Transistor Recommendation

Tim-M · 2009-04-01 14:56

I could use some help with a transistor recommendation for a 555 based pwm circuit. The application is automotive so all is powered by nominal 12 VDC. The transistor in question is driven by the pwm output signal and is the switch for a resistive load of about 12 Amps maximum. The 555 signal driving the output transistor is also a nominal 12 VDC. Since the load is resistive, it doesn't matter if it is high-side or low-side switched. Unless you guys say otherwise, I'm thinking that a MOSFET with really low on-resistance would work well here. The goal is full on and off so it needs to be matched to the 12 V drive signal and it wouldn't hurt to have some additional 'head room' above and beyond the 12 Amp maximum load it will see. I figure that the higher rated it is, the less it has to work, and the more bullet proof the circuit will be. The switching frequency is approximately 2500 Hz.

Any help would be appreciated,

Tim

Philldapill · 2009-04-01 15:55

Tim, you might also need a MOSFET driver chip. These work by taking an input signal, like your PWM signal, and outputing a high current source/sink signal to the mosfet gate. The reason I suggest this, is that you are switching the mosfet at a fairly high rate - 2.5kHz. The driver allows the mosfet to turn on/off as quickly as possible so that it generates the least amount of heat. Try TI.com. They have some samples that you can get for free(for evaluation of course...). The part number I'd suggest is UCC37322. It's able to deliver a 9A source/sink which should be overkill for most mosfets, but like you said, make it bulletproof.

As for mosfets, you are right. Use one that has the lowest R_DS(on) characteristic. If you are only using one, and not more in parallel, then you don't need to worry about gate charge. Another "free" evaluation mosfet I'd suggest is an STD60N55F3. It comes in a DPAK and TO-220 package.

Phil Pilgrim (PhiPi) · 2009-04-01 16:16

Take a look at the IRFZ34. It has an R_DS(on) of 0.040 ohms with a 10V gate drive and a 700pF gate capacitance. Your siwthcing frequency is low enough that you could use a small series resistor to limit transient gate current and still not spend a ponderable fraction of time in its linear region.

-Phil

Tim-M · 2009-04-01 17:51

Philldapill and Phil,

Thank you for the MOSFET driver information and part numbers. I was not aware of driver specifically for MOSFETs... I'll be checking into more detail about this. Do I also understand that a driver IC is designed to keep the MOSFET from operating in the linear region as little as possible, such as Phil is alluding to?

Phil, You mention that the 10V gate drive and series current limiting resistor... Do I understand correctly that if the gate current is limited, you can drive the gate with more than the rated 10 volts? Would a driver IC be best used with the IRFZ34, or are you thinking that it may not be necessary?

Would a pull-up or pull-down resistor on the gate be a good idea to limit the switching time (linear region), or would that effect the 555 pwm drive signal too much?

Tim

Phil Pilgrim (PhiPi) · 2009-04-01 19:21

Gate drivers are used with high-capacitance gates to shorten the switching by providing a high current drive for the short duration required to charge or discharge the gate. For MOSFET gate drivers check www.micrel.com.

Check the IRFZ34's datasheet. There you will see that the absolute maximum V_GS range is +/-20V. The series resistor is not there to protect the MOSFET, and there is virtually no gate current once the gate capacitance is fully charged. It's only there to protect the driving circuitry from an overcurrent condition on each transition, since a capacitive load can look like a brief, but dead short until it's charged (or discharged) to the driving voltage.

At 2500 Hz, with 50% duty cycle, you will be spending 2000usec in each state. Now lets say you use a 1K series resistor to drive the IRFZ34's 700pF gate. The time constant then is 700nsec and the surge current is limited to 12mA with a 12V drive. Since 700nsec is such a small fraction of the 2000usec spent in each state, the MOSFET will be spending most of its time either in saturation or in nonconductance, thereby dissipating very little power in the form of heat.

Therefore, I don't think you'll need a MOSFET driver in this app, since a series resistor appears to be adequate.

-Phil

Tim-M · 2009-04-01 20:10

Wish I had your knowledge and skills Phil, thanks for the explanation and calc work!

The 555 output is capable of driving quite a bit more than 12mA, so is there an advantage in switching speed to drop the series resistor value and drive the gate harder than that, or is the switching time plenty low enough not to bother?

Thanks for the link to Micrel for MOSFET drivers.

I'll have another project in the works soon that this will be good information for also.

I appreciate your time and help, very much.

Thank You,

Tim

Philldapill · 2009-04-01 20:32

Tim, the 555 can indeed drive the gate of the MOSFET, but it can't drive it as FAST as a driver chip. For this application, you probably won't need a driver chip since your frequency is so low, but if the MOSFET is getting hot, you might try a driver and see if it runs a little cooler.

Raising the gate drive voltage won't really help all that much. If you are driving it with 10V, then the mosfet will already be fully ON at 10V. The only thing it might do, is get there a little more quickly but that is because you will have more current to the gate.

Phil Pilgrim (PhiPi) · 2009-04-01 20:33

You could certainly drop the series resistor value to get more instantaneous drive current. (The LM555 appears capable of 200mA source and sink outputs, after all.) At a 50% duty cycle and 2500Hz it probably won't matter much. But for more extreme PWM values, it would help with the shorter high or low pulses.

-Phil

Peter Jakacki · 2009-04-02 01:09

Tim,
You say you are switching a resistive automotive load of 12A but that smacks of head-lamps which although being resistive are non-linear in respect to their average current. So if that is the case then 2500Hz is rather high and you could run at below 100Hz by relying upon the thermal inertia of the resistive/incandescent load. At the lower frequency you needn't worry about drivers and just let the 555 do the work.

I know I might be out on a limb but no use going down either track unless you/we have the facts straight which depend upon the WHAT (the load) and the WHY (2500Hz).

*Peter*

Phil Pilgrim (PhiPi) · 2009-04-02 01:33

Peter,

I'm glad you caught that possibility. This would also require the current-handling capability of the MOSFET to be overspec'd by quite a bit more, in order to handle the turn-on surge.

-Phil

Tim-M · 2009-04-02 02:04

You're right Peter,

Here's more of the story...
(I don't want to give it all away since I think the idea has market potential)

The load is in the range of that of head-lamps, true, but it happens to be a resistive heating element.

As far as the frequency is concerned, I understand that 2500 Hz is higher than what is needed for the heating application, but for the prototype I'm using a pre-built 555 pwm circuit that I have laying around and it happens to be fixed at that frequency and I thought that would be OK for a first run. In the end, the frequency can be whatever is best for the job and it's just 2.5kHz for convenience right now. Thank you for your help!

Phil,

The drive current is what I was thinking about, I just didn't know how much difference more would make in this situation. Thanks for more explanation.

Everyone,

What do you think would be an appropriate frequency for this type of load? I'm not looking to hold a specific temperature or anything, just to allow reasonable adjustability of the heat level. Given these parameters, I don't think that the frequency is all that critical... maybe a range from every few seconds to a few hundred Hz?

Thanks again for your help and input,

Tim

Post Edited (Tim-M) : 4/2/2009 2:27:52 AM GMT

Phil Pilgrim (PhiPi) · 2009-04-02 02:09

You can estimate the maximum required drive current by measuring the resistance of the heating element when it's cold. As it heats up and becomes incandescent, the resistance will increase, so the current at a given voltage will decrease.

-Phil

Tim-M · 2009-04-02 02:23

Good idea Phil. I don't own this system so I'll take a cold resistance heater measurement the next time I have access to it, which will be in the next 3-5 days. The cold condition should be the lowest resistance or worst case start-up current draw, right?

By the way, this is a low level heating system with a maximum temperature of 150-180 degrees F, so I don't know if you'd say it becomes incandescent or not... I wouldn't think so.

Tim

Philldapill · 2009-04-02 03:29

Tim, if you're doing HHO, don't use a 555. Use a quad opamp to generate an adjustable PWM signal. The freq is adjustable, and so is the duty. I've made a prototype unit that I'm going to start selling on ebay next week. If you want some info on better ways to handle it all, PM me.

KatyBri · 2009-04-02 03:34

Could someone post an example schematic using the UCC37322 and a MOSFET
·with a STAMP?

I looked at the data sheet and the other comments in this thread, but did not see one. Thanks

Philldapill · 2009-04-02 05:09

katyBri, I can't post a schematic, but the circuit is very simple. The UCC37322 takes one input signal(on the "input" pin), and outputs a high current gate-drive signal for the mosfet. The datasheet has a couple very good common application schematics in it. Just feed the stamp output into the input of the UCC37322.

Phil Pilgrim (PhiPi) · 2009-04-02 05:18

Phil,

Isn't the UCC37322 the chip that gave you such heartburn with your headlight dimming circuit? What was the final outcome of that?

-Phil

Philldapill · 2009-04-02 14:04

PhiPi, you're correct. That is the chip that would spontaneously explode - only with a particular load. If I don't use THAT headlight, the chip works great. I still haven't figured out WHY the chip would blow up when I used that headlight, and to keep my sanity, I gave up on figuring out why. [noparse]:)[/noparse]

Great chip otherwise.

Tim-M · 2009-04-03 14:19

Philldapill,

After a little homework online, I figured out what HHO was. No, I'm not trying to generate hydrogen from water to increase gas mileage, but I am interested in your quad op-amp PWM approach.

Tim

Philldapill · 2009-04-03 16:06

Tim, I'm glad you aren't interested in the whole HHO thing. Personally, I think the crowd that IS into, are just fools with no understanding of the most basic laws of physics. However, I'm glad they'll buy my stuff! [noparse]:)[/noparse]

Tim-M · 2009-04-07 16:40

Do I understand correctly that the IRFZ34 (N-Channel) MOSFET transistor that Phil Pilgrim suggested is intended as a low or ground side switch, as opposed to the positive side? I have read the data sheet and it looks that way, but I admit that I don't understand the test circuits they show there very well.

Thanks,

Tim

Phil Pilgrim (PhiPi) · 2009-04-07 16:47

Tim,

Generally, nMOSFETs, like the IRFZ34A, are considered to be low-side (current-sinking) switches. However, there are some MOSFET drivers which use voltage boosting circuitry to ensure that V_GS remains positive when the transistor is used as a high-side switch. This is not possible without a special driver designed for that application, though.

For high-side switching (current-sourcing), it's usually much simpler to use a pMOSFET.

-Phil

Tim-M · 2009-04-07 21:31

Thank you for the continued MOSFET education Phil, I do appreciate it. As best I could tell, the data sheets were trying to show me that they are for low-side switching. I have a few IRFZ34A's arriving later this week and I wanted to be wiring them correctly for the trial run... don't want to let the smoke out, you know. For this application, it doesn't matter which side is switched.

Tim

Post Edited (Tim-M) : 4/8/2009 12:23:22 AM GMT

Transistor Recommendation

Comments