Post by thread:[EE] FET question

Hi Mauricio, Although switching the ground side of the Rabbit might not be how I would do it, it is probably OK. But only if you make sure that you are not driving any of the Rabbit pins with 0 volts when the Rabbit is unpowered. This can cause latchup and also other weird stuff. I have found that chips are more sensitive to latchup when driven negative than positive. But that is just my personal observation. An example of weird stuff that isn't latchup is an LM324 opamp when inputs are driven below the negative supply, it creates huge output swings. Cheerful regards, Bob On Thu, 24 Jul 2008 18:50:41 -0300, "Mauricio Jancic" <spam_OUTlistasTakeThisOuTjanso.com.ar> said: {Quote hidden}> Hi, > I have a circuit with a MOSFET driving a RCM4400W (A Rabbit module > with wifi, that need 80mA on startup and 450mA when the wifi module is > on), > and it is not working. > > The power supply is fine, I think, because if I connect the Rabbit > directly to the PSU it works. > > The mosfet seems to be working, because I have 3.3V on the Rabbit > terminals when it is on. > > The FET is driven with a PIC pin with a 5K6 resistor to gate. > > The source of the FET is to GROUND and the drain is connected to the > NEGATIVE terminal of the Rabbit. The POSITIVE terminal of the rabbit is > directly connected to 3.3V. > > There is nothing else in the circuit, shouldn't it be working? > > The FET is an FDV305N, 0.9A and the Vgsth is 1.5V. -- http://www.fastmail.fm - Accessible with your email software or over the web

Hi Mauricio, A few questions: 1) Besides 3.3V and the mosfet drain, what else is connected to the rabbit module? (I suspect there may be some kind of ground leakage path) 2) What voltage is your PIC running at? (to determine how it is driving the FET gate) I assume that the PIC shares the same ground as the FET? 3) Have you tried measuring the voltage across the rabbit module's power input (pos to neg) when the FET is turned on? (trying to determine whether there is a static on-resistance problem with the FET) 4) Does the rabbit module have large capacitors on its input or a switching power supply which may have a high inrush current? (This may cause the voltage to drop sharply at startup, which may in turn trigger the module to shut itself down immediately) Also, I'd suggest replacing the rabbit module with a 7 ohm resistor (approx) to simulate a 450mA load, then turn on your FET and measure how much voltage there is across the FET and across the resistor. You might also try removing the 5.6K resistor and just driving the FET gate directly from the PIC. Sean On Thu, Jul 24, 2008 at 5:50 PM, Mauricio Jancic <.....listasKILLspam@spam@janso.com.ar> wrote: {Quote hidden}> Hi, > I have a circuit with a MOSFET driving a RCM4400W (A Rabbit module > with wifi, that need 80mA on startup and 450mA when the wifi module is on), > and it is not working. > > The power supply is fine, I think, because if I connect the Rabbit > directly to the PSU it works. > > The mosfet seems to be working, because I have 3.3V on the Rabbit > terminals when it is on. > > The FET is driven with a PIC pin with a 5K6 resistor to gate. > > The source of the FET is to GROUND and the drain is connected to the > NEGATIVE terminal of the Rabbit. The POSITIVE terminal of the rabbit is > directly connected to 3.3V. > > There is nothing else in the circuit, shouldn't it be working? > > The FET is an FDV305N, 0.9A and the Vgsth is 1.5V. > > Regards, > > Mauricio Jancic > Microchip Design Partner > Janso Desarrollos > > -

> 1) Besides 3.3V and the mosfet drain, what else is connected to the > rabbit module? (I suspect there may be some kind of ground leakage > path) There is no other connection in or out the rabbit. I removed everything to make tests. The only connections to the Rabbit are the 3.3V line straight from the MCP1826 voltage regulator and the ground line going from the rabbit to the GND path through the FDV305N FET. > 2) What voltage is your PIC running at? (to determine how it is > driving the FET gate) I assume that the PIC shares the same ground as > the FET? Yes, it is the same ground. The everything is running at 3.3V (PIC16F67J10) > 3) Have you tried measuring the voltage across the rabbit module's > power input (pos to neg) when the FET is turned on? (trying to > determine whether there is a static on-resistance problem with the > FET) There are 3.3V as expected. The ON time of the FET is 200 uS. > 4) Does the rabbit module have large capacitors on its input or a > switching power supply which may have a high inrush current? (This may > cause the voltage to drop sharply at startup, which may in turn > trigger the module to shut itself down immediately) It doesn't seems to. However, it takes 450mA when the wifi module is on. I know the rabbit works while the wifi is off, but it resets when the power consumption increases (I should have mentioned this earlier, I know...) > Also, I'd suggest replacing the rabbit module with a 7 ohm resistor > (approx) to simulate a 450mA load, then turn on your FET and measure > how much voltage there is across the FET and across the resistor. I've done it with a 5.6 ohm. The weird thing is that it works just fine (550+ mA). With the actual load (the rabbit) the PIC resets sometimes when it is turned on, the 5.6ohm resistor has no negative effect however. > You might also try removing the 5.6K resistor and just driving the FET > gate directly from the PIC. Donne that. This causes the PIC to reset as soon as the GATE line is taken high. I put the resistor to solve this. I also placed a 1000 uF capacitor in the output of the MCP1826 voltage regulator, to improve this reset problem, that I trace down to the regulator not being able to regulate on the sudden change of load. What is puzzling is that with the 5.6 ohm resistor everything works as expected, there are 3.3V as needed, but the rabbit won't work. Regards, Mauricio Jancic Microchip Design Partner Janso Desarrollos

part 1 1781 bytes content-type:text/plain; charset="gb2312" (decoded quoted-printable) something I can think of: Is your power supply current limited?the gate resistor is way too high for me, but it's not a problem since I assume the load is always on but not switching, right? another thing is how much current your PIC can give at the output, maybe it's working in linea region or so. Why not connect your rabbit POSITIVE to the FET source and NEGATIVE to ground? I don't know how it should be working, though. > From: .....listasKILLspam.....janso.com.ar> To: EraseMEpiclistspam_OUTTakeThisOuTmit.edu> Subject: [EE] FET question> Date: Thu, 24 Jul 2008 18:50:41 -0300> > Hi,> I have a circuit with a MOSFET driving a RCM4400W (A Rabbit module> with wifi, that need 80mA on startup and 450mA when the wifi module is on),> and it is not working.> > The power supply is fine, I think, because if I connect the Rabbit> directly to the PSU it works.> > The mosfet seems to be working, because I have 3.3V on the Rabbit> terminals when it is on.> > The FET is driven with a PIC pin with a 5K6 resistor to gate.> > The source of the FET is to GROUND and the drain is connected to the> NEGATIVE terminal of the Rabbit. The POSITIVE terminal of the rabbit is> directly connected to 3.3V.> > There is nothing else in the circuit, shouldn't it be working?> > The FET is an FDV305N, 0.9A and the Vgsth is 1.5V.> > Regards,> > Mauricio Jancic> Microchip Design Partner> Janso Desarrollos> > -- > http://www.piclist.com PIC/SX FAQ & list archive> View/chan ge your membership options at> mailman.mit.edu/mailman/listinfo/piclist _________________________________________________________________ MSNʷʫ��Ƭ��ճ��ȿ�� http://im.msn.cn/ part 2 35 bytes content-type:text/plain; charset="us-ascii" (decoded 7bit)

At 06:38 PM 7/24/2008, you wrote: {Quote hidden}> > 1) Besides 3.3V and the mosfet drain, what else is connected to the > > rabbit module? (I suspect there may be some kind of ground leakage > > path) > >There is no other connection in or out the rabbit. I removed everything to >make tests. The only connections to the Rabbit are the 3.3V line straight >from the MCP1826 voltage regulator and the ground line going from the rabbit >to the GND path through the FDV305N FET. > > > 2) What voltage is your PIC running at? (to determine how it is > > driving the FET gate) I assume that the PIC shares the same ground as > > the FET? > >Yes, it is the same ground. The everything is running at 3.3V (PIC16F67J10) > > > 3) Have you tried measuring the voltage across the rabbit module's > > power input (pos to neg) when the FET is turned on? (trying to > > determine whether there is a static on-resistance problem with the > > FET) > >There are 3.3V as expected. The ON time of the FET is 200 uS. > > > 4) Does the rabbit module have large capacitors on its input or a > > switching power supply which may have a high inrush current? (This may > > cause the voltage to drop sharply at startup, which may in turn > > trigger the module to shut itself down immediately) > >It doesn't seems to. However, it takes 450mA when the wifi module is on. I >know the rabbit works while the wifi is off, but it resets when the power >consumption increases (I should have mentioned this earlier, I know...) > > > Also, I'd suggest replacing the rabbit module with a 7 ohm resistor > > (approx) to simulate a 450mA load, then turn on your FET and measure > > how much voltage there is across the FET and across the resistor. > >I've done it with a 5.6 ohm. The weird thing is that it works just fine >(550+ mA). With the actual load (the rabbit) the PIC resets sometimes when >it is turned on, the 5.6ohm resistor has no negative effect however. That's not weird, it's expected. The module probably has a bunch of bypass capacitors and you're momentarily killing the supply charging them. I suggest a separate regulator with shutdown input is probably the best way to handle this (effectively a high-side switch with current limiting) The problem with switching the low side is that any logic pins that connect to the PIC or whatever could get pulled way below "ground" from the perspective of the the Rabbit module (eg. the PIC is "low", a great deal of current will flow through the protection diodes on the Rabbit module, possibly damaging it or causing latchup which would result very quickly in damage due to high currents. {Quote hidden}> > > You might also try removing the 5.6K resistor and just driving the FET > > gate directly from the PIC. > >Donne that. This causes the PIC to reset as soon as the GATE line is taken >high. I put the resistor to solve this. I also placed a 1000 uF capacitor in >the output of the MCP1826 voltage regulator, to improve this reset problem, >that I trace down to the regulator not being able to regulate on the sudden >change of load. > >What is puzzling is that with the 5.6 ohm resistor everything works as >expected, there are 3.3V as needed, but the rabbit won't work. > >Regards, > >Mauricio Jancic >Microchip Design Partner >Janso Desarrollos > > >

Unfortunally, we already have 100 boards made. We can make small changes, like putting a FET on the high side of the rabbit, but making big changes will be difficult. I will see tomorrow, while I'm at the office, if it is possible to change the layout easily from low side switch to High side switch. by the way, if I connect the rabbit directly to ground (or with a switch or a relay) it works just fine. So, the problem is with the MOSFET... Regards On Thu, Jul 24, 2008 at 10:43 PM, Spehro Pefhany <@spam@speffKILLspaminterlog.com> wrote: {Quote hidden}> At 06:38 PM 7/24/2008, you wrote: > > > 1) Besides 3.3V and the mosfet drain, what else is connected to the > > > rabbit module? (I suspect there may be some kind of ground leakage > > > path) > > > >There is no other connection in or out the rabbit. I removed everything to > >make tests. The only connections to the Rabbit are the 3.3V line straight > >from the MCP1826 voltage regulator and the ground line going from the > rabbit > >to the GND path through the FDV305N FET. > > > > > 2) What voltage is your PIC running at? (to determine how it is > > > driving the FET gate) I assume that the PIC shares the same ground as > > > the FET? > > > >Yes, it is the same ground. The everything is running at 3.3V > (PIC16F67J10) > > > > > 3) Have you tried measuring the voltage across the rabbit module's > > > power input (pos to neg) when the FET is turned on? (trying to > > > determine whether there is a static on-resistance problem with the > > > FET) > > > >There are 3.3V as expected. The ON time of the FET is 200 uS. > > > > > 4) Does the rabbit module have large capacitors on its input or a > > > switching power supply which may have a high inrush current? (This may > > > cause the voltage to drop sharply at startup, which may in turn > > > trigger the module to shut itself down immediately) > > > >It doesn't seems to. However, it takes 450mA when the wifi module is on. I > >know the rabbit works while the wifi is off, but it resets when the power > >consumption increases (I should have mentioned this earlier, I know...) > > > > > Also, I'd suggest replacing the rabbit module with a 7 ohm resistor > > > (approx) to simulate a 450mA load, then turn on your FET and measure > > > how much voltage there is across the FET and across the resistor. > > > >I've done it with a 5.6 ohm. The weird thing is that it works just fine > >(550+ mA). With the actual load (the rabbit) the PIC resets sometimes when > >it is turned on, the 5.6ohm resistor has no negative effect however. > > That's not weird, it's expected. The module probably has a bunch of > bypass capacitors and you're momentarily killing the supply charging them. > > I suggest a separate regulator with shutdown input is probably the best > way to handle this (effectively a high-side switch with current limiting) > > The problem with switching the low side is that any logic pins that > connect to the PIC or whatever could get pulled way below > "ground" from the perspective of the the Rabbit module (eg. the PIC > is "low", a great deal of current will flow through the protection > diodes on the Rabbit module, possibly damaging it or causing latchup > which would result very quickly in damage due to high currents. > > > > > > > > > You might also try removing the 5.6K resistor and just driving the FET > > > gate directly from the PIC. > > > >Donne that. This causes the PIC to reset as soon as the GATE line is taken > >high. I put the resistor to solve this. I also placed a 1000 uF capacitor > in > >the output of the MCP1826 voltage regulator, to improve this reset > problem, > >that I trace down to the regulator not being able to regulate on the > sudden > >change of load. > > > >What is puzzling is that with the 5.6 ohm resistor everything works as > >expected, there are 3.3V as needed, but the rabbit won't work. > > > >Regards, > > > >Mauricio Jancic > >Microchip Design Partner > >Janso Desarrollos > > > > > >--

> I have a circuit with a MOSFET driving a RCM4400W (A > Rabbit module > with wifi, that need 80mA on startup and 450mA when the > wifi module is on), > and it is not working. FET should tolerate a zero ohm connection from PIC. Try this. - Set up system as described. - Add switch across FET drain-source. - Open switch - Enable system via FET. - Close switch. IF it works under that condition the FET is suspect as an on switch. FET should have about 0.15 ohm on resistance as described. Should be OKish although the module spec is 3V3 +/- 5% so you are outside that range at 450 mA. If it fails to work then the system is probably being disabled in some manner by the load side open switch. eg having the module connected to Vdd and stray paths from I/O to ground may do funny things to it. Also: RF may be getting into FET. All bets off if that happens :-). Russell

I have tried connecting and disconnecting the ground wire of the rabbit, directly, to GND and it successfully starts every time. The 450 mA at 0.15 ohm represent a voltage drop of 0.0225 V, that is 0.68% of 3.3V, so the Rabbit should work with that voltage, shouldn't? I'm trying not to consider RF yet... On Thu, Jul 24, 2008 at 11:00 PM, Apptech <KILLspamapptechKILLspamparadise.net.nz> wrote: {Quote hidden}> > I have a circuit with a MOSFET driving a RCM4400W (A > > Rabbit module > > with wifi, that need 80mA on startup and 450mA when the > > wifi module is on), > > and it is not working. > > FET should tolerate a zero ohm connection from PIC. > > Try this. > - Set up system as described. > - Add switch across FET drain-source. > - Open switch > - Enable system via FET. > - Close switch. > > IF it works under that condition the FET is suspect as an on > switch. > FET should have about 0.15 ohm on resistance as described. > Should be OKish although the module spec is 3V3 +/- 5% so > you are outside that range at 450 mA. > > If it fails to work then the system is probably being > disabled in some manner by the load side open switch. > eg having the module connected to Vdd and stray paths from > I/O to ground may do funny things to it. > > Also: RF may be getting into FET. All bets off if that > happens :-). > > > > Russell > > -

Apptech wrote: {Quote hidden}>> I have a circuit with a MOSFET driving a RCM4400W (A >> Rabbit module >> with wifi, that need 80mA on startup and 450mA when the >> wifi module is on), >> and it is not working. >> > > FET should tolerate a zero ohm connection from PIC. > > Try this. > - Set up system as described. > - Add switch across FET drain-source. > - Open switch > - Enable system via FET. > - Close switch. > > IF it works under that condition the FET is suspect as an on > switch. > FET should have about 0.15 ohm on resistance as described. > Should be OKish although the module spec is 3V3 +/- 5% so > you are outside that range at 450 mA. > > If it fails to work then the system is probably being > disabled in some manner by the load side open switch. > eg having the module connected to Vdd and stray paths from > I/O to ground may do funny things to it. > > Also: RF may be getting into FET. All bets off if that > happens :-). > > > > Russell > > The Fet itself might be ok with a 0 ohm drive but the PIC probably wont be ;-> you could well be asking for a few amps of drive current (for a very short time). I'm wondering given the OP is using a 5.6K resistor if the miller charge is turning the fet off again as the rabbit comes up to power (as somebody else said its probably charging some caps and asking for a few amps of current) To limit the initial current you want about a 170 ohm resistor and that should hopefully be stiff enough to handle any of that other weirdness that goes on with fets. Your sure the fet is into its saturated region at 3.3V gate drive? that's quite low, not impossible just uncommon is all. Don't know how it'll work with the rabit, but try sticking a 1 ohm resistor in line with it, might help reduce the switch on thump.

That particular Rabbit core has several power supplies for the radio, and while the radio may be off the power supplies do start up on power up, IIRC. This means that the core has a significant inrush current due to all the bypass capacitores used for filtering - 0.9A may not be enough, the Ron of the FET may not be low enough, etc, etc. Measure the voltage across the FET with a high speed scope to see what the inrush current looks like in combination with the turn on time of the FET. Also, are you able to reset the Rabbit when it doesn't come up using the FET to power it? If it resets ok, then you may need to control the reset along with the power suplly. If it doesn't reset then you'll need to beef up the power supply and switching mechanism (the FET and FET driver). You may need to drive the FET very hard to get it to turn on quickly enough especially since you don't have much overhead. You might consider trying the same thing with another Rabbit 4000 core to determine if it's the added radio circuitry, or if there's a problem with the reset and cpu itself. It should just plug right in. Good luck! -Adam On 7/24/08, Mauricio Jancic <RemoveMElistasTakeThisOuTjanso.com.ar> wrote: {Quote hidden}> Hi, > I have a circuit with a MOSFET driving a RCM4400W (A Rabbit module > with wifi, that need 80mA on startup and 450mA when the wifi module is on), > and it is not working. > > The power supply is fine, I think, because if I connect the Rabbit > directly to the PSU it works. > > The mosfet seems to be working, because I have 3.3V on the Rabbit > terminals when it is on. > > The FET is driven with a PIC pin with a 5K6 resistor to gate. > > The source of the FET is to GROUND and the drain is connected to the > NEGATIVE terminal of the Rabbit. The POSITIVE terminal of the rabbit is > directly connected to 3.3V. > > There is nothing else in the circuit, shouldn't it be working? > > The FET is an FDV305N, 0.9A and the Vgsth is 1.5V. > > Regards, > > Mauricio Jancic > Microchip Design Partner > Janso Desarrollos > > -