Best multiple/many individual MOSFET driver IC?

[artis]

I have a need for a driver ic, here is what I want to do. I have several (10) mosfets of the IRPF type if I remember correctly with the TO-220 type package, I would need a driver with which I can switch those mosfets on and off one at a time in such an order that at least two fets are on at the same time at any instant.

Simply think of it this way, there is a line of mosfets each connecting +ve to a load (light bulb), what I want to do is to have the bulbs light up and then down moving from one side to the other and as they reach the end the first one lights up again. So I need to drive all those mosfets but for any given moment only two fets need to be switched on.

It would also be great if I could control the speed at which the controller runs through the mosfets so that kmy running lights could go slower and faster?

 

[berkeman]

I have several (10) mosfets of the IRPF type if I remember correctly with the TO-220 type package, I would need a driver with which I can switch those mosfets on and off one at a time in such an order that at least two fets are on at the same time at any instant.

Can you post a link to the datasheet? What supply voltages are you going to use? How much current will each light bulb take? Is there a reason you aren't using LED lights instead and simple BJT low-side current drive transistors?

 

[Borek]

Sounds like job for Arduino.

 

[Windadct]

Which MOSFET? You will need a driver that supplies the MOSFET with the correct gate voltage. I can be done with a variety of devices ( opto, level shiftier, etc)
-- Why do you need 2 MOSFETs on at the same time?

Diagram / Schematic ?

 

[artis]

Well I am just sort of toying around with some leftover gadgets but I will also later use this same project for a small tabletop generator project that I have. But let's leave that for another topic.

Why mosfets you ask? Because I'm dealing with low voltages (12v and lower) and they have a low resistance and I have a bunch of irf540 leftover on my desk.
Why two on at the same time? Because I want continuity, if only one bulb is on at any given time it doesn't seem so seamless and also i need such an arrangement for the other thing that I will do with this if it will work.
So two mosfets on at any time, as the next mosfets comes on the one before the previous switches off and so on and on in a continuous progression.

here is a simple schematic, the grey circles are the bulbs the circles with third leg are the mosfets. what do you think how could I best accomplish this?

 

[Rive]

This task does not require a real 'mosfet driver' since that name typically means a bipolar high current driver capable of high speed switching (few amps, up to MHz or so). At the aimed speed any typical logic output would do which can support 5V (a serial resistor and maybe a zener is needed to protect the output). So the actual question would be more about 'driver logic'.
I too suggest a microcontroller (with 5V capable outputs or with level shifters).

 

[Windadct]

Yes - as pointed out - good arduino project, use a level shifter, or actually a basic switching transistor ( 2n2222). Connect the lamps to the +DC, and have the MOSFETs switch to the Gnd. That way all of the electronics are referenced to GND ( the arduino will not want to "see" more then 3.3 or 5V on its terminals)

The MOSFETS want/need > 8V on the gate, is the V+ only 12 V? You can switch the full 12V to the Gate.

Also - for the IRF540 - I would only give it about 3A load without a heat-sink.

 

[artis]

well since this is needed for a test not a real disco lamp or anything like that the amps and volts will be low so heatsinking cheap mosfets is not a problem here.
the speed, well it would be great if i had the option to control it, increase it if necessary but I'm sure couple of KHz will be the highest i will need this to work.

I guess I have to start looking more into arduino type stuff as I have no experience with them.

 

[gneill]

If you run out of outputs on your arduino I suppose you could "tack on" some shift registers and expand the number of controlled lights indefinitely. You'll lose a few arduino outputs to drive the shift register clocking, etc., but you add 8 outputs with each shift register chip, and they can all share the common functions like clocking and reset.

You could clock in different patterns to the shift registers and then run them around your "light loop". Use bi-directional shift registers to run the loop forwards or backwards... Sounds like you can have some fun with this.

 

[Windadct]

I was thinking just turning the lamps on and off - not PWM... that is more difficult, as each channel needs a PWM output if in the KhZ range.. If they are incandescent, you will not need a switching frequency of more than 50 Hz or so where you do NOT need a PWM output from a uController, it is doable in processor cycle, without high-speed timer, even for a smooth rise and fade if desired... For incandescent there is no need to go to kHz,I can not think of a reason - unless you are doing something you have not revealed. (LEDs would be a little different, but then the circuit is different.)

Are you looking to just make "chase" light display? -- Or be able to program essentially any pattern?

What is the wattage of each bulb.

 

[Rive]

Frequent switching might require heatsink (and/or) real mosfet driver, since the dissipation will depend on transient time and frequency. Without driver the transient time might be relatively long.

 

[Baluncore]

There are many hidden traps in this poorly specified project. It is very easy to throw together an unreliable filament globe light chaser.

1. You will need to use a driver between the logic and globe to switch efficiently. If you do not switch the mosfet capacitive gates quickly, the mosfets used will have source-drain voltage and current at the same time, so they will dissipate significant heat. Bigger mosfets cost more but switch slower, so they need faster and higher current drivers.

2. Incandescent filament globes momentarily require 10 times their normal operating current when cold to reach operating temperature. Filament resistance is proportional to temperature which changes from 300K to 3000K.

3. Incandescent filament globes do not last long when subjected to repeated large thermal expansion-contraction cycles. So as to reduce the turn-on current surge and the thermal shocks each cycle, you need to keep the filament a dull red while it is supposed to be off. That suggests using PWM.

Exact filament lamp voltage and wattage specifications will be needed before specifying the drivers. Until then we are all in the dark. Alternatively, rather than using incandescent filament lamps in a digital light chaser, it suggests using LEDs.

 

[artis]

I am very happy to see so many answers here.
To be honest I know that incandescent lamps don't like sharp frequent turn on/off because of the metal filament expanding and contracting etc but that is not an issue here. This is just a test and it is intended for me as a means to verify whether my generator works or not. Mosfets will be fine, we are talking about 12 volts and very small current, the bulbs are small 12v ones so no biggie.
Now since I have limited space and the lamps will be fitted on my rotor and will have to physically rotate I actually thought about using a simple DC/universal motor commutator for this and a brush assembly, this will accomplish mostly what I need as the brush is stationary and so only those bulbs will light up which pass the slots on the commutator that are touching the brush at that moment.

I will write why I'm doing this later if this will work at all.

 

[Baluncore]

Mosfets will be fine, we are talking about 12 volts and very small current, the bulbs are small 12v ones so no biggie.

Now I am wondering why you need to use and drive mosfets in TO-220 packages if there is almost no current? Low voltage mosfets in TO-220 packages will typically switch currents between 10 and 100 amps. They also require a momentary 50mA to 500mA current to charge the gate capacitance in order to switch quickly before they overheat.

Dismissing the relevance of a specification as “no biggie” is irresponsible and has no place in engineering. We each have a different interpretation of what a “very small current” might be. If you do not know the operating wattage rating of your 12 volt globes then you will need to measure it before you get others to recommend an appropriate driver for the unspecified mosfets.

Engineering deals with numbers, not guesses. If you want to guess then you must take “ownership” of that guess by specifying a maximum limit to the globe wattage as a number. You might estimate maximum globe wattage based on the physical size of the globe and the base used, but you must specify a numerical wattage and take responsibility for the value you specify.

Now you suggest dismissing the electronic design and using a physical commutator. I wonder what happened that caused the wind change. Maybe it all got too difficult. That could be because the project was never fully specified. There is a real limit to the number of unknowns that can be accommodated before a design process goes overbudget or collapses into futility.

 

[Borek]

Now I am wondering why you need to use and drive mosfets in TO-220 packages if there is almost no current?

My understanding from the first post is he just happen to have these lying around, so they seem an obvious choice for some quick proof of concept tests.

 

[Baluncore]

My understanding from the first post is he just happen to have these lying around, so they seem an obvious choice for some quick proof of concept tests.

Just because something is lying around does not mean it must be included, indeed there is a strong argument that it represents a liability in a new design. This is especially true when construction must be repeated to make ten channels.

Until the lamp power is specified, and the stocked mosfet part number is known, we can only sit and watch as the confusion increases.

As an example, in this case it could be easier to drive several low-power 150mA lamps directly with one or two TPIC6B595 chips. http://www.ti.com/lit/ds/symlink/tpic6b595.pdf They feature an internal 8 stage shift register and can be cascaded, so can be driven by a few pins on any $10 assembled controller board such as an arduino trinket. https://www.adafruit.com/product/1501

 

[Rive]

Now I am wondering why you need to use and drive mosfets in TO-220 packages if there is almost no current?

At the company we have many alternatives with far better parameters but we are still using these monsters mostly because:
- we have some thousand pieces on stock
- they are still through hole components, easy to fit into protoboards
- if no protoboards at hand, you can still solder their legs
- they are so big that at small currents there is no need for heatsink...

 

[artis]

Ok , sorry guys for the confusion. Borek already took the word out of my mouth, yes I use them because those are the only ones I have at hand leftover from earlier projects. TO220 package with small currents will let me use them without heatsink. They are N channel IRF540.
The bulbs will be 12v 10w car signal lamps.

I know from distance this may sound as me backing out because microcontrolers are more complex than a bunch of copper and carbon but I actually understood that for me testing out whether my idea works at all with respect to the laws of physics it is more convenient for me to use as few as possible cheap and simple parts and since space is also limited as I'm doing this in an old small generator housing I modified to suit my needs I settled for the commutator as the means of achieving mosfet driving needs. as for the gate charge I have no problem with that as I will feed the commutator with a sufficient 12V source with enough current.


I might ask a question, now since I don't have a driver IC for this , given that the switching frequency won't be too high (not more than few Khz at best) what should be the value of resistor that I would need to connect the gate to ground in order for fast enough discharge?

 

[Rive]

the switching frequency won't be too high (not more than few Khz at best)

That is already high enough to require engineering, and we don't have enough details to provide any worthy guess. What kind of driving output will you use?
Also, you have to be aware that multiple switchings will multiply the dissipation on the switching element: you have to check if heatsink is needed or not.
Switching 12V 1A at a few Hz can be already more than enough to burn a standalone TO220, depending on the switching times.

 

[artis]

Well the mosfets may get a bit hot but then again I just need to run the thing and measure my results for a short period and then I will know whether it works or not , if it indeed does work the way I intend then I have no problem of heatsinking the transistors later on.

since i will be using a commutator then the driving will be simple, +12v applied to he commutator brush and as the rotor spins the slot contacting the brush at the time will deliver those +12v to the gate of the corresponding transistor as the slot will move away from the brush the transistor should in theory switch off if my gate discharge will be made correctly.
Should be simple.
See this whole thing is not so much about the transistors and loads as it is about me testing out whether a device I made works the way I think it should. The so called engineering will come later if it will be necessary at all because just in case I have gotten my physics wrong then there won't be any need for extra engineering and heatsinking transistors.

 

[Baluncore]

Why do you not drive the globes directly from the commutator which would cut out all the static sensitive semiconductors?

 

[artis]

because my intent is to primarily test whether my generator works, the bulbs are just a confirmation so that it is easier for me to test because the rotor is rotating.

after i get my commutator and do the test I will then probably make a thread here, if the approach doesn't work out then we could figure out together why as I feel it will be an interesting question with regards to the laws of physics and special relativity.

 

[berkeman]

See this whole thing is not so much about the transistors and loads as it is about me testing out whether a device I made works the way I think it should.

I feel it will be an interesting question with regards to the laws of physics and special relativity.

Say what?

 

[artis]

Is that "Say what" ? meant in a "lil John" style or simply a curious question? :D

 

[berkeman]

How did you go from a simple FET light bulbs DIY project to a simpler commutator-based light bulbs DIY project to testing Special Relativity?

 

[artis]

Well I actually did not go from that to that, I had the intention to do all these things I need in the first place just that in order not to make the thread very lengthy I asked piece by piece.
I would love to make a different thread about what exactly I'm doing but that has to be done some time in the future because I first need to test out everything I want and see what happens and then sit down get all the facts straight before I make any comments or threads.

 

[Javier Lopez]

There are low voltage mosfets that can be activated directly from 5V logic devices.
Here are some of them: BSL215CH6327XTSA1
In order to calculate maximum power you should multiply lamp average current by mosfet Rds on resistance.

Low voltage mosfets

 

[Baluncore]

since i will be using a commutator then the driving will be simple, +12v applied to he commutator brush and as the rotor spins the slot contacting the brush at the time will deliver those +12v to the gate of the corresponding transistor as the slot will move away from the brush the transistor should in theory switch off if my gate discharge will be made correctly.
Should be simple.

One problem with carbon brushes on metal commutators is that they need to run a minimum of about one amp to keep them clean and conductive. For that reason it is more efficient to drive 10W / 12V filament lamps directly from the commutator than it is to insert a MOSFET as a lamp driver.

When a MOSFET gate is controlled by a commutator there must be a resistor to provide the 1 amp of contact wetting current or the gate voltage can float so the MOSFET will remain at half voltage * half current which may be more destructive than no voltage * maximum current, or maximum voltage * no current. The situation can be aggravated by the higher currents drawn by cold filament lamps.

 

[artis]

No worries, this is just a test setup, my homemade generator current and voltage is so low the MOSFET could be half open all the time and won't self destruct. We are talking about maybe 3V and few mA of current at best. An while I believe what you said applies to constantly running devices , mine will run for few mins probably which will be enough to determine the result.

Speaking of the result, I just got my commutator and I plan on setting up my setup today, now what would be the best value resistor to put between gate and ground in order for the mosftet to switch off quickly enough? My gate drive will be 12V from a car battery, which will be applied to the gate via the commutator. the ground to source. mosfets are N channel IRF540

 

[Baluncore]

now what would be the best value resistor to put between gate and ground in order for the mosftet to switch off quickly enough?

12 ohm. I would suggest you use a 12 volt filament lamp.

 

[Rive]

On the other hand, if current is flowing and it is used for control purposes, then switching off is no simple matter. In such case I would rather choose the limited reliability/lifetime.

I don't know what are you trying to do with relativity, but better do the math first because any (mechanically feasible) effect is usually too small to be measured.

I have a feeling that this won't end well. But please keep us updated anyway.

 

[artis]

I did not understand your concern Rive?

 

[Rive]

Which one?

Regarding relativity, it is exactly as I said. The effects are usually too small (needs math to see how small) to be measured, and even attempts needs rigorous accuracy both in design and actual doing. Mechanical switching - well, no way accurate enough. But since I have very little information about what are you trying to do, I can't say much.

Regarding switching: mechanical switching is an art in itself.
As an example, see this or this.
Since you are trying to control a MOSFET directly, the dissipation will inevitably increase, but with the available information there is no way to give an estimate. You have to be careful.

 

[Baluncore]

MOSFETs do not mix with brush commutators without signal conditioning networks and bipolar gate-drivers having stable power supplies.

My concern is that artis is asking for technical advice on how to build a leaning-tower of left-overs, something that will continue to rise in height and complexity until it collapses under the number of components to be mounted, along with the profusion of electrical connections that will have to be made.

The KISS principle says, make it work now with the minimum, a commutator and the light globes is all that is needed.