Interpreting Datasheets

[shekhar_dandya]

Hi Everyone,

The electrical characterstics of common components (like diodes,transistors,opamps,regulator chips etc etc...)
are mentioned at perticular test conditions of voltages and currents.Our circuit may not have these test conditions.
Then how do we decide the value of a perticular parameter based on conditions of our circuits?
Is it that we can safely rely on values in "Typ" coloumn for our circuit or there is any other way of selecting.
Help needed.

Thanks
-shekhar

 

[ante1]

Hi shekhar,

You always have to choose your components on the

 

[Kevin Weddle]

That is an interesting subject. It can be desireable to bias a component outside the datasheet. My gripe is with the opamp parameters at high frequencies. Why operate at such a high frequency just because you can. I think it's too nonlinear. It's not great that you want to attenuate the spectrum by a mixed amount. Sending the high frequencies to some realm without accountability. Tossing the frequencies to whatever gain. 20db per decade. Whatever!

 

[audioguru2]

Hi Guys,
I almost always design for "worst case", and use minimum spec's so that my designs always work with any part that meets the spec's minimum which is guaranteed.
I was going to provide a sample circuit with a BC547 transistor that has a DC current gain from 210 to 900 and ask how you would bias it. But I have a better example.

Let's say you want to make an oscillator/timer with the CD4541. You need a supply voltage of 5V, and you need to choose the divider ratio and clock frequency.
Its typical clock oscillator frequency is 2.5MHz, but some (which might be many) can't oscillate beyond only 1MHZ. It is a big gamble to design it for 2.5MHz, some will work and some won't. If you make just one circuit, how many ICs are you going to try before you find one that works at 2.5MHz? They are all guaranteed to work at 1MHz. Your choice.

View attachment 36351

 

[shekhar_dandya]

Hi All,
Actually I was trying to make a small signal voltage amplifier from descrete transistors for audio frequency range.I used a small signal h-arameter equivalent circuit of the transistor.I decided to use a BC107 for it.Then I used the formulae from my electronics textbook to calculate the current and voltage gains.these were:

Av=-hFE(Rc||RL)/hie
Ai=-hFE*Rc*Rb/((Rc+RL)(Rb+hie))
Ri=hie||R1||r2
Ro=Rc||RL
I used a voltage divider bias to bias the Q point.(Vcc=10V,Vc=Vcc/2,Ic=5mA)
Rb=R1 || R2 (R1,R2=Resistors of voltage divider)

From the datasheet of BC107,I looked for the hFE v/s Ic curve to find that beta of X'tor was 250 @Ic=5mA.

The 'electrical characterstics' table did not specify h parameters at these conditions.

Now that the X'tor I use has a beta of 100.Also the Gains mentioned above depend on h parameters.Do these parameters differ from X'tor to X'tor? ie peice to peice? Can it be that the datasheets specify value of hie as 4Kohm but the X'tor piece you use has a hie about half of that?In that case above voltage gain would double.
How does one calculate amplifier parameters(Ai,Av,Ri,Ro)to meet your specifications?Are there any other formulae to calculate these parameters to see the circuit meeting your design conditions?
How much reliable it is to rely on the small signal equivalent model of the X'tor to design an Amplifier?How do you design one to meet the necessary conditions?


Kevin, do you want to say that assume the test conditions specified in datasheets in your circuit and then calculate the remaining parameters?I find this inflexible.

more help needed.

-Thanx
shekhar

 

[audioguru2]

Hi Shekhar,
The spec's for most transistors vary up to 8 times from guaranteed minimum to maximum. You can buy them with the current gain selected with a narrower range if you pay more.
You can't even bias the base of a transistor at 0.7V with a voltage divider, because one transistor will saturate with 0.7V while another of the same part number will be cutoff. The biasing will also change a lot with temperature changes.
If you use a transistor as a high gain voltage amplifier with its emitter grounded, its output will be extremely distorted.

The solution to this mess is to use negative feedback. An emitter resistor does wonders for biasing, temperature and distortion problems, although reduces the gain. Biasing a transistor from the collector instead of the supply also helps a lot, but reduces input impedance.
With negative feedback, the different h-parameters from piece to piece don't matter much. In fact, the h-parameters don't matter anymore with a lot of negative feedback, so voltage gain becomes simply Rc divided by Re. Bandwidth is increased a lot too.

IC's use emitter resistors to help biasing and temperature problems and a high impedance current-source instead of a collector resistor for high gain and linearity, since the current in the transistor doesn't change.

 

[shekhar_dandya]

Hi Audioguru,
Merry Christmas.

I have put an emitter resistor to ground and also biased the base of
x'tor from collector.This improoves dc stability.But for ac
conditions,emitter is bypassed to ground by an emitter capacitor to
enhance ac voltage gain.This would again make the gain formulae
dependant on h parameters.

 

[audioguru2]

Hi Shekhar,
Thanks. Merry Christmas to you too!

If you insist on cap-bypassing the emitter resistor of your transistor for a high gain, circuits using it will have variations in gain and frequency response due to the wide range of each individual transistor's h-parameters and also will have high distortion, reduced bandwidth and a fairly low input impedance.
I suggest using emitter resistors on two lower-gain transistors that are cascaded for high predictable gain, wide bandwidth and high input impedance. Since the effect of variable h-parameters won't make much difference due to the negative feedback provided by the emitter resistors, transistor substitutions when required also won't make much difference.

A constant current source instead of a collector resistor will offer a high gain even with an emitter resistor and also low distortion, at the expense of reduced bandwidth and a very high output impedance.

 

[surajbarkale]

Check out
Practicle Transistor Circuit Design & Analysis
By - Willems E. Gerald
(PS - I am a looser when it comes to spelling ;D )

If you can lay your hands on it. It is the best book to learn transistor design from. Of cource Basic Electronics by Malvino is also a good book but he explains design in very short manner. Better refer to it after you have read the first book.

 

[shekhar_dandya]

Hi all
A Happy New Year

Thanks Suraj for the book.I will definitely try to find it in my library.
AudioGuru,
Thanks for such a valid suggestion.