3 constant current source questions

[Jef Patat]

hi everybody,

in the past i build some audio amplifiers, mostly circuits of which i read some good reviews, but i never managed to understand them completely. Now i would like to construct an amplifier which i understand to the bone.

At the same time i would like everybody to be able to do so, even people quite new to the subject. I bought myself Douglas Self's Audio power amplifier design handbook and G. Randy Slone's High-power audio amplifier construciton manual. I have learned a great deal about amplifiers, but to a newbie those books aren't the easiest stuff. That's why i started to build a website to explain things to the bone, but of course only for one amplifier.

At the moment I'm stuck on the current source thing and i've got three questions. I refer to my website for schematics: http://users.pandora.be/jef_patat/au...t%20source.htm .
Now my questions are:
1. why is this design with two transistors better than the one with (zener)diodes (something G. Randy Slone says but doesn't really explain)
2. what is the purpose of R3
3. why is the capacitor connected between R2 and R4 and not above R2 so R2 and R4 could be one resistor?

I hope somebody can help me out with these small questions, many thanks in advance and kind regards, Jef

[jan.didden]

Originally posted by Jef Patat
[snip]I refer to my website for schematics: http://users.pandora.be/jef_patat/au...t%20source.htm .
Now my questions are:
1. why is this design with two transistors better than the one with (zener)diodes (something G. Randy Slone says but doesn't really explain)
2. what is the purpose of R3
3. why is the capacitor connected between R2 and R4 and not above R2 so R2 and R4 could be one resistor?[snip]

Hi Jefke,

Q1: Don't know. Do you have a schematic of a cs with the zener? Then we could talk.

Q2: No idea. Your schematic is an isolated cs that really doesn't do anything (except consume some current). In a practical application there may be a use for R3.

Q3: That would put the cap across the part that does the actual correction of the constant current, and would prevent the reg transistor to vary its Vce to keep the cs constant against noise and other quick variations. The way it is now, the cap keeps noise etc away from the reg transistor collector, where it might harm the constantness of the current, but still leaves the reg transistor to work to keep the current constant.

Jan Didden

[jaycee]

Originally posted by Jef Patat


1. why is this design with two transistors better than the one with (zener)diodes (something G. Randy Slone says but doesn't really explain)
2. what is the purpose of R3
3. why is the capacitor connected between R2 and R4 and not above R2 so R2 and R4 could be one resistor?

http://www.4qdtec.com/csm.html is an excellent site about current mirrors. The two-transistor version is merely more precise, also zeners tend to generate more noise than another transistor does.
Quite commonly LED's are used instead of a zener, but apparently these are effected by photovoltaic effects ?

At a guess, R3 is just to simulate a load.

In amplifiers, the most common types of current source you will see will be figure 3 and figure 4 on the 4QD site. More commonly 3, but higher-end stuff may use 4. Older stuff may use 2 where the diode is a reverse-biased Zener.


I belive the circuit formed between R2, R4 and C1 is an RCR filter to reduce power supply ripple.

[azira]

R3 represents the "load" that you're feeding current into. It's just being represented as a resistance. In real application it'll be some part of your amplifier such as the collector of an NPN transistor.
--
Danny

[Jef Patat]

thanks for the replies janneman and jaycee,
i made a little mistake in my post, strange no-one noticed it.
my second quest should be: what is the purpose of R5 and not R3
i know R3 simulates the load, i even changed it to prove how constant the current is.
so: what is the purpose of R5?

Q1: Don't know. Do you have a schematic of a cs with the zener? Then we could talk.

the url jaycee posted is a good example: why is circuit 4 better than 2 or 3?

and as to jaycee's post

The two-transistor version is merely more precise, also zeners tend to generate more noise than another transistor does.

why more precise, because of less noise? and why do zeners generate more noise, what kind of noise (temperature dependant, current, voltage, ...) is this and how large?

thx for the help !

[johnnyx]

"The Art of Electronics" by Horowitz and Hill explains this very well, and includes current mirrors, which is another "building block" of audio amps. This book is another "Must Have".

edit; spelling (of course!)

[Jef Patat]

thx for the reply johnnyx, now i've got a good idea for a newyears present. Of course it might take a couple of weeks up to a month before i have the book. Can't you just say why or is it to complicated?

thx for the suggestion, jef

[jaycee]

Originally posted by Jef Patat

why more precise, because of less noise? and why do zeners generate more noise, what kind of noise (temperature dependant, current, voltage, ...) is this and how large?

thx for the help !

The 4QD article does explain why its more accurate... the idea is both transistors work together to limit the current as well as the extra transistor serving as a reference.

As for zener noise, i belive it is affected by temperature but it is also to do with the way zeners are constructed. An LED is less problematic in this regard and is less affected by temperature changes also.

[thanh]

johnnyx! Is this book at internet?

[peranders]

Originally posted by Jef Patat
the url jaycee posted is a good example: why is circuit 4 better than 2 or 3?

Normally you want a "stiff" current source meaing high output impedance meaning constant current regardless of load. Noise is normally a number 2 priority.

Originally posted by Jef Patat
why more precise, because of less noise? and why do zeners generate more noise, what kind of noise (temperature dependant, current, voltage, ...) is this and how large?

Precise = higher output impedance.

A good way to learn is actually to simulate. I recommend that you download the free LTSpice from www.linerar.com and start fool around with this program. Just draw the circuit and start "measure" on it. Simulation can be very educational and you don't have to dig so deep into theoretics.