Hi all
Any suggestions? I plan to run off a Lipo, and only need to switch up to 1A (a lipo charger and GSM modem), so heat should not be a big issue.
I seem to be struggling?
Also do I need a resistor between Gate and Source to stop the Gate from floating?
I have some of these, that I now have working (powering a 30ma GPS); https://www.sparkfun.com/products/10213
I also have some 2N7000.
Assume I know nothing.
Why this here and not in an ardunio forum... The whole point is to save power!
Thanks
Greg,
UPDATE:
I just noticed that you said that you want to switch power to a LiPo charger and GSM radio, but you're asking for a 'N' channel MOSFET. Why do you think you need an 'N' channel? Aren't you switching the positive rail of the power supply? If so, you'll need a P channel FET and, if the voltage you're switching is greater than the power supply to your processor, you'll need to have two transistors configured in what is commonly known as a 'load switch'.
Previous Post:
I suggest that you get some SOT-23 breakout boards (both Sparkfun and Low Power Labs sell these) and use these to convert very commonly available SOT-23 MOSFETs to Through Hole, Dual Inline, footprint. These may look small, but they're only 3 pins and fairly easy to solder once you've tuned your technique.
The problem isn't current as much as it is the gate/source threshold voltage (Vgsth in most specs). Running from a 3.3V processor (where the GPIO output is typically lower than 3V) you'll have a hard time finding a through hole FET that will reliably switch Amps at this voltage. There are a bunch of SOT-23 transistors that will. Si2302DDS, for example, will switch 4 Amps at about 1.3V gate/source voltage.
Tom
Thanks Tom.
At this stage I just want to switch the GPS and GSM modules. Later I want to switch a 7-36v->5v regulator to the charger/lipo.
You've given me a diagram before where a N channel Mosfet drives a P-Channel.
For the life of me I do not understand the difference between the two. It seems to be about where you put a switch, but if you put it either side of the load what difference does it make, as neither option will result in a circuit unless they close.
Thanks for the tip on the Si2302DDS.
Quote from: gregcope on May 09, 2016, 05:13:37 PM
Thanks Tom.
At this stage I just want to switch the GPS and GSM modules. Later I want to switch a 7-36v->5v regulator to the charger/lipo.
You've given me a diagram before where a N channel Mosfet drives a P-Channel.
For the life of me I do not understand the difference between the two. It seems to be about where you put a switch, but if you put it either side of the load what difference does it make, as neither option will result in a circuit unless they close.
Thanks for the tip on the Si2302DDS.
The difference is not in powering the device you're switching, but with the REST of the signal pins going between the device and your processor. If you switch off the ground side of the device then all the signal pins will be pulled to the positive voltage giving your processor a headache and perhaps a very short life. On the other hand, if you switch the positive supply to the device, then all the signal pins will be pulled to ground which your processor is much more capable of managing.
The key is where signals are referenced, if they are referenced to ground it is better to switch off the positive supply.
Tom
Mr. Perky is using a nice load switch: FDC6420C which also fits on an SOT-23 breakout board, although there are 6 pins, rather than 3. Mr. Perky might even be a 'neighbour' of yours and might be able to advise you as to where you can get these locally.
Tom
Ah. Sort of understand.
For the GPS, which has a TX/RX set of pins, do I still need to use a P channel? I was going to switch it on GRD with an N Channel.
GSM is more complex as it has TX/RX, Enable, and VIO. I was going to switch it on GRD with an N Channel.
The GSM has a lipo charger and Lipo battery but I have not thought through that bit yet.
Quote from: gregcope on May 09, 2016, 06:04:01 PM
Ah. Sort of understand.
For the GPS, which has a TX/RX set of pins, do I still need to use a P channel? I was going to switch it on GRD with an N Channel.
GSM is more complex as it has TX/RX, Enable, and VIO. I was going to switch it on GRD with an N Channel.
The GSM has a lipo charger and Lipo battery but I have not thought through that bit yet.
If any of the devices are powered from the SAME voltage as your processor, then which type of transistor you use is somewhat arbitrary. However, for any devices that are powered from a higher voltage than your processor you should switch off the positive rail and this is easiest with a P Channel device. The Si2343CDS is a decent P channel FET with 30V breakdown and 20V gate/source breakdown and the combination of the Si2343 and Si2302 makes a decent load switch if you want to reduce inventory.
Tom
My go-to FETs are:
P-Channel: BS250P (TO-92), or NDS325AP (SOT-23).
N-Channel: 2N7000, BS170 (both TO-92), or BSS138, 2SK1824 (SOT-23).
Quote from: captcha on May 10, 2016, 01:42:24 AM
My go-to FETs are:
P-Channel: BS250P (TO-92), or NDS325AP (SOT-23).
N-Channel: 2N7000, BS170 (both TO-92), or BSS138, 2SK1824 (SOT-23).
Thanks, just what I wanted.
Quote from: TomWS on May 09, 2016, 06:28:19 PM
If any of the devices are powered from the SAME voltage as your processor, then which type of transistor you use is somewhat arbitrary. However, for any devices that are powered from a higher voltage than your processor you should switch off the positive rail and this is easiest with a P Channel device. The Si2343CDS is a decent P channel FET with 30V breakdown and 20V gate/source breakdown and the combination of the Si2343 and Si2302 makes a decent load switch if you want to reduce inventory.
Tom
Ah. Ok. So;
1. GPS runs off moteino 3.3v rail and has TX/RX pins to moteino, is ok for a GRD N Channel mosfet
2. GSM modem, that runs of a 3.7V lipo, with a VIO to set the logic Level converter (to 3.3v). I was thinking of an N Channel on the GRD. But it has an enable pin that can drop it to a sleep mode @50uA. Seems like it might not be worth the hassle?
Chip data sheet; https://cdn-shop.adafruit.com/datasheets/sim800h_hardware_design_v1.00.pdf (https://cdn-shop.adafruit.com/datasheets/sim800h_hardware_design_v1.00.pdf) - Page 60 for power consumption
Fona overviewhttps://learn.adafruit.com/adafruit-fona-mini-gsm-gprs-cellular-phone-module/overview
(https://learn.adafruit.com/adafruit-fona-mini-gsm-gprs-cellular-phone-module/overview)
3. GSM has a 5V LIPO charger, which I plan to run from a 7-36V regulator, with a P channel mosfet. However if the LIPO charger only does this, does it need a P, could I use an N? as it has no connections back the processor?!?!
Would make life easy to just use one type of mosfet.
Quote from: gregcope on May 10, 2016, 03:37:49 PM
Quote from: TomWS on May 09, 2016, 06:28:19 PM
If any of the devices are powered from the SAME voltage as your processor, then which type of transistor you use is somewhat arbitrary. However, for any devices that are powered from a higher voltage than your processor you should switch off the positive rail and this is easiest with a P Channel device. The Si2343CDS is a decent P channel FET with 30V breakdown and 20V gate/source breakdown and the combination of the Si2343 and Si2302 makes a decent load switch if you want to reduce inventory.
Tom
Ah. Ok. So;
1. GPS runs off moteino 3.3v rail and has TX/RX pins to moteino, is ok for a GRD N Channel mosfet
2. GSM modem, that runs of a 3.7V lipo, with a VIO to set the logic Level converter (to 3.3v). I was thinking of an N Channel on the GRD. But it has an enable pin that can drop it to a sleep mode @50uA. Seems like it might not be worth the hassle?
Chip data sheet; https://cdn-shop.adafruit.com/datasheets/sim800h_hardware_design_v1.00.pdf (https://cdn-shop.adafruit.com/datasheets/sim800h_hardware_design_v1.00.pdf) - Page 60 for power consumption
Fona overviewhttps://learn.adafruit.com/adafruit-fona-mini-gsm-gprs-cellular-phone-module/overview
(https://learn.adafruit.com/adafruit-fona-mini-gsm-gprs-cellular-phone-module/overview)
3. GSM has a 5V LIPO charger, which I plan to run from a 7-36V regulator, with a P channel mosfet. However if the LIPO charger only does this, does it need a P, could I use an N? as it has no connections back the processor?!?!
Would make life easy to just use one type of mosfet.
Well, you could make it easy or you could make it right.
Here is my recommendation:
ALL of your connections can use an N channel device: GPS for on/off switching, GSM & LiPo charger as level shifter. However, since you want to be able to run up to 36Volts on your LiPo charger you will need a device with a DrainSource breakdown voltage >36V. You'll have to search for that. This kind of voltage moves you into the 'trickier to do' category, stay below 20V and you stay in the 'easy peasy' category - the Si2302 can't handle 36V. Vishay Siliconix website has a good parametric search engine if you know what you're looking for. You need three N Channel devices with a Vgsth of <3V at whatever current your GPS device draws.
TWO of your connections require a P channel device: GSM & LiPo charger. You need two P channel devices with a DrainSource breakdown voltage of >36Volts and will need a current protection resistor in the gate leg of the 36V device as its GateSource break down voltage is probably 20 volts.
Tom
Would this irf9z14pbf be able to switch 36v?
Data sheet suggests drain source breakdown voltage of -60?
http://www.irf.com/product-info/datasheets/data/irf9z14pbf.pdf
I plan to drive the gate from an N channel mosfet.
I want to power side switch up to 32v.
Hi Greg,
The IRF9Z14 is a pretty easy to use FET.
QuoteWould this irf9z14pbf be able to switch 36v?
Sure, a 60v device should easily handle the 36v from your power source.
Just be certain that the source never goes over 60v, even for a moment.
This FET is also rated at 4.7a (@100°C).
Be sure that you are staying well under the disipation ratings.
For example, with no heatsink, it can only handle ½ to 1W.
Also, the gate is quite fragile. Never exceed 20v source to gate.
Plan on including a zener or (my preference) a couple of LEDs to limit
the gate voltage. There is little advantage to driving the gate with
more than 10 to 15v.
However, you will need to exceed 6 to 8v to adequately turn
on the FET. Refer to the datasheet graphs for the details.
Good Luck
John AE5HQ
Quote from: gregcope on July 10, 2016, 04:55:52 PM
Would this irf9z14pbf be able to switch 36v?
It depends on how much current you want to switch at this voltage. If it's just a couple of Amps, then yes. 10Amps, no.
Quote
Data sheet suggests drain source breakdown voltage of -60?
http://www.irf.com/product-info/datasheets/data/irf9z14pbf.pdf
I plan to drive the gate from an N channel mosfet.
I want to power side switch up to 32v.
The problem you will have is the Gate/Source breakdown voltage which is maximum of 20V. If you drive it from a ground referenced N Fet then you'll be driving 32V into the gate. This transistor won't like that... If you use a 2 stage driver, where the first stage pulls the P Channel Gate to >16V and the N channel FET pulls the intermediate P channel transistor's gate to Ground then this will work. You won't need much drive capability for the intermediate transistor, but you will probably need a capacitor at the intermediate transistor's drain if you use a resistor divider to create the intermediate voltage. The capacitor is needed to offset the main P Channel's gate/drain capacitance. You will also need a drain/source breakdown voltage >32 volts in your N Channel FET.
Update: John4444's suggestion of using a Zener or series of LEDs to limit the Gate/Source voltage on the P Channel FET is a good one and eliminates the need for an intermediate transistor.
Update Update: if your supply is fixed at 32 volts then a simple R/R resistor divider on the leg between the 32V supply and the drain of the N Channel FET will give you a safe 16V at the gate of the P Channel transistor. I would add a small cap between the P channel gate and N channel drain to offset the P Channel gate's capacitance.
Tom
I'd just add that due to the parasitic capacitance between the P channel's drain and its gate, and the fact you're switching 32V with large dV/dt, there's a danger that this could couple through from the drain to the gate and damage the gate. In which case a zener protector is probably a good idea. I'd seriously consider slew rate control as well if you're switching 32V, you could end up with current spikes on switching which due to parasitic inductances could also damage things. Here's a useful document on slew rate controlling load switches:
www.onsemi.com/pub/Collateral/AND9093-D.PDF
Mark.
Thanks All, learning loads.
More detail.
The input voltage (VCC?) can be anything from 11v (flat 12v nominal) to 32v (24v being charged). This goes to a 7-36v to 5v regulator that then goes to a lipo charger. I plan on voltage divider in VCC with MegOhm resistors as a low power divider so that I can measure the VCC on an Analog pin on the motineo. this would be before the P channel as I want to (nearly) constantly measure/record this. If voltage is above X and lipo below Y switch on P channel to charge lipo.
I like the idea of LEDs as these would be on when the device is on.
Going back to basics - i think I need some sort of voltage divider to ensure the P channel gate is less than 20v, but more than VGSon? IE something between 3v and 16v?
Do LEDs show resistance?
Here is a good intro article about LEDs: http://www.theledlight.com/LED101.html
Electrically, an LED looks like a 'soft', high voltage drop, silicon diode, it will not conduct (much) until there is sufficient voltage to forward bias the diode. Then, as current builds up, the voltage will increase with increased current. A green LED, will begin to conduct around 2volts, so, if you want a 6 volt protection, you'll need at least 3 Green LEDs in series. Note that each color/efficiency rated LED has its own voltage drop so 2V would be unique to Green, 'normal' efficiency.
Now, as to 'seeing' the LEDs light, your difficulty is the wide range of voltage that you need to deal with. At 11V VCC you'll have 5 volts across the resistor (assuming 3 Green LEDs). At 32 volts you'll have about 20 volts across the resistor (due to 2.5V drop at each LED due to increased current), which will give you a 4 to one range of current. Max operating current on the LED will probably be around 25mA so your resistor will be R=V/I or 20/0.025 = 800 ohms (820ohms is close enough) and should be a 1W resistor to handle worst case power. This resistor will give you approximately 6mA at your LEDs with 11V VCC. This should be enough to light them, but is obviously 4 times dimmer than max brightness (your eye won't notice this much brightness change, however).
Finally, note, to 'see' that the charger is on, you are throwing away 25mA in the worst case.
A 10V zener diode (1N5240B), on the other hand, would provide 8V with only 3uA of bias, allowing you to substantially decrease the power wasted by the circuit. If you put one of these (cathode to VCC, anode to Gate) on your transistor, supply 10uA at 1V drop resistor (100K ohm) with a capacitor matching the DrainGate parasitic capacitance of the power transistor in parallel with the 100K resistor, you won't draw more than 220uA in the worst case condition. You should add a 1M ohm resistor Source-Gate on the power transistor to ensure that it is off when the N Channel FET is off.
Tom
Update: changed diode analogy and a few cleanup touches.
I think I am nearly there.
I am missing a resistor. Not sure where the 100K goes?
See attachement for a really rough sketch of a circuit
Re-reading for the nth time, I think I am missing a capacitor in parallel with the 100K too.
Quote from: gregcope on July 11, 2016, 04:52:38 PM
I think I am nearly there.
I am missing a resistor. Not sure where the 100K goes?
See attachement for a really rough sketch of a circuit
Not bad, Greg, not bad. The 100K resistor goes between the Drain of the N Channel FET and the Gate of the P Channel power FET. A small cap across this resistor will improve the switching speed but probably isn't essential given your light load. The cap should match the parasitic capacitance of the P Channel FET. I'd have to take a look at the datasheet to see if they spec this. As a really rough SWAG, I'd use between 100pf-470pf.
Tom
Many thanks Tom!
V1.1 including 100K.
Also the Data sheet has this, which I think it what you are refering to (Capacitance wise);
Quote from: gregcope on July 12, 2016, 02:20:14 AM
Many thanks Tom!
V1.1 including 100K.
Also the Data sheet has this, which I think it what you are refering to (Capacitance wise);
I don't see the capacitor in V1.1. Yes, Figure 5 is the correct data and you want to use Cgd (Crss in the chart). I'd use 100pF from this data and it should be put in parallel to R4 (100K).
Maybe you'll get your monitor operating this season! Happy boating!
Tom
There is an argument that speeding up the switching is actually what you don't want with load switches, and instead it should switch in its linear region using slew rate control. Otherwise you're going to get large current spikes as the switched in caps are charged up. Also there's a linear regulator rather than a buck regulator and I don't know what the power dissipation of that could be as it could be dropping a lot of volts. If it's charging a LiPo I presume the current is not negligable.
Mark.
Quote from: perky on July 12, 2016, 07:59:15 AM
There is an argument that speeding up the switching is actually what you don't want with load switches, and instead it should switch in its linear region using slew rate control. Otherwise you're going to get large current spikes as the switched in caps are charged up. Also there's a linear regulator rather than a buck regulator and I don't know what the power dissipation of that could be as it could be dropping a lot of volts. If it's charging a LiPo I presume the current is not negligable.
Mark.
Valid points, Mark. In this case, since we have so little data on what is actually being switched, I erred on the side of keeping the P FET in SOA rather than worrying about surge on a rather hefty supply (bank of SLAs). Also, its easier to snip a capacitor than add one later on (after the P FET is toasted ;)
Tom
Will add a capcitor on v1.2 when I get home later!
Also the 7-36v regulator that is being switched is the OKI one that Felix recommends;
http://power.murata.com/data/power/oki-78sr.pdf
Quote from: gregcope on July 12, 2016, 12:59:24 PM
Also the 7-36v regulator that is being switched is the OKI one that Felix recommends;
http://power.murata.com/data/power/oki-78sr.pdf
OK, a switching regulator. This will have a capacitor on board and one on board on its output. Personally I would use a slew rate controlled load switch circuit for this just to keep the spikes down on turn on in case it is a problem, but YMMV.
Mark.
Added 100pF to as C2. Version 1.2