question re: transistor selection

[spamiam]

I have blown up more transistors in my life than I care to think about, so I am going to ask before I buy some more.....

I want to use a general purpose NPN transistor.

I will be driving a peizo buzzer from the main 13.6v automotive supply.

I will be driving the base with the output of the ZX24 through a resistor. I was thinking something like a 10K ohm.

So, when I look at the specs for a certain transistor, I see that a 4volt absolute max emitter-base voltage. Since the ZX will be sending 5 volts, maybe I am going to exceed the base voltage. But since I have a 10K diode, I would assume that the base voltage will NEVER get anything close to this.

So, is this transistor going to be OK to use with the ZX?

Also, since I want to be able to deliver at least 40ma through the collector (just to be sure, the buzzer draws 10ma or so), I would need an Hfe of
0.040*10K/5 = 80. Is this correct?

I was thinking that a transistor with 70-100 Vceo, 4Vebo, 500mA Imax, 100 Hfe would be about right for use with a filtered (1000uF cap & 27 volt TVS after a rectifier diode) automotive power supply.

-Tony

[mikep]

Aha I see your problem Tony - it's those 10K diodes that get you every time.

I have 3 general-purpose NPN transistors in my parts box:

• 2N3904 for signals
  
• 2N2222 metal can for low power driving
  
• TIP31 for high power driving

I haven't checked all the specs but I usually find these are adequate for most purposes.

[twesthoff]

Tony,
Mikes transistor list is good. Add to that a 2N3906 for when you need a PNP transistor, and you should be OK.

The emitter-base voltage you are refering to is the reverse biased voltage and for your purposes you can ignore it. You will be forward biasing the emitter-base junction (+ on base, emitter grounded), so the voltage should never be higher than .7V anyway.

Your calculations are close, and will work OK, but they are approximations. To design your circuit properly, you would need to know the exact characteristics of your load (the piezo) and work backwards from there.
A 2N3904 will work just fine. Usually you only double the amount of current you will need for switching applications, not 4X, so pretty much any 2N3904 will work. These days they are at least a gain of 100, so you are good to go on that account. The 10K is fine, but if you would like to see the more exact calculations, let me know the specs on your load, and I (or someone else) will show you how to do it.
Tom

[dkinzer]

[pdubinsky]

[spamiam]

Thanks for the feedback.

I appreciate the input. I will look for the transistors you specify. It is good to hear that I was on track with my guesses.

As for using the conditioned power from the OBD of my car, well, I will probably be using this in an OLD car with no sophisticated electronics at all, and I may have to deal with solid copper core ignition wires. So spikes and transients are expected.

This is why I was planning on something more than a 40v transistor. I was thinking 80v or so. As for the base resistor, I made a mistake. I am using a 2.2K with a 10K pull-down resistor. So, with an Hfe of 80-100, I think I will be able to driv ethe piezo buzzer properly.


The TVS I am using is a 1.5KE27CA. 1500 watt bidirectional. As I said, this is after the rectifier diode, and is in parallel with a 35v 1000uF cap.

So far, in a modern car, with power out of the cigarette lighter it has not glitched in a test unit. In that one I was using a 500w TVS, but I decided to go with a heavier one.

-Tony

[JC]

Schmatic and thoughts attached.
2N3904 may well be out of spec for this application.
Do not forget THERMAL considerations, esp in a hot, automotive environment. You may need a small heat sink on your transistor.
JC

[dkinzer]

One thing to keep in mind is that the collector current is Beta times the base current only in the linear region. When the transistor is saturated (i.e. the collector-emitter voltage is near zero), you can increase the base current all you want (up to the device maximum) and the collector current won't increase appreciably, if at all. If the piezo buzzer is spec'd to draw 10mA, you can drive the base with 1mA, 5mA or even 10mA of current and the collector current is still only going to be 10mA.

With regard to heat, the power dissipated by a transistor operating as a switch (cutoff or saturated) is very small. Consider, for example, the saturated region. Even at 100mA of collector current, if Vce(sat) is 0.4 volts, the power being dissipated is only 40mW. On the other hand, there is significant power dissipation as the transistor moves through the linear region as it goes to and from cutoff and saturation.

Fortunately, bipolar transistors switch pretty quickly and consequently they don't spend much time in the linear region. It is true, however, that overdriving the transistor does lengthen the switching time somewhat but it may not have a significant effect. If you're only occasionally switching the transistor, power dissipation isn't much concern at all. If you're switching it at medium to high frequencies, power dissipation is more of a concern.

[twesthoff]

JC,
Although your calculations are correct, they are not applied to the problem correctly. Usually, you start with the Load (piezo) current and work backwards to the base resistor, selecting the value of resistor that properly drives the base for the load you are using. The 2N3904 with a 10K base resistor will work fine. If the transistor is in the passenger compartment heat will not be a problem in this applicaton.
I may have missed it, but the exact specs for the piezo buzzer have not been given, so the exact calculations can not be made. Only educated guesses based on experience with similar devices.
Tom

[spamiam]

Well, the piezo buzzer does not have a coil in it as far as I know. The current draw is rated at 10mA.

Don said that driving the transistor to 2X that should be fine. I was planning on 4X just to be sure I was pretty well saturated.

It is not being used in high frequency switching. On for a few tenths of a second, off for a few. Essentially some morse code beeps.

It is going to be in the passenger compartment, but, as you know, they can get pretty hot or cold when no one is in it. The electronics COULD be running at that time.

If a jump start were ever needed, it probably would be turned off, can't guarantee that, though. The 27V TVS might get a real workout if the car is jumped with 24V. The switching power supply is rated for something like 40v input, so except for transients, 24v would be no problem.

Thanks for the input. I will be using an SOT-23 device, and I saw a MMTA06 (I think that was the #) that looked pretty good and reasonably cheap in small quantities. 80V, Hfe of 100+ as I remember.

-Tony

[JC]

Wow. Tough audience!
I agree with all of the comments made above.
Spamian stated that he has blow up lots of transistors and wanted some guidance, presumably both for this specific application, and in general.
My goal was to give him some general thoughts on things to consider to help keep from frying more transistors.

I think we all agree a 2N3904 can switch a non-inductive small collector current load without difficulty, but if he is burning up transistors, then somewhere along the line a mistake is being made, and one of our current assumptions is invalid!!!
I agree it is difficult to design the circuit without specific load specs. But rather than say "Can't do", and move on, some suggestions can clearly be provided.
The schematic was draw to confirm the actual circuit being used. The 10 k diode and pull down resistor descriptions justify confirming the basic circuit.
The pull down resistor could be anywhere, if it was place in the emitter lead it is clearly not indicated, and is detrimental to this circuit's purpose. If it is tied to the transistor's base, it also isn't needed, and makes calculating the base current, when driven by a voltage divider, more involved than the simplistic model calculations above. If it is tied to the ZX port, it again is unnecessary, but is basically immaterial.

The reference to past transistor episodes make me want to include the inductive load reverse EMF diode, and comment. There are many small electro-mechanical buzzers on the market, and they require protecting the driver transistor. Additionally, perhaps in the past his load was a relay, or other inductive source. Spamian has himself asked about these diodes, on the ZX forum, previously, when discussing "Noise filtering in Automotive environment". Nobody, myself include, actually drew him a schematic and illustrated exactly where to put the diode, its orientation, and a suggested part number at that time. Hence, this time I did.

It is difficult to know if the transistor will be in linear or saturation mode when no spec's are given for the load. If he has had problems, another possible cause is that he is not running the transistor in saturation mode. This could quickly give rise to either instant or early failure. Only if it is in saturation mode will the load be the current limiting component. The calculations emphasize that the 2N2222 has a significantly higher load capacity than the 2N3904.

If either this, or any of his prior circuits have been in linear mode, then a discussion of heat is warranted. Many designs look great on paper, and may even work in an air conditioned lab on a protoboard, only to fail prematurely when enclosed in a small, non-ventilated case, in a hot environment. The passenger compartment is much better than sitting on top of the engine, but it is still a hostile environment compared to an indoor setting. In Ohio, certainly not the extremes in the US or elsewhere, my passenger compartment routinely goes over 100 'F, sometimes over 115 'F in the Summer.

Our collective (ZX Forum) goal, given Spamian's inquiry, ought to be to provide him with a bullet-proof design that will prevent him from "blowing up" more transistors.

If his original inquiry, and my response, encouraged several people to re-think this common task, that is a good thing.

Tony, good luck with your project.
JC

[spamiam]

All the input is appreciated. I have not been lblowing transistors for THIS project, but I have before. On some of them it may have been due to lack of a pull down between the base and ground.

I do plan on using a pull down on the base beause the power will be supplied to the transistor before the ZX is fully booted. Also, I do not want to worry about peizo activity if the ZX pin is set to input tristate.

Is it bad to have the pulldown on the base? I think I could use MUCH higher resistance than 10K. I would expect the 100K would be sufficient. Why not use the pull down, and how high can I go before it is too high a resistance to be effective?

-Tony

[JC]

Doing a controlled power up can be a non-trivial task. This could generate a much longer thread than your initial question!

For this application, an audio alert in your car, is its state critical during the first << 1 mSec of initial power up? If one of the first items your program does is define the I/O pins, and their initial states, will you even hear your piezo chirp? Do you intend to chirp it on power up to indicate to the user that it is present and (presumed) functional? (Equivalent to the dash panel briefly lighting all of the warning lights on power up.)

That said, you can use a high value pull down.
JC

[spamiam]

Well, one of the first things that happens as my programming powers up is to set the I/O pins. I make no assumptions about their states as the program starts.

I am sure that a 1ms chirp would not be noticeable.

when all the intializations is complete, including communication with a local and a remote LCD, then there is a triple short beep of the buzzer.

Lack of communication with the remote display (a legal condition) is signaled by 4 long beeps.

Any worse than that will not result in any sounds, but no displays light up.

However, I have not simulated failures of hardware subsystems such as RTC (other than lost backup clock power), certain external logic chips, etc. For the most part failures of subsystems will either make it totally non-functional (power supply, RTC time base), or it will only lose certain functions, like switch senses.

I will change the specifications to use a 100K pull down, and it will be an optional part. I do not HAVE to solder it in! The cost of the part is only about 6 cents in small quantities. Soldering the 0603 surface mount part is "free" in the cost for my time and solder.

-Tony