Post by thread:PIC MOSFET control

Hi all, I have a puzzler which is related to PICs only because I use a PIC to control the circuit. If you want to save your mail reading bandwith for PIC only questions skip to the next message now..... I am using a PIC to switch a full-wave rectified signal to a load through a MOSFET. The actual circuit is a PIC 16C57 output driving a CMOS 4050 buffer which drives the gate of an IRF520 n-channel MOSFET. The switched signal is a full-wave rectified 24vac (rms) so that means roughly 12volts rms (upper half of the waveform). Here is the circuit: ^ _ _ _ | / \ / \ / \ 12vrms / \ / \ / \ ____|______/_____\/_____\/_____\______ 24vac (rms), full-wave rectified | | load (4 ohm, pure resistance) | | drain } PIC-------->MHC14050------->IRF520 gate } rds(on)=0.27ohm } source | | ground (common to PIC and rms waveform) I use the PIC to sense the zero of the rectified signal and turn on the MOSFET at low voltage (approx. 1 volt). The MOSFET turns off at the appropriate point to provide the desired power to the load. This all works like a champ. My problem is that the MOSFET is getting very hot. It is rated at 8 amps and 40 watts with rds on of 0.27 ohm (at 5 amps). Thus far I have only used about 30 percent duty cycle to the load. By my calculations that should be just over 2 watts in the MOSFET even at 100% duty cycle. But that little puppy is cooking. I hesitate to display my ignorance but am I just missing something basic about the use of MOSFETs? I did not expect to be near the limit for power usage at all. If any of you see an obvious solution please let me know. It is probably most appropriate to respond directly to me at spam_OUTmlkTakeThisOuTasu.edu. This is not just PIC related so let's not clog the list bandwidth with discussion of MOSFET physics. Thanks all for your patience, Martin Kirk Arizona State University .....mlkKILLspam@spam@asu.edu (602) 582-5718

At 16.48 1995-12-01 -0700, Martin Kirk wrote: -- chop -- chop -- > My problem is that the MOSFET is getting very hot. It is rated >at 8 amps and 40 watts with rds on of 0.27 ohm (at 5 amps). Thus far I have >only used about 30 percent duty cycle to the load. By my calculations that >should be just over 2 watts in the MOSFET even at 100% duty cycle. But >that little puppy is cooking. The thermal resistance of air is rather high so I suggest you put the MOSFET onto a heat sink with some "silicon grease" between the MOSFET and the heat sink. It was a while since I did some calculations with power dissipation but one guideline is: Tj - Tamb = P * (Kjm + Kmh + F * Kh) where Tj = temperature inside MOSFET. Tamb = ambient temperature (air). P = power dissipation from MOSFET. Kjm = thermal resistance between MOSFET (inside) and it's package, internal, may be found in the MOSFET:s data sheet. Kmh = thermal resistance between the package and the heat sink depends on what you put between your MOSFET and the heat sink, may also be found in the MOSFET:s data sheet. Kh = thermal resistance between the heat sink and the ambient air, may be found in the heat sink data sheet. F = factor representing air flow onto the heat sink. F = 1 if air velocity = 0 and decreases exponentially with increasing air velocity. -- Conny

Look at the gate of the MOS FET with a scope and see how fast you are turning the device on, it may have an ON resistance of 0.27 ohms but this is when it is fully ON, if the rise time of the gate drive voltage is slow then you will be disipating a lot of power during this time. It may only be for a few mS but you would not believe how much power can be disipated in that time. MOS FET's have a gate capacitance of about 1600pf and you need to be able to drive them with a current source capable of supplying 1.5A+for a few nS. Hope this helps. Ken. +-----------------------------+----------------------------------+ | ken hewitt | Email kenKILLspamwelwyn.demon.co.uk | +-----------------------------+----------------------------------+

> My problem is that the MOSFET is getting very hot. It is rated >at 8 amps and 40 watts with rds on of 0.27 ohm (at 5 amps). Thus far I have >only used about 30 percent duty cycle to the load. By my calculations that >should be just over 2 watts in the MOSFET even at 100% duty cycle. But >that little puppy is cooking. > > I hesitate to display my ignorance but am I just missing >something basic about the use of MOSFETs? I think so. I think this is like a problem I had using the same MOSFET to drive a transformer primary. You need to turn the gate on with a voltage higher than the circuit Vcc. Try creating a seperate power supply a volt or so higher to power the PIC and the 4050. See if that reduces the voltage drop across the MOSFET to a level that would result in the .3 ohm rds on range. I think this is whats called a "high-side" drive.

Thanks to all who have responded to my PIC drives MOSFET question. Some of the best info was regarding the voltage drive levels to the MOSFET gate. It appears that 5V may not be enough gate drive. It also appears that I may need enough current drive to insure that the gate turns ON and OFF (source and sink) very quickly so that I spend very little time in the high resistance region. That could also account for excessive MOSFET heating. So here is the prize question... the winner will have my eternal gratitude (I'll pull a thorn out of your foot someday): Does anyone know of a single-chip solution to interface the PIC output (5V) to the MOSFET gate (around 12V) with plenty of source and sink capability. I hear the MOSFET gate capacitance can be as high as 1600pf (I do not have a data sheet on the IRF520) so current drive should be: I = C*(deltaV/deltaT) = 1600pf * (12V/100nsec) = 192 mA A bipolar transistor voltage translation could work. I need 8 of these for 8 channels of MOSFET drive. So single-chip solution means all 8 channels. Thanks all, Martin Kirk Arizona State University .....mlkKILLspam.....asu.edu (602) 582-5718

{Quote hidden}>Thanks to all who have responded to my PIC drives MOSFET question. Some >of the best info was regarding the voltage drive levels to the MOSFET >gate. It appears that 5V may not be enough gate drive. It also appears >that I may need enough current drive to insure that the gate turns ON and >OFF (source and sink) very quickly so that I spend very little time in the >high resistance region. That could also account for excessive MOSFET heating. > > So here is the prize question... the winner will have my eternal >gratitude (I'll pull a thorn out of your foot someday): > > Does anyone know of a single-chip solution to interface the PIC >output (5V) to the MOSFET gate (around 12V) with plenty of source and >sink capability. I hear the MOSFET gate capacitance can be as high as >1600pf (I do not have a data sheet on the IRF520) so current drive should be: > > I = C*(deltaV/deltaT) = 1600pf * (12V/100nsec) = 192 mA > > A bipolar transistor voltage translation could work. I need 8 of >these for 8 channels of MOSFET drive. So single-chip solution means all >8 channels. > Maxim has a few high side mosfet driver chips (some with internal charge pumps). I think the MAX620 series might be what you are looking for. You can call Maxim at 800-998-8800, they have been extremely helpful to me in the past supplying literature and samples. On another note. I believe there was a fairly comprehensive article in Circuit Celler Ink in the last few months doing something very similar to what you want to do. I've lent out all my CCI for the last year to a friend, so I can't look up the article myself. Some other pic-list reader might have a better memory. Hope this helps. carl ---------------------------------------------------------------- Henry Carl Ott N2RVQ | talk/chat carlott@204.74.7.186 EraseMEcarlottspam_OUTTakeThisOuTinterport.net | http://www.interport.net/~carlott/ ---------------------------------------------------------------- "On justice and on friendship, there is no price, but there are established credit limits."

On Sat, 2 Dec 1995 mlkspam_OUTASU.EDU wrote: > > Does anyone know of a single-chip solution to interface the PIC > output (5V) to the MOSFET gate (around 12V) with plenty of source and > sink capability. I hear the MOSFET gate capacitance can be as high as > 1600pf (I do not have a data sheet on the IRF520) so current drive should be: > > I = C*(deltaV/deltaT) = 1600pf * (12V/100nsec) = 192 mA > > A bipolar transistor voltage translation could work. I need 8 of > these for 8 channels of MOSFET drive. So single-chip solution means all > 8 channels. > Electronics Now magazine has a column called Hardware Hacker by Don Lancaster. A while back, maybe a year ago, he had some info on "high side drivers". He is very good with cross references and bibliographies. He is on the Genie service in the PSRT (postscript round table) also. Good luck to you, I spent a lot of time pulling my hair out over a similar, though higher current circuit. morris beverly @spam@avpresKILLspamworld.std.com

NOTE: This is the second attempt to post because the list bounced the first one. ---snip ---snip {Quote hidden}> So here is the prize question... the winner will have my eternal >gratitude (I'll pull a thorn out of your foot someday): > > Does anyone know of a single-chip solution to interface the PIC >output (5V) to the MOSFET gate (around 12V) with plenty of source and >sink capability. I hear the MOSFET gate capacitance can be as high as >1600pf (I do not have a data sheet on the IRF520) so current drive should be: > > I = C*(deltaV/deltaT) = 1600pf * (12V/100nsec) = 192 mA > > A bipolar transistor voltage translation could work. I need 8 of >these for 8 channels of MOSFET drive. So single-chip solution means all >8 channels. > Maxim has a few high side mosfet driver chips (some with internal charge pumps). I think the MAX620 series might be what you are looking for. You can call Maxim at 800-998-8800, they have been extremely helpful to me in the past supplying literature and samples. On another note. I believe there was a fairly comprehensive article in Circuit Celler Ink in the last few months doing something very similar to what you're attempting. I've lent out all my CCI for the last year to a friend, so I can't look up the article myself. Some other pic-list reader might have a better memory. Hope this helps. carl ---------------------------------------------------------------- Henry Carl Ott N2RVQ | talk/chat carlott@204.74.7.186 KILLspamcarlottKILLspaminterport.net | http://www.interport.net/~carlott/ ---------------------------------------------------------------- "On justice and on friendship, there is no price, but there are established credit limits."

... > So here is the prize question... the winner will have my eternal >gratitude (I'll pull a thorn out of your foot someday): > > Does anyone know of a single-chip solution to interface the PIC >output (5V) to the MOSFET gate (around 12V) with plenty of source and ... Well, close, but I won't claim the price since you'll need two of them: Maxim has a few chips that do exactly that, e.g. the MAX62x series (there may be others, but I just had a look in the '94 databook). These contain 4 drivers and a charge pump to increase the driver voltage to 11 Volts above supply (IIRC, MOSFET supply). Hope this Helps. Bye Markus

cut...cut > Does anyone know of a single-chip solution to interface the PIC >output (5V) to the MOSFET gate (around 12V) with plenty of source and >sink capability. I hear the MOSFET gate capacitance can be as high as >1600pf (I do not have a data sheet on the IRF520) so current drive should be: > > I = C*(deltaV/deltaT) = 1600pf * (12V/100nsec) = 192 mA > > A bipolar transistor voltage translation could work. I need 8 of >these for 8 channels of MOSFET drive. So single-chip solution means all >8 channels. > cut...cut Try a photo FET optocoupler from QT Optoelectronics, H11F1 or H11F2. Digikey sells these for about $2.00. Rob

In a message dated 95-12-03 15:21:41 EST, you write: >> So here is the prize question... the winner will have my eternal >>gratitude (I'll pull a thorn out of your foot someday): >> >> Does anyone know of a single-chip solution to interface the PIC >>output (5V) to the MOSFET gate (around 12V) with plenty of source and >>sink capability. I hear the MOSFET gate capacitance can be as high as >>1600pf (I do not have a data sheet on the IRF520) so current drive should >be: >> >> I = C*(deltaV/deltaT) = 1600pf * (12V/100nsec) = 192 mA >> >> A bipolar transistor voltage translation could work. I need 8 of >>these for 8 channels of MOSFET drive. So single-chip solution means all >>8 channels. >> > International Rectifier makes a line of MOSFETs designed specifically to be driven at logic levels. They have a much lower gate threshold voltage than their "regular" counterparts. The equivalent device to the IRF520 would be the IRL520. The gate capacitance for the IRL520 is 490 pF. Hope this helps- Ken Pontbriant

> Electronics Now magazine has a column called Hardware Hacker by Don > Lancaster. A while back, maybe a year ago, he had some info on "high > side drivers". He is very good with cross references and > bibliographies. He is on the Genie service in the PSRT (postscript round > table) also. Good luck to you, I spent a lot of time pulling my hair out > over a similar, though higher current circuit. I don't have that magazine; I'm trying to figure a good way to switch as many outputs per chip as I can, at least 20 mils per and preferably (at least for some) 50 [driving IR led's]. Currently I'm using 5484's for low side and 5484's with external pullups for high side. Unfortunately, sinking all the unused pullups uses an objectionable amount of current and causes the power supply to sag. If I can't find a good chip, I'm thinking of using NPN transistors as emitter-followers for the high-side drive. These would have the advantage that I could hook the collector up to an unregulated supply and the output of the transistors would still be stable [so my regulator wouldn't have to supply too much current]. Any thoughts on this approach [good/bad/indiffer- ent?]

On Mon, 4 Dec 1995, John Payson wrote: > low side and 5484's with external pullups for high side. Unfortunately, > sinking all the unused pullups uses an objectionable amount of current and > causes the power supply to sag. > > If I can't find a good chip, I'm thinking of using NPN transistors as > emitter-followers for the high-side drive. These would have the advantage > that I could hook the collector up to an unregulated supply and the output > of the transistors would still be stable [so my regulator wouldn't have to > supply too much current]. Any thoughts on this approach [good/bad/indiffer- > ent?] I don't see any need for a tightly regulated drive voltage: you need to turn them on, but you aren't switching anywhere near fast enough to need great concern over the rise and fall times. For that matter, if I recall this application correctly, you're always turning the gate drive on at the zero crossing, aren't you? In that case the gate risetime is unimportant: there won't be any current flowing until the unfiltered supply voltage rises, so FET turn-on time should have negligible effect on power dissipation. However, turn-off will matter. So I would suggest that you should use the open-collector drivers with pullups to the higher voltage (to insure full turn-on) but you can use relatively large resistors - the time when you need to move a lot of charge on the gate is at turn off, when the open-collector drivers should turn on and do the job very nicely. Luck!

On Mon, 4 Dec 1995, John Payson wrote about driving MOSFETs: > low side and 5484's with external pullups for high side. Unfortunately, > sinking all the unused pullups uses an objectionable amount of current and > causes the power supply to sag. > > If I can't find a good chip, I'm thinking of using NPN transistors as > emitter-followers for the high-side drive. These would have the advantage > that I could hook the collector up to an unregulated supply and the output > of the transistors would still be stable [so my regulator wouldn't have to > supply too much current]. Any thoughts on this approach [good/bad/indiffer- > ent?] I don't see any need for a tightly regulated drive voltage: you need to turn them on, but you aren't switching anywhere near fast enough to need to be greatly concerned about the the rise and fall times. For that matter, if I recall this application correctly, you're always turning the gate drive on at the zero crossing, aren't you? In that case the gate drive risetime is unimportant: there won't be any current flowing until the unfiltered supply voltage rises, so FET turn-on time should have negligible effect on power dissipation. However, turn-off will matter. So I would suggest that you should use the open-collector drivers with pullups to the higher voltage (to insure full turn-on) but you can use relatively large resistors - the time when you need to move a lot of charge on the gate is at turn off, when the open-collector drivers should turn on and do the job very nicely. Luck!

> >Thanks to all who have responded to my PIC drives MOSFET question. Some > >of the best info was regarding the voltage drive levels to the MOSFET > >gate. It appears that 5V may not be enough gate drive. It also appears > >that I may need enough current drive to insure that the gate turns ON and > >OFF (source and sink) very quickly so that I spend very little time in the > >high resistance region. That could also account for excessive MOSFET heating. {Quote hidden}> > > > So here is the prize question... the winner will have my eternal > >gratitude (I'll pull a thorn out of your foot someday): > > > > Does anyone know of a single-chip solution to interface the PIC > >output (5V) to the MOSFET gate (around 12V) with plenty of source and > >sink capability. I hear the MOSFET gate capacitance can be as high as > >1600pf (I do not have a data sheet on the IRF520) so current drive should be: > > > > I = C*(deltaV/deltaT) = 1600pf * (12V/100nsec) = 192 mA > > > > A bipolar transistor voltage translation could work. I need 8 of > >these for 8 channels of MOSFET drive. So single-chip solution means all > >8 channels. > > > > Maxim has a few high side mosfet driver chips (some with internal charge > pumps). I think the MAX620 series might be what you are looking for. You > can call Maxim at 800-998-8800, they have been extremely helpful to me in > the past supplying literature and samples. > > On another note. I believe there was a fairly comprehensive article in > Circuit Celler Ink in the last few months doing something very similar to > what you want to do. I've lent out all my CCI for the last year to a friend, > so I can't look up the article myself. Some other pic-list reader might have > a better memory. > > > Hope this helps. > > carl > > ---------------------------------------------------------------- > Henry Carl Ott N2RVQ | talk/chat carlott@204.74.7.186 > RemoveMEcarlottTakeThisOuTinterport.net | http://www.interport.net/~carlott/ > ---------------------------------------------------------------- > "On justice and on friendship, there is no price, > but there are established credit limits." > Linear Technology make a LTC1155 DUAL High Side Driver chip. This allows using low cost N channel FET's for High Side switching applications. These chips can operate off a 5V supply and safely drive the gates of virtually all FET's. Ideally suited for low voltage applications,and compatible with Standard logic families. Available in 8-Pin SO Package. Hope this helps. -Sandy.