Choosing a Q- Calculating hFE

[BrendanP]

I have to choose a Q to switch on a pump in a project. The pump draws around 5amps @ 13.8V running and 12amps stalled.

I am using a 40X2 3.3V version to control the project.

I have been using this formula here to choose the correct part and R.
http://www.kpsec.freeuk.com/trancirc.htm

I had thought to use ST micros BU941ZT.
http://www.st.com/stonline/books/pdf/docs/5288.pdf

I am having trouble understanding if it is possible to use this part in my app due to the figure I am getting using the formula contained in the Q info web site.

Applying the formula it seems a I need a Q with hFE >3750 ie. 5X 15000/20=3750

The BU941ZT data sheet states the part has a min. hFE of 300.

Any advice welcome guys.

 

[BeanieBots]

I have to choose a Q to switch on a pump

Eh?? Do you mean you want a suitable transistor?

Darlington type transistors for high current = bad idea because they will get very hot.
Use a FET.
The problem is that you only have 3v logic but you might be able to find one that will turn on enough to do the job.
If not, (and assuming you don't want PWM control) you could use an intermediate transistor to level shift but I have no idea what other voltage(s) you have available to do that with.

EDIT:
As for calculating the required hfe.
Assume maximum PICAXE OP current = 20mA
You want 12A collector current.
Hence hfe > 12,000/20 = 600
To ensure good reliable switching for a bipolar device, you should therefore be looking for one with an hfe of about 1000. Good luck. Use a FET!

 

[BrendanP]

Thanks BB.

My understanding is that with FET the gate V has to be higher than the switched V?

If so I was trying to avoid that issue. The device is powered by a 13.8V gell cell. I don't have acces to higher V than that.

 

[Janne]

BrendanP,

Fet gate voltage does not need to exceed the supply voltage. Even the most stubborn ones are usually fully open with 10V supplied at the gate, so 13.8V from the gel cell would be great.

Suggest you get a suitable FET, for example IRFZ44. Easiest way to drive the gate would be to use a pull-up resistor, and an NPN transistor, like Beaniebots suggested. If you need to have the motor stopped when the circuit powers up, use a pull-up resistor on the NPN transistor's base the keep the gate voltage low until it's ready to go.

Don't forget to also add a free wheeling diode to your motor (a diode connected paraller with the motor in blocking direction), otherwise the inductive voltage spike might toast your FET when the motor turns off.

 

[Michael 2727]

The Old standard run of the mill MOSFETs only need around 10V (gate) Max to
fully switch up to their rated Voltage / Current.
Logic level ones do it at 5V or are supposed to.
At lower than the rated (typ) gate V they will still switch but not quite as much current.

Get some Logic Level MOSFET Data Sheets and have a scan of the graphs down the page
to see if any will give you enough current at the intended Gate voltage you wish to use.

At 3.3v there should be plenty that will work for you.

EDIT:
At less than fully turned ON they can run a little hotter, check, use a heatsink if needed.
And what Janne said about Doides for back EMF protection etc.

 

[BrendanP]

I'll look at at some data sheets as suggested, I have to use smd part and the pcb is very tight hence not much ground plane for heat sinking.

If not Ill got the route of the secondary BJT to switch the FET.

 

[BeanieBots]

The BU941ZT is MASSIVE. (and it would need a large heatsink)
You should be able to find a much smaller logic level FET that will give enough current with 3v3 gate drive. (might still need a heatsink).
You could also consider parallel FETs to avoid the need for a heatsink and hence keep the footprint smaller.

 

[Dippy]

As said, use a MOSFET. Avoid Darlingtons wherever possible.

"My understanding is that with FET the gate V has to be higher than the switched V? "
... nearly but not quite.



Opening the Gate.

The resistance of a MOSFET varies with the volatge applied to the Gate , but remember, WITH RESPECT to the voltage at the SOURCE terminal.
With an N chan this is positive.
With a P chan this is negative.


Keep it simple.

If your N chan MOSFET is placed as a LOW sided switch , i.e. between device(Drain) and Ground(Source), you need to raise the Gate voltage to a value as specced in the Data Sheet.
The convenient default for the gate should be 0V. This can be achieved by the driving the micro's I/O low , or setting it to input-mode and having a resistor (say 10K) between Gate and Ground (=Source).

If your P chan MOSFET is placed as a HIGH sided switch , i.e. between V+(Source) and Device(Drain), you need to lower the Gate voltage as per Data Sheet.
This is slightly more complicated as the MOSFET's 'OFF' setting requires the Gate to be nominally the same voltage as the Source. Therefore, any direct connection to a PICAXE pin could mean a destructive voltgae at the I/O.
The safest solution is to put a resistor from Gate to Source, and use a simple signal level transistor to switch the Gate.
Luckily, this inversion means you don't have to think about inverting your code.

You can get Logic Level, so Vgs of 5V prodcuse a nice low Rds.
And there are a number of low voltage gate MOSFETs kicking around for 3V operation.


If you use an N chan as a HIGH sided switch then it gets a little more complicated.
You have to drive the Gate voltage to Vsupply + X volts.
It's easy to do , esp. as there are some nice driver chips that will do it for you.
This method does have certain advatanges in certain circuits as a P chan High sided switch is limited by Vgs max. and low R N-chans are cheap.


For High Speed PWM using a MOSFET switch you will need a driver chip or circuit which has been covered to the point of nauseum over the last couple of years. There are dozens of chips available.



I doubt that much/any of the above will get read, but it's useful stuff to remember and would mean you don't have to wait for replies here and can go straight into design

 

[BrendanP]

Thanks Dippy, I do carefully read your replies and will apply the information you have supplied here.
I know FET questions have been covered on the forum previously.

I am switching the low side of the pump. I only need to switch it on and off. No speed control needed.

 

[Dippy]

OK, low sided. Nice and easy. Just give the Gate a signal. Vgs will be the voltage applied to the Gate.

For a low sided MOSFET consider BSC026N02KS.
http://www.infineon.com/dgdl/BSC026N02KS+G+Rev1.03.pdf?folderId=db3a3043163797a6011637c252b10018&fileId=db3a3043163797a6011637c2a4280019

or 1 or 2 paralleled
STD30NF06L
http://www.st.com/stonline/books/pdf/docs/7292.pdf

The first is ideal but an awkward SMD.
The second may need a pair to keep the heat down but look reasonable.

Always treble-check when looking at Data Sheets.
Never take just the Gate Threshold value as your defining parameter.
Look at the Data Tables and the "Output Characteristics" graph.

There are probably many other suitable MOSFETs. It just requires Eclectic to search them out . (Let him do it , he loves it and it is Christmas!)

Obviously you'll have to incorporate some transient protection/suppression.


Of course, if the above is awkward, then just using a couple of BJTs will allow you to apply 12V nom to the Gate. It all looks pretty nice and simple. 10 minutes with breadboard should have the driving sussed

 

[cdngunner]

Marked for future reference

 

[BrendanP]

Many thanks Dippy, thats an excellent part.

I've spent the last hour or so looking over the data sheet and applying various FET selection formulas google has turned up for me.

Double check my understanding of this please mate.

The pump will draw ~5 amps.

From the "6 Typ. drain-source on resistance" chart when the gate is driven with 3.3V there is Rds of ~.25

therefore
Pd= 5 squared X .25= 6.25 watts

Looking at "1 Power dissipation" chart it appears that the part can handle ~6.25watts even up to Tc of 140C.

So far so good I think.

I don't understand how to calculate Tc rise @Ta 25C in the part if it is dissipating 6.25 watts. I see the formula "R thJA=45 K/W2)" Is that what is used? If so can you explain it please.

One other question, I can't see data for the gates current draw. I presume I don't need to use a R between the picaxe pin and the gate?

 

[Dippy]

Sorry, have lost my marbles ; it's not often someone says I have an excellent part

 

[John West]

I assume he's using the "Q" reference to mean transistor, as that is often the prefix applied to their reference designator in schematics and BOM's. But this is the first time I've personally seen it referred to as such - since the invention of the semiconductor junction. I was thinking he was going to ask about inductor Q, or Q switched lasers.
-

A gate input resistor would, in this case, just make matters worse as you are already under-driving the FET as far as applied gate voltage is concerned.

The voltage drop across .25 Ohm at 5A is 1.25 Volts. With a source of ~12 from a battery it's a reasonable overhead voltage loss, but it's not so good as far as total power used (and dissipated.)

It would be far better to use a FET with a lower on resistance at your drive voltage. 6W is a lot of heat and wasted power. If circuit size and component count is critical, then finding a low on-resistance FET at your drive voltage is the thing to do. It will eliminate a lot of heat and heatsinking problems

I'd either find myself a FET with better gate voltage/on resistance specs, or use one of the other driving methodologies suggested. Running A power FET as a switch at less than full "ON" resistance levels, (lowest resistance) isn't especially good design practice and can cause unusual drive and thermal problems later on.
_
P.S. I always fully and carefully read Dippy's posts, if for no other reason than to prove wrong his comment about his posts not being fully read. That, and the fact that they are invariably quite informative and accurate.

 

[Dippy]

Ah well, when I saw the size of my first post I thought to myself "Lordy, that's too long and boring". I tend to switch off after paragraph 5 if each para uses too many lines


I can't quite see where you get 0.25 Ohms ?
I'm looking at the Output Characteristics for the STD30N.
Say 10A on the Vgs = 3V line.
I see that Vds = <1V.
So, that's a much lower Rds. Maybe I've had too much Sherry.
And it will be lower at 3.3V , judging by the huge difference between 3V and 4V.
And at 5A it's a little tricky to follow the lines, but I'd estimate an Rds of < 0.1 Ohms.
You can parallel to spread load and heat.

And with the Infeon the Rds would be 0.003 Ohms .... beauty!

Alternatively, as I mentioned before, using a couple of transistors you could arrange it to drive the gate from your 13.8V. Then the STD MOSFETs would be in the 0.02 Ohm range.

But feel free to search around. I only spent 10 minutes, so I'm sure there are many MOSFETs better than the STD30N.

Sadly for hobbyists, these days the better specced mid-range MOSFETs are SMD. Might be time to bite the bullet and go for a surface mount.

 

[BeanieBots]

I have to use smd part and the pcb is very tight hence not much ground plane for heat sinking.

Dippy, you need to read the whole thread

 

[Dippy]

Damn that Harvey's Bristol Cream is strong.

Yes, it's time I retired. I'll leave it to you youngsters.
Well, if talking about that SMD one then the Rds is really low.

Have a good Christmas et cetera.

 

[BrendanP]

Sorry guys, I didn't make it clear, I propose to use the Infinion BSC026N02KS. The numbers I posted are based on its data sheet.

When I see schematics I see transistors marked as Q1, Q2, etc. It might be an anglo/au type thing not used in the US?

 

[John West]

In the US they do often label transistors on schematics as Q this and Q that as a reference designator. However that's just a common convention, not a hard and fast rule. Just as U is often the prefix on an IC.

I've also seen transistors with a U prefix, though not as often as Q.

However, they could label them anything they felt like and no one would think the worse of them for it. There is no standard I'm aware of, just a typical human tendency to do as others do. Go along to get along.

As I indicated earlier, the letter Q in electronics does indeed refer to other things though, notably, the "Quality Factor" of an inductor.

Around here most posters simply refer to "FET's," when they really mean power MOSFET's as switching devices. JFET's are FET's as well, but due to their architecture they are more often used as small signal amplifiers.

We throw around a lot of misconceptions and incorrect terminology everyday as we attempt to communicate with one another. It's the nature of language. We're lucky we do as well as we do, all things considered.

Just spell things out as completely as you can, and call it good.

 

[BrendanP]

Thanks John, I'll refrain from that abbreviation here on in.

OK, my maths/decimal point placing was defective, I'll just round it off to .003 Rds like Dippy and it makes it a lot clearer.

Id = 5 amps.

Pd= 5X5 X .003 = .075

Which seems like a very small amount of heat to dissipate.

 

[John West]

That's a very low dissipation alright. It will work just fine. I was calculating Pd from the value of .25 Ohm you'd mentioned above.

 

[Dippy]

Well, I couldn't quite see where Brendan got that value from.
After all, on PAGE 2 it quotes 3.4 milliOhm with a Vgs of 2.5V.
So, when shoving in over 3V it's got to be better than that.

Anyway, jobs a good 'un . Good old Christmas.
I'm off, life is too boozy to be stuck on here.
Have a great time ... and put the nerdy stuff away for a day or two eh?