from 3 different books, Much too long
Too long indeed LOL
, I think that if you were able to follow my reasoning so far you have some third year electronics theory education.
I do not.
I also think that If you are reading this sentence you are either glassy eyed or seriously interested in the topic of impedance matching.
Actually the latter is more true.
This subject is very complicated as the available gate drive current is a function of the Rise Time of The Driving Pulse not it's Amplitude since we are driving a capacitor. Remember that Rise time is piece-wise equivalent to frequency and that XC (Gate impedance) is an inverse function of frequency.
In my experience, microcontroller pins usually have decently sharp rise time since I have put them on a scope before.
Now lets hope you have experience with OpAmps and current sensing since its another topic I am pretty new to and will probably start another thread soon.
Yes I can handle that... It's not too difficult. The Mosfet thing was something I learned in an advanced theory course and had to re-remember when I first started working with the devices in the late 80's. I DO Appreciate you being able to understand what I wrote it's the first time I had occasion to try to put into words something that is almost instinctive for me. The work is almost like a bicycle once you learn it it applies to a lot of different concepts. My hardest subject was Fourier analysis or at least the frequency/time relationships... which in a nutshell say that if a pulse has 0 rise time it has infinite current, a good part of what I was trying to explain involve that concept. The Math is easy... and remember that reciprocal idea, it is applicable to a great deal of electronics especially for deriving solutions to network matching or network solutions. Finally thank you for a thoughtful review of what I wrote.
Doc
Docedison:
My hardest subject was Fourier analysis or at least the frequency/time relationships...
That subject tends confuses me as well. which in a nutshell say that if a pulse has 0 rise time it has infinite current[/quote]
But in practice a pulse of zero rise time does not exist since everything has resistance.
The Math is easy... and remember that reciprocal idea, it is applicable to a great deal of electronics especially for deriving solutions to network matching or network solutions.
I do my best to remember it.
Finally thank you for a thoughtful review of what I wrote.
I am not 100% on all concepts yet but I am learning. The concept is not completely foreign to me.
