driving N channel MOSFET with high side chip

[hax]

On the topic of MOSFETS (again).......sorry......


I didn't want to ask this question in the other current threads as it would just confuse the situation with talk of the different ways you can control a MOSFET.


So here is my situation.... I want to control an N-channel mosfet (IRL3803) on the high side of a 12V power supply. I will be dimming a LOT of LEDs with PWM.

I want to do it properly to minimise heat dissipation so I will be going for a dedicated N-Channel MOSFET IC from International Rectifier. I have chosen the IRS2001.

The IRS2001 has two outputs, one for a high side N-channel fet and another for a low side fet.

My understanding is that for the fet to turn on while in the high side, its gate has to be supplied with a voltage that is 10V higher than the power supply (in this case its a 12V power supply so I need 22v on the gate)

Stop me if this is incorrect....

Now my question is, looking at the first diagram on page one of the datasheet attached, how on earth can the IRS2001 generate a voltage greater than VCC without actually being connected to the high side voltage????? There is no talk of a charge pump in the description and in any case it does not know what potential the fet is connected to...

 

[moxhamj]

What a cunning little chip. The secret is probably in the "HV level shift" part of the functional diagram, and may use a very high frequency and small caps that are on the silicon die. It appears to store that charge in the external capacitor across VB and VS. It would not need very much current at all to do this and could have a charge pump internal to the chip. List price is only $1.06

 

[BeanieBots]

Neat indeed. I would suspect something along the lines of what Dr. describes but it is very difficult to get much capacitance on a die. (Otherwise op-amps, regulators etc would have them built in).
What I would suspect is that there is enough to make the FET start conduction and the rest is done by boot-strapping.
Only my guess, I honestly don't know for sure.

 

[Michael V]

Max 620/621

Hi Haxby,
In the search for high side n- Channel (know what that is now) control on the high side i came across the MAX 620/621 chip. I found it at futurlec, http://www.futurlec.com.au/Maxim/MAX621CPN.jsp it wasn't at farnell.

This chip actually has four of high side drivers for N channel mosfets, and takes the input direct from a microcontroller. Apart from the Mosfets the only external components are a few capacitors. The data sheet specifically talks about a charge pump to get the voltage above the gate (hope i said that right) so that might be the answer to your question.

Sounds like a very handy device for PWM and stepper motor control if you want to or need to be on the high side, costs $13, not surprisingly about four times the IRS2001.

At the same site the MAX 622 is a single high side driver, also talks about pumping the charge. Bit like the petrol prices at the moment.

The dates on the data sheets are 1994, so it sounds like these things have been around for a while.

 

[hax]

Thanks for the heads up Michael. I'll be experimenting with a whole bunch of setups. I have some N channel and P channel fets coming from Futurlec at the moment. (hoping they are not fakes as the price is 1/5 that of other sellers)

 

[sjremington]

Now my question is, looking at the first diagram on page one of the datasheet attached, how on earth can the IRS2001 generate a voltage greater than VCC without actually being connected to the high side voltage????? There is no talk of a charge pump in the description and in any case it does not know what potential the fet is connected to...

My reading of the data sheet, which is pathetically unclear, is that the driver chip must be connected to voltage levels comparable to the high side voltage via the Vb, Vs and Vho pins. Note from the Absolute Maximum Ratings table that Vb can go as high as 225 V and that Vs and Vho cannot be less than Vb - 25 volts. The "typical connection" diagram is very misleading because the high side voltage cannot be greater than 50 V, when Vcc is connected to Vb as shown.

The pulse generation circuit shown in the equivalent circuit is almost certainly a charge pump. This would explain the diode in the schematic. It is a pity that they cannot suggest a value for the cap between Vb and Vs. I suggest that you try to find the Application Notes and Design Tips referred to in the typical connection diagram.

Cheers, Jim

 

[hax]

Yikes, I thought it was too good to be true. I've bitten the bullet and I will go with a standard P channel Fet for high side switching. Less headaches.

The high side driver does make things more complicated, and for the saving of a few milli ohms of on resistance (for an N channel fet) its not worth the effort in my application.

Now to interface with the P channel fet, I was looking at a silicon chip article

http://www.siliconchip.com.au/cms/A_109490/article.html

That circuit uses a TLP251 which is a mosfet driver and opto-coupler in one. The diagram shows it in a H bridge foration, but if I only wanted to switch one channel, am I right in connecting pin 8 to 12V+, pin 5 and 3 to gnd, pin 2 to picaxe PWM with resistor and pin 6 to the fet Gate?

This would drive the fet very hard. I would have to watch the voltage and current on the output, and maybe have a resistor and zener diode on the gate to ensure that the current does not go over specifications for the TLP251 (400ma from memory) and not over 10V.

Any other comments?

 

[moxhamj]

Re So here is my situation.... I want to control an N-channel mosfet (IRL3803) on the high side of a 12V power supply. I will be dimming a LOT of LEDs with PWM.

Given this is getting complicated, what is the reason the driver has to be on the high side?

 

[hax]

There is other circuitry on the low side which selects which LEDs should be on through darlington transistors. So I only have one common high side connection to all of the LEDs. Hence I need a high side switch.

 

[BCJKiwi]

So having decided to go with a P-channel MOSFET, does the one you have (or are going to purchase) have a logic level gate?

If so, you won't need the driver chip setup as the PICAXE will be able to drive the gate directly. OK the gurus in this area will advise how to protect the PICAXE port but the current required is negligible and the MOSFET will be hard on at < 5V.

I will look forward to the Gurus' recommendations as I have a P-Channel MOSFET being PWM'd in response to an LDR to dim a bunch of LEDS hanging off I/O expanders.
Right now this is all on a breadboard and there is just the direct connection from the pwmout right to the gate. Has been working fine. MOSFET is a Fairchild FDP4020P

 

[moxhamj]

The advantage of mosfets over transistors is they consume less power, but if you already are using darlingtons on the low side then why not use transistors on the high side as well? They are simple to wire up - use a PNP on the high side with a NPN pulling the base down.

Then again, if you have darlingtons on the low side and mosfets on the high side, why not have 5V gate drive mosfets on the low side and PNP/NPN transistors to drive the high side?

 

[sjremington]

Rethinking the driver chip circuit

Haxby:

After rethinking the "typical circuit" for the driver chip, I think it is OK as advertised. That external diode is the key. There is no charge pump.

To analyze the circuit action begin with the high side power mosfet off--since Vs is connected to ground through the load, it is at ~0V. The output mosfets in the driver chip won't be on at the same time, so the capacitor will charge to Vb = ~Vcc through the diode and the load, while the gate of the power mosfet is held at Vs = ~0V.

When the high side mosfet in the driver chip turns on, Vb will be applied to the gate of the power mosfet; it will begin to conduct and Vs will float up to the high side supply. In turn this will bring Vb up to (high side supply + Vcc), so the power mosfet stays on. The cap could be quite small, e.g. 0.1 uF or less depending on the gate switching current and frequency.

Vcc should not need to be more than about 12V in the worst case. Try it and let us know!

Cheers, Jim

 

[hax]

Hi BCJ, I have some IRF4905 P channel fets on their way but I don't think they are suitable for direct switching with PWM output of the picaxe.

Their gate capacitance is 3400 pF which is very high. Hence they need a beefy driver.

Your FDP4020P part has only 665pF which makes it easier to drive from the picaxe directly. But your part won't be suitable for my application as Rds on is 4 times higher at 80m ohms. Hence it will require a heat sink when switching more than an amp or two.

As with most electronics component choices, it's all about weighing up the pros and cons. A high current fet has high gate capacitance to overcome. A low current fet has lower capacitance, but higher internal resistance when on. That seems to be the trend from what I have found.

In my application I'd prefer not to have a heat sink and just use a beefy fet with a beefy driver.

Hence my question whether anyone can see a problem with wiring up the TLP251 in the attached way...

The picaxe is isolated and is just driving an LED.
The TLP251 switches between 12V and Gnd. To ensure that it does not go over its rated 400ma current capability, a resistor is used, low value say 100 ohm.

A zener just greater than 10V is used to ensure the gate of the P channel fet is not over driven...

 

[Dippy]

I honestly think there are some slight misunderstandings driving MOSFETs particularly Ps.

Yes, BCJ's logic level MOSFET can be driven OK direct from PICAXE if the Vss is 5V too, but you have potential (pun intended) problems at higher Vss.
BUT stick a nice load onto that MOSFET and 'scope the Gate and drain. I bet you that you won't see a sub-microsecond transition squared wave. But for lower power and lower F it will prob be fine - but, as teachers used to say to me "could do better".
MOSFETs are more than just voltages at the gate to trigger them.
You all know they capacitance, so if you don't get that charge out/in quickly you won't switch that FET quickly and it will spend a significant amount of time acting as a resistor. For Power applications that means hot. Well we all know what charge/time equals eh?

That TLP251: With an absolute max of 400mA it really is only capable of driving the lower end of the Power MOSFET spectrum. The driver I use is capable of nearly 2 amps. And the TLP251 switching time of 1uS - no doubt better than most home brew drivers, but to get the best out of any MOSFET you really should be looking at gate transitions of sub-250nS.

That Si-Chip article shows a nice driver, but boy it's over-complicated IMNSHO.

And generally in many (not all) examples I've seen PWMing a motor: why PWM the 'top' and 'bottom' elements of each H arm/leg? Pointless.

 

[hax]

Dippy, which driver do you use?

 

[Dippy]

Well keep it a secret as it's a pukka job, but for SM driver I'm using a TC4431. I only remember that as I'm CADding the circuit now Memory retention of a goldfish.

For the H-Bridge driver I can't remember offhand, I'd have to find out - but compared to that Si Chip circuit it replaces all the TLP251s and logic with one chip. I'll have to search as its not in front of me at the moment and i can't remember the number as even I have not reached full anorak status yet.

Edit: Just realised I ought to be clear here, I am talking driving P channel with a driver... don't want to cause confusion, but probably have already.

 

[hax]

Righto, I'll go for the TC4420CPA from Futurlec.

It's a 6A driver !! If that's not plenty to switch the fet, I don't know what is.

 

[Dippy]

Yes, that is the one I would have used but I needed a higher supply voltage figure.

Just one single thing to think about: the supply bypass capacitor. DO have a look at the Data Sheet on that subject as it IS important. And to get the best performance a nice tight PCB layout is always best.
It'll work a treat.

"If that's not plenty to switch the fet, I don't know what is. "
- well all I'll say is that someone thinks a bigger one is necessary
Check out the TC4421.

Hopfully, someone will find an even bigger one and we can all go wow!
(NOT).