Transistor issue

@Patduino, dhenry: I thought it was some kind of interference, could you please elaborate more, or perhaps link me to a page where it explains what is going on? I'm not exactly sure what voltage floating is. Also, advice on how to stop this would be appreciated.

Connect a 10k resistor from the base to ground. That'll keep the base voltage 0 when not connected to the Arduino. It's a good idea to keep it there even when it is connected. Voltage floating just means that the base isn't tied directly to any voltage source, so the voltage fluctuates in value, sometimes it has enough to trigger your transistor. The 10k pull-down resistor will tie it to ground so it won't do that. Google pull-down resistor for more. This is a very common problem.

Also, when you connect the Arduino GPIO pin to the base, put a 1k resistor in between them.

BillO:
... I would recommend a Darlington arrangement which is basically 2 transistors, althogh you can buy power Dalingtons in a single device.

See the attached schematic. It will handle a motor that draws up to 10A and is pretty cheap.

  1. The flyback diode is on the wrong place in that schematic, it doesn't protect the transistor when the motor switches off. For that, you need a flyback diode connected in parallel with the motor.

  2. IMO darlingtons should almost never be used, because they have high saturation voltage. If the motor draws a substantial current, this results in a lot of power being dissipated in the darlington, so it will get hot and need a heatsink. A power mosfet is a much better solution in such a case. If the current is not so high (say 500mA or less), then a single transistor will do the job more efficiently and with less heat generation than a darlington, especially if the transistor is a high gain low saturation type such as the ZTX851.

dc42:

  1. The flyback diode is on the wrong place in that schematic, it doesn't protect the transistor when the motor switches off. For that, you need a flyback diode connected in parallel with the motor.

Agreed. Thanks for catching this, maybe a bit too early on Sunday for me. Updated schematic attached.

  1. IMO darlingtons should almost never be used, because they have high saturation voltage. If the motor draws a substantial current, this results in a lot of power being dissipated in the darlington, so it will get hot and need a heatsink.

Not sure I agree it would be a factor in this particular case but generally, I do agree.

A power mosfet is a much better solution in such a case.

I agree. The MOSFET would be my first choice as a switch like this. We were talking about transistors though.

If the current is not so high (say 500mA or less), then a single transistor will do the job more efficiently and with less heat generation than a darlington, especially if the transistor is a high gain low saturation type such as the ZTX851.

Yes, very true. The ZTX851 looks like a pretty good choice.

For what it's worth, the updated schematic now shows the ZTX851.

I agree. The MOSFET would be my first choice as a switch like this. We were talking about transistors though.

When did a MOSFET stop being classified as being a transistor? That's what the T in MOSFET is all about! :smiley:

Lefty

Another thing you should learn about is the proper way to figure out the transistor base current, comes in handy when designing something
You take the switching current and divide it by the minimum hfe(gain) of the transistor and maybe add 5/10percent
thhen you take your base voltage(5v from arduino) minus .7v(the voltage drop acrosss the transistor) and divide that by the number figured out from the current, that will be your resistor to fully saturate the transistor with that load to minimize losses and without wasting extra saturation current or even worse not fully saturating the transistor and creating extra heat

divide it by the minimum hfe(gain) of the transistor

hFE is a concept for linear amplification and has very little meaning for switching applications. Typically, you design a circuit assuming the switcher is driven to saturation: Ic / Ib < 10.

retrolefty:

I agree. The MOSFET would be my first choice as a switch like this. We were talking about transistors though.

When did a MOSFET stop being classified as being a transistor? That's what the T in MOSFET is all about! :smiley:

Lefty

Touche!

BillO:
I agree. The MOSFET would be my first choice as a switch like this.

Mine too, if I was making a PCB for it and could use a mosfet in an SMD package, for example http://uk.farnell.com/diodes-inc/zxmn2f30fhta/mosfet-n-sot-23/dp/1583664RL. Unfortunately, there do not seem to be any medium-current mosfets available in non-SMD packages. It surely wouldn't be difficult to make a 1A or even 2A mosfet in a TO92 package, but all that seems to be available is the 2N7000, which has a continuous drain current limit of only 200mA. Mosfets in TO220 packages are a lot more expensive and physically too large in some applications.

dhenry:

divide it by the minimum hfe(gain) of the transistor

hFE is a concept for linear amplification and has very little meaning for switching applications. Typically, you design a circuit assuming the switcher is driven to saturation: Ic / Ib < 10.

IC/IB < 10 is a just rule of thumb that should work for any transistor, but may not always be practical or the right thing to do. It's not a bad rule for most cases, but you can certainly do a lot better if you take the specifications of the transistor being used into account.

While hFE is a variable quality and is only valid during linear operation, it can be used to help find a more reasonable saturation bias to get the most out of your transistor. A transistor is in saturation when both the BE and BC junctions are forward biased. This will occur when:

VBE>VCE, which beings to occur when IBE x hFEmin > IC. Where the value for hFEmin is minimum for the expected IC.

In other words, saturation occurs when an increase in base current will no longer significantly increase collector current.

To insure adequate saturation we include a factor of 1.5 so that all we need is:

IBsat > (1.5 x IC) / hFEmin

If we use the transistor quoted above, the ZTX851, and we need to drive a 1000ma motor, we can get the hFEmin from the spec sheet. Then we have:

IC = 1000ma
hFEmin=100 (between 10ma and 2A)

Therefore IBsat > (1.5 x 1000) / 100

Or, a base current of just over 15ma, which is a whole lot better for MCU use than the 100ma as prescribed by the rule of thumb. In many cases you can get away with even less current, but this would involve testing each transistor for actual hFE at the required collector current.

BTW, this is the method and assumptions by which I arrived at the 240 ohm base resistor in my second schematic which is a conservative selection. 270 ohms would still get the job done.

IC = 1000ma
hFEmin=100 (between 10ma and 2A)

What's the corresponding Vce?

What's the power dissipation?

What's the temperature rise?

Your transistor would have exploded at that point.

The Vce(sat) vs. Ic chart provides you with better information.

dc42:
Mine too, if I was making a PCB for it and could use a mosfet in an SMD package, for example http://uk.farnell.com/diodes-inc/zxmn2f30fhta/mosfet-n-sot-23/dp/1583664RL. Unfortunately, there do not seem to be any medium-current mosfets available in non-SMD packages. It surely wouldn't be difficult to make a 1A or even 2A mosfet in a TO92 package, but all that seems to be available is the 2N7000, which has a continuous drain current limit of only 200mA. Mosfets in TO220 packages are a lot more expensive and physically too large in some applications.

It seems the manufacturers do not take much notice of the amateur/hobbyist market anymore. Well, not that they ever used to, but at least back 20 years ago there were a lot more through-hole parts available. That being said, the sexier new parts just were not around then. I guess it's was hard to miss what didn't exist.

I have used the IRLU3410 quite a bit. Sure, it's over-kill for 90% of what I use it for, but is a decent MOSFET at a decent price (about $0.75) and is much more compact than a TO220.

dhenry:
Your transistor would have exploded at that point.

What are you on about?

VCE would be about 0.05v @ 1amp with a base drive of about about 18ma. That's about 0.05W in my book.

Are we looking at the same device???

Page 3, conditions for hFE include Vce = 1v, for Ic = 10ma - 2amp.

That's 1w dissipation for Ic = 1amp.

Thermal resistance for to92 is 100 - 200c/w (max 150c/w for this device) -> 150c temperature increase.

The Vce(sat) chart provides you with better indication (typical) of driving the chip into saturation: 10x - 50x. So your drive requirement is anywhere from 100ma - 20ma, typical.

No margin for error.

BillO:
I have used the IRLU3410 quite a bit. Sure, it's over-kill for 90% of what I use it for, but is a decent MOSFET at a decent price (about $0.75) and is much more compact than a TO220.

I have used the IRLU8726PBF, which comes in the same packages - as you say, much more compact than a TO220 - and very low Rds(on). But I've recently discovered the ZVN4306A N-channel mosfet (1.1A continuous Id, 0.45ohm max Rds(on) @ 5V) which is in the Zetec E-line "TO92-compatible" package. There are some P-channel mosfets in the same package, but with much higher Rds(on).

@dhenry

As I said hFE is only valid in linear mode. They are quoting linear conditions for the hFE. My example is in saturation mode. I only use the hFEmin as a guide to calculate a suitable base current to drive the device into saturation under the conditions I quoted.

Please, take your own advice and look at the VCEsat vs. IC plot. with the 240 ohm resistor we are driving at about 4.3V/240 = ~18ma, which would put VCEsat at bit over 0.05V.

@dc42

The ZVN4306A looks like a nice compact device, but Mouser lists them at more than twice the price of the IRL3410 in the small quantities I buy for experimentation. I wonder why they are so costly?

Yes, they are a little expensive. I guess the demand for them is lower than for the surface mount parts. if you can tolerate a high Rds(on), there are some less expensive parts in the same range, for example ZVN4206A.

with the 240 ohm resistor we are driving at about 4.3V/240 = ~18ma

That's not how that chart works.

You take your desired Ic, divide it by the 10x or 50x curve to get the Ib (100ma - 20ma). From that, you calculate your base resistor for 5v drive -> 43ohm - 200ohm.

Alternatively, your Ib with a 220ohm base resistor is 18ma, yielding a Ic of 18ma * 10x - 18ma * 50x, or 180ma to 800ma.

For switching applications, you should simply forget about hFE.

dhenry:
For switching applications, you should simply forget about hFE.

Look, it's not that cut an dried. They are only showing two lines, IC/IB = 10 and 50. There a an infinite number of possible lines you could draw in there, and they behave in a fully predictable manner. Just because they only show 2 lines does not mean that other lines do not exist. Or that if you wander off the space between those two lines by a tiny bit, the transistor will behave in a hugely different fashion. A line for a ratio of 55 would be just a tiny bit higher than the line showing the ratio of 50.

My calculations are right. The transistor, will be in saturation, VCE will be approximately 0.055V with IC @ 1amp, and it will be at less than 10 degrees above ambient temperature.

You are obviously set on this, so I will not try to convince you further. This argument is just starting to get inane. If you want to prove me wrong, get a ZTX851, drive the base from 5V through a 240 ohm resistor, allow 1 amp IC (5V through 5 ohms, perhaps) and take some measurements. If it blows up, or even get's luke-warm, I'll play for the transistor and send you $5 for your trouble. Okay?

Other than that, let's agree to disagree.