Trouble finding a high powered mosfet

Im wanting to make a dc to dc buck converter for a solar panel array. Sparkfun has a couple 60 v 30 amp p and n channel mosfets. Which would allow my to wire 3 solar panels on in series. Problem is, i was really hopeing to be able to wire my solar panels in a 4 panel series which would require an 80 volt p and n channel mosfet preferably logic level and i cant seem to find one.

Does anybody know of an 80v logic level p channel mosfet? There are several 80 volt mosfets to be had but getting the gate voltage up to 80 volts is an issue in a complementary pair. For those that dont understand complementary pair. Im wanting the gates of both the p channel and n channel to be connected to the same pwm output.

If there is no such thing as a 80 v logic level p channel fet. How much of a logistical nightmare am i looking at to get an 80 v p channel gate up to 80 volts without destroying the gate of a n channel.

I just woke up and my brain feels kinda fried from thinking to much last night so pardon if i confused n channel with p channel.

What are youo trying to get out of this solar array? A squarewave? battery charging?

regards

Allan

What is your experience in designing buck converters?

Why does the p-channel mosfet need to be logic level when you have up to 80V available to turn it on (especially when the gate to source voltage of most mosfets is 20V maximum) ?

[qoute]What are youo trying to get out of this solar array? A squarewave? battery charging?

regards

Allan [/qoute]

A battery charging system.

[qoute]Why does the p-channel mosfet need to be logic level when you have up to 80V available to turn it on (especially when the gate to source voltage of most mosfets is 20V maximum) ? [/qoute]

Again i was wanting the mosfet to funtion in complement to the n channel mosfet.

On lower voltage circuitry, correct me if im wrong, you can connect a p channel and an n channel gate to the same output pin. So that if pwm i/o 1 is high the n channel mosfet is on and the p channel is off. When the i/o 1 is low the n channel mosfet is off and the p channel mosfet is on.

Yes one can use the 80 volt power rail with drop down resisters to control the p channel fet with an n channel fet. The issue with this, os you then require 2 pins which will cause an open short do to latency issues.

Post a schematic of what you want to do.

It sounds like you want something like a high side gate driver, but since you admit that your brain is fried right now I don’t want to make any further assumptions about anything.

You’ll need gate drivers even for the low-side FETs too. High power, high speed PWM like what’ll be happening with this converter is not something you do with logic level FETs driven by a microcontroller’s output buffer. The specs of the switch and inductor are the two biggest factors in switching converter efficiency, so you want to get switch losses down as low as possible. That means finding the lowest RDSon that you can, driving it with the highest gate voltage you can muster (more gate voltage above the threshold = less switch resistance), and slamming them on and off as quickly as possible with high current pulses so they spend minimal time in the linear region.

Find a suitable non logic-level >100v P-channel >30A mosfet - 8v gate turn-on or similar would be fine.
And a big diode capable of >30A. And a small high voltage NPN transistor

Then look at the attached. Turn on the fet when solar volts> battery volts and the battery isn’t fully charged… I’d build in a bit of hysteresis. Turn off the charge once in a while to check the volts…
If you want to PWM it be my guest.

If you want a fancy buck/boost convertor I suggest you buy one - they’re not trivial.

regards

Allan

solcharge.pdf (21.2 KB)

On lower voltage circuitry, correct me if i'm wrong, you can connect a p channel and an n channel gate to the same output pin. So that if pwm i/o 1 is high the n channel mosfet is on and the p channel is off. When the i/o 1 is low the n channel mosfet is off and the p channel mosfet is on.

Yes, you can do this if the supply voltage is the same as the logic supply, but not when you are trying to switch 80V from a 5V logic level.

My drawimg capability sucks when it comes to schematics. But since a few people here know a think or two about buck converters somewhere else someone brought up the idea of replacing the diode with a mosfet. Question was on lower powered systems would this have better efficiency than a diode.

On a high powered system i was wondering if this might be scalable.

Would using a high powered p fet. Drop down resistors and a low powered n fet be a better way of switching the high powered fet?

Ninjarider2003: My drawimg capability sucks when it comes to schematics. But since a few people here know a think or two about buck converters somewhere else someone brought up the idea of replacing the diode with a mosfet. Question was on lower powered systems would this have better efficiency than a diode.

If you have high output currents (several amps), the power dissipated by the diode can become significant, up to a few watts. If the ratio between Input and Output voltage is very high (say, 12V -> 1.8V), the power wasted in the diode is a HUGE hit on efficiency.

Replacing the diode with a MOSFET is called a synchronous converter. The lower voltage loss in the MOSFET improves efficiency.

What is your input and output voltage, and output current requirement?

On a high powered system i was wondering if this might be scalable.

Scaleable to what? Higher voltages? Higher currents? More panels?

If you're asking if something is "scaleable", you need to specify what you're planning on scaling.

Would using a high powered p fet. Drop down resistors and a low powered n fet be a better way of switching the high powered fet?

Use gate drivers on every FET.

Ninjarider2003: Im wanting to make a dc to dc buck converter for a solar panel array. Sparkfun has a couple 60 v 30 amp p and n channel mosfets. Which would allow my to wire 3 solar panels on in series. Problem is, i was really hopeing to be able to wire my solar panels in a 4 panel series which would require an 80 volt p and n channel mosfet preferably logic level and i cant seem to find one.

Does anybody know of an 80v logic level p channel mosfet? There are several 80 volt mosfets to be had but getting the gate voltage up to 80 volts is an issue in a complementary pair. For those that dont understand complementary pair. Im wanting the gates of both the p channel and n channel to be connected to the same pwm output.

If there is no such thing as a 80 v logic level p channel fet. How much of a logistical nightmare am i looking at to get an 80 v p channel gate up to 80 volts without destroying the gate of a n channel.

I just woke up and my brain feels kinda fried from thinking to much last night so pardon if i confused n channel with p channel.

Why would one use p-channel MOSFETs for this? Do you understand the tradeoffs and design principles of power electronics? I suspect not and you are guessing. High voltage power conversion is not a beginners project, you will at best blow up lots of MOSFETs, at worst give yourself a life-threatening DC shock (80V is well into the heart-stopping dangerous range).

The basis of all power conversion circuitry are switches/inductors/capacitors, the control logic and the protection circuitry. At 80V the protection circuitry is all important and you have to worry about dV/dt issues in the switches, stray capacitance, shoot-through, inductor saturation and such like. If you get things wrong you blow up your circuit. The energy stored in the DC bus capacitor at 80V is enough to do damage in microseconds when things go wrong.

If you want to build your own design you have to take this on, be prepared for failure, and take care not to electrocute yourself. I would recommend starting at 12V, gradually build up to higher voltages, higher powers, and you will need an oscilloscope to check on the switching waveforms.

Why would one use p-channel MOSFETs for this?

Correct me if im wrong but i thought the rule of thumb for mosfets was if the fet has a clear line to hot use a p channel. If the fet has a clear line to ground use an n channel.

The better rule of thumb is use the standard MOSFET drivers and all n-channel because they are three times better in performance than p-channel. Standard MOSFET drivers are designed to drive all n-channel as a consequence of this fact, despite it necessitating bootstrapped floating gate driver supply.

Also the floating gate driver design will work independently of the main power supply voltage, which can be less or more than the 12V normally used for MOSFET drivers.

What is your input and output voltage, and output current requirement?

Input voltage 80v 5 amp, output 14v at 40 amps

MarkT so your suggesting using an n channel fet with a bootstrap capacitor and try to accurately keep the capacitor at 15 volts. And when i need it to swing up make sure not to over 100 volts. Think i trust myself with a p channel fet cause i have no idea how to even design such a rig. Maybe some drop down resistors 10:2 ratio diode tap and a fair cap. Would give me 16 volts in the cap. But the inductor makes it all screwy.

If you're charging a 14v lead acid battery and have 4 20v solar panels, why not parallel them?

If they give max output 5A each you'd have a max charge of 20A, and a variant of my circuit would be fine.

And no nasty high voltages.

regards

Allan.

Ninjarider2003: Input voltage 80v 5 amp, output 14v at 40 amps

MarkT so your suggesting using an n channel fet with a bootstrap capacitor and try to accurately keep the capacitor at 15 volts. And when i need it to swing up make sure not to over 100 volts. Think i trust myself with a p channel fet cause i have no idea how to even design such a rig. Maybe some drop down resistors 10:2 ratio diode tap and a fair cap. Would give me 16 volts in the cap. But the inductor makes it all screwy.

I suggest you study how chips like the HIP4081 work from the datasheet - the bootstrap cap is managed for you by high-low gate driver chips. Typically the gate voltage used is 12V for MOSFETs, 15V for IGBT's

And I'll repeat this is non-trivial, gate driver chips exist for a very good reason, driving high power MOSFETs without them is going to be a learning experience. If you go for the learning experience definitely opto-isolate your ARduino from the power stuff.

Allan.

Factor 1. For simple cost of installation. Higher voltages low current can transmit through thinner wire. Low voltage high current needs a thicker wire. Im going to start of with 1 panel and work my way up as i go. During early spring we buy poultry that requires heart lamps 24/7. I have estimated that i may need up to 16 panels during my peak usage months. I could do 18 volts at 80 amps, Or i could do 80 volts at 20 amps. For maximum light saturation installed on pole with a sun tracking system i could be looking at 40- 50 ft of wiring. 80 amps to travel 50 ft with a 3%volt loss requires 3/0 wiring while 20 amps traveling the same distance would need 4 gauge. Now the aluminium 4/0/3 wrote at the hardware store might come out cheaper than using 4 gauge and will also play a factor.

Factor 2. At 18 volts i have to wait longer in the day to achieve a voltage capable of charging. At 80 volts i can acheive a charging voltage just after sun up.

Now i have played with the idea of getting a rotary tumbler switch made turned by a stepper motor to change the panels back and forth from 4 series to 4 parallel.

MarkT i will look into that chip. Got a stashed on it pulled up already.

OK - have your panels in series....

But now you still have a problem with the ?-80v input, 14v ?amp convertor - as I mentioned earlier, such devices aren't trivial - it might be cheaper to buy thicker wire by the time you've blown up a few prototypes and maybe yourself.....

2.5^mm solid copper as used in the UK for house wiring is reckoned ok to 22A.

The output voltage of a solar panel rises fairly quickly with illumination - it's the internal resistance which drops as the light gets brighter, so you may not gain that much.

Do you want something that works, or months of frustration and magic smoke trying to get something a bit better..? How many high power switchmode psu's have you designed? There's a lot more to it than just driving the switches.....

regards

Allan.

I'd say series everytime, much less copper needed. You know aluminium wire is approx 3 times poorer at conducting than copper (size for size) don't you?

Remember that wiring in series requires reverse diodes across every panel to prevent back-driving when a panel is in shade. Depending on what type of panel these may already be fitted. You also need a diode in series with the whole array to prevent current flowing the wrong way when dark. Again some panels may already one per panel already to make wiring up simpler.

If your charge controller can handle MPPT from a wide range of input voltages then all in series is a good approach (though you might want to limit the highest voltage in the system for safety reasons)

Hi MarkT..

of course you're right - perhaps you could provide us all with a detailed debugged design and layout for a 5-90v in , 14v 40A+ output buck/boost convertor .

Wouldn't mind such a gadget myself. Nice bit of kit.

regards

Allan

Maybe combine a 5V and 9V with stacked (series) outputs http://www.mpja.com/5-Volt-Power-Supply-40A-240W-Switching-Hengfu/productinfo/18434+PS/ http://www.mpja.com/9-Volt-Power-Supply-17A-150W-Switching-Hengfu/productinfo/16019+PS/ and split the circuit so it's not all feeding off the same 14V source?