PNP bias/voltage divider

[James]

Hi,

I'm sure this is a repeat question, but searching hasn't revealed an
answer to my specific issue. If there's a thread that addresses this
already, please direct me to it. I'm not an EE, so forgive me if the
answer is simple


I have chosen a BC327 to use in a low power switching regulator and I'm
trying to bias the base for -0.7V and 100mA of collector current. The
emitter voltage will be anywhere from 6.5-8.9V. Assuming an Hfe of 150,
the base current should be about 660uA. I'm calculating my voltage
divider using 6.5V, anything more should only 'overbias' the base. I've
calculated the R2 resistance to be about 10K ((6.5-0.7)/.00066 =
9.66K). I used this to calculate R1 to divide 6.5V into 5.8V and 0.7V.
This came out to be about 90K (6.5*9666/0.7 = 89.755K). Is this the
correct way to apply the math in the circuit below if I want base
voltage of -.7V at 660uA?

6.5-8.9V
O----------+--------------+
| |
\ |
/ R1 |
\ 90k |
/ |
\ |
| /
| <-660uA |V_
+-----------| BC327
| |\
\ \
/ R2 |
\ 9K |
/ | 100mA ->
\ +-----------
|
|
|
_____
___
_

 

[Tam/WB2TT]

Hi,

I'm sure this is a repeat question, but searching hasn't revealed an
answer to my specific issue. If there's a thread that addresses this
already, please direct me to it. I'm not an EE, so forgive me if the
answer is simple


I have chosen a BC327 to use in a low power switching regulator and I'm
trying to bias the base for -0.7V and 100mA of collector current. The
emitter voltage will be anywhere from 6.5-8.9V. Assuming an Hfe of 150,
the base current should be about 660uA. I'm calculating my voltage
divider using 6.5V, anything more should only 'overbias' the base. I've
calculated the R2 resistance to be about 10K ((6.5-0.7)/.00066 =
9.66K). I used this to calculate R1 to divide 6.5V into 5.8V and 0.7V.
This came out to be about 90K (6.5*9666/0.7 = 89.755K). Is this the
correct way to apply the math in the circuit below if I want base
voltage of -.7V at 660uA?

6.5-8.9V
O----------+--------------+
| |
\ |
/ R1 |
\ 90k |
/ |
\ |
| /
| <-660uA |V_
+-----------| BC327
| |\
\ \
/ R2 |
\ 9K |
/ | 100mA ->
\ +-----------
|
|
|
_____
___
_

I have problems vieving this, The view that looks the best appears to show
the emitter of the transistor connected to the + supply. If that's wrong,
you have other problems and ignore the first two paragraphs.

You have some basic problems here. First, having a base voltage of -.7V
(wrong number) does not mean an absolute -.7V, but rather -.7 with respect
to the emitter. So, if the emitter was at +6.7V, the base would have to be
at +6.0. Your resistor values are interchanged.

You don't bias a bipolar trtansistor that way. At 100 ma, the base-emitter
voltage is going to be some number between about .60 and .68. Haven't looked
at your transistor, but you want it to work with all transistors having
betas between about 50 and 300. You could probably get it to work somewhat
by returning R2 to the collector of the transistor instead of ground. and
assuming the collector voltage is 1/2 of VCC. You would have to make R2 a
variable resistor, and hope the temperature does not change a whole lot.

If this is a switching regulator, why is there *any* bias on the transistor?

Tam

 

[Tom Bruhns]

Hi,

I'm sure this is a repeat question, but searching hasn't revealed an
answer to my specific issue. If there's a thread that addresses this
already, please direct me to it. I'm not an EE, so forgive me if the
answer is simple


I have chosen a BC327 to use in a low power switching regulator and I'm
trying to bias the base for -0.7V and 100mA of collector current. The
emitter voltage will be anywhere from 6.5-8.9V. Assuming an Hfe of 150,
the base current should be about 660uA. I'm calculating my voltage
divider using 6.5V, anything more should only 'overbias' the base. I've
calculated the R2 resistance to be about 10K ((6.5-0.7)/.00066 =
9.66K). I used this to calculate R1 to divide 6.5V into 5.8V and 0.7V.
This came out to be about 90K (6.5*9666/0.7 = 89.755K). Is this the
correct way to apply the math in the circuit below if I want base
voltage of -.7V at 660uA?

6.5-8.9V
O----------+--------------+
| |
\ |
/ R1 |
\ 90k |
/ |
\ |
| /
| <-660uA |V_
+-----------| BC327
| |\
\ \
/ R2 |
\ 9K |
/ | 100mA ->
\ +-----------
|
|
|
_____
___
_

To begin with, the base current is an exponentail function of the b-e
voltage, and in addition it's quite temperature dependent. That means,
for a particular base current, it's not a good idea to try to achieve a
particular b-e voltage. The nominal b-e voltage just lets you know the
base voltage above ground (that is, about 0.7V lower than the supply),
so you can find the value of R2 to give you the required base current.
Then R1 just insures that the base current goes to essentially zero
when you are NOT pulling R2's bottom end down to ground potential,
assuming you want to turn it off. R2 must supply the required
transistor base current AND whatever current is in R2 when it has about
0.7V across it.

Second, the transistor DC current gain (HFE) is a strong function of
the particular sample, and also of temperature. If you really want the
collector to source 100mA to some fairly close tolerance, there are
better ways to do it. If you just want the collector to pull up to the
supply voltage (less a small drop) and the maximum current will be
100mA, then you're OK. But it's usual to "overbias" the base, and go
for a ratio of base to collector current of 20:1 or even 10:1.

You didn't say anything about turning the transistor on and off, but if
you want to do that, there may be some other considerations: speed,
and how much voltage swing the thing that drives the on/off signal can
supply.

Cheers,
Tom

 

[Tom Bruhns]

I have problems vieving this, The view that looks the best appears to show
the emitter of the transistor connected to the + supply. If that's wrong,
you have other problems and ignore the first two paragraphs.

You have some basic problems here. First, having a base voltage of -.7V
(wrong number) does not mean an absolute -.7V, but rather -.7 with respect
to the emitter. So, if the emitter was at +6.7V, the base would have to be
at +6.0. Your resistor values are interchanged.

You don't bias a bipolar trtansistor that way. At 100 ma, the base-emitter
voltage is going to be some number between about .60 and .68. Haven't looked
at your transistor, but you want it to work with all transistors having
betas between about 50 and 300. You could probably get it to work somewhat
by returning R2 to the collector of the transistor instead of ground. and
assuming the collector voltage is 1/2 of VCC. You would have to make R2 a
variable resistor, and hope the temperature does not change a whole lot.

If this is a switching regulator, why is there *any* bias on the transistor?

Tam

Resistor values interchanged?? 6 volts across his 9k R2 value does
give about 667uA, and the current in R1 at 90k and 0.7V would be about
7uA, leaving 660uA for the base. There is some question then about how
fast it would turn off, if that's the desire, if you open the
connection to R2, and even some question about just what the purpose of
the transistor is, but looks to me like the calcs for the values are OK
with the given assumptions.

The assumptions, on the other hand, may be off-base. Should the
transistor saturate, or should it act like a 100mA current source?
Will it be turning on and off? Occasionally, or rapidly? Over what
temperature range must it operate?

Cheers,
Tom

 

[Tam/WB2TT]

Resistor values interchanged?? 6 volts across his 9k R2 value does
give about 667uA, and the current in R1 at 90k and 0.7V would be about
7uA, leaving 660uA for the base.

You are right. I guess it is a coincidence, but it struck out at me that if
you reversed the R values, you would get the correct base voltage if you
ignored the loadg. Still, current biasing is not the way to do it either.
I've been there, and had the factory scream bloody murder when the nominal
beta=40 transistors started arriving with betas of 385.

Tam

There is some question then about how

 

[James]

First of all, thanks for all the comments, it really helps out and I
appreciate everyone's time.

The regulator is modeled after an idea that I read about a few years
ago. That circuit could only source 20mA or so and required a higher
supply voltage, somewhere around 12V I believe. Using the same concept,
I've created a regulator that can be driven from a 9V battery (6.5V to
8.xV), is fairly efficient (95% or so simulated) that provides 5.49V
with <1.5% ripple. I'm trying to replace an LM7805 in a microprocessor
application that requires up to 85mA and do it on the cheap The
reason I don't want to use the linear regulator is that the application
is portable and the LM7805 is wasteful, especially with higher
voltages. Plus, the dropout is unacceptable for a 9V battery in my
application.

I am including a couple of links to both the entire schematic for my
regulator as well as a plot of the voltage at the Q3 collector-L1
junction.

http://www.jamescleland.com/content/5VReg/circuit.JPG
http://www.jamescleland.com/content/5VReg/emitter-L.JPG

Description:
R1 turns on Q1, which turns on Q3, causing voltage to build across L1.
Eventually D1, a 5.1V zener, breaks down and turns on Q2 which turns
OFF Q1 and, subsequently, Q3. The inductor rings through D1 and D2,
with C2 holding the output voltage at approximately 5.49V. Voltage
across D1 drops, Q2 turns off and the whole cycle starts over again.
Q3's switching frequency is about 50KHz.

I vary the load between 55 and 220 Ohms, simulating between 25 and
100mA load at 5.49V. So far, it's fairly stable, but I haven't
prototyped it yet.

 

[john jardine]

Hi,

I'm sure this is a repeat question, but searching hasn't revealed an
answer to my specific issue. If there's a thread that addresses this
already, please direct me to it. I'm not an EE, so forgive me if the
answer is simple


I have chosen a BC327 to use in a low power switching regulator and I'm
trying to bias the base for -0.7V and 100mA of collector current. The
emitter voltage will be anywhere from 6.5-8.9V. Assuming an Hfe of 150,
the base current should be about 660uA. I'm calculating my voltage
divider using 6.5V, anything more should only 'overbias' the base. I've
calculated the R2 resistance to be about 10K ((6.5-0.7)/.00066 =
9.66K). I used this to calculate R1 to divide 6.5V into 5.8V and 0.7V.
This came out to be about 90K (6.5*9666/0.7 = 89.755K). Is this the
correct way to apply the math in the circuit below if I want base
voltage of -.7V at 660uA?

6.5-8.9V
O----------+--------------+
| |
\ |
/ R1 |
\ 90k |
/ |
\ |
| /
| <-660uA |V_
+-----------| BC327
| |\
\ \
/ R2 |
\ 9K |
/ | 100mA ->
\ +-----------
|
|
|
_____
___
_

Transistors are not precise devices. You've a cat in hell's chance of
pinning down that 100ma.
It's usual to swamp out the Vbe voltage variations by adding some emitter
resistance. Changes in Vbe due to varying temperature and currents will now
be small compared to the larger voltage across the emitter resistor.
A la ...

6.5 to 8.9 ---o---------o
| |
.-. .-.
| | | |
| | | |4.7
'-' '-'
|220 |
| |6V
| 1ma |<
o-------| BC327
|5.3V |\
| |
5ma| |
| o
.-.
| | 100ma load
| | o
'-' |
| 1k |
0V---o---------o
(created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)

 

[John Popelish]

I am including a couple of links to both the entire schematic for my
regulator as well as a plot of the voltage at the Q3 collector-L1
junction.

http://www.jamescleland.com/content/5VReg/circuit.JPG
http://www.jamescleland.com/content/5VReg/emitter-L.JPG

Description:
R1 turns on Q1, which turns on Q3, causing voltage to build across L1.
Eventually D1, a 5.1V zener, breaks down and turns on Q2 which turns
OFF Q1 and, subsequently, Q3. The inductor rings through D1 and D2,
with C2 holding the output voltage at approximately 5.49V. Voltage
across D1 drops, Q2 turns off and the whole cycle starts over again.
Q3's switching frequency is about 50KHz.

I vary the load between 55 and 220 Ohms, simulating between 25 and
100mA load at 5.49V. So far, it's fairly stable, but I haven't
prototyped it yet.

First of all, the transistor will have a reduced gain as it
saturates. The lower you need the saturation drop to be,
the more base current you have to provide. So lets say you
provide a base current that is 1/50th of the collector
current, to drive a normally higher gain (say, beta 150)
device well into saturation, to keep the on state conduction
losses low. This would be 100 mA/50= 3mA.

Your R3 has to supply that current ans also the current that
bleeds through R2 while the transistor is on. The only
purpose for R2 is to drain the base stored charge away, to
quickly turn the transistor off when the current through R3
is cut off. I usually try to keep the on state R2 current
at least 1/10th of the on state base current. That would
put it around .3 mA. If the on state base emitter voltage
is about .7 volts, that means that R2 would have a
resistance of about .7/.3 mA = 2.3k. Lets say we use 2.2k,
a standard value. So R3 has to provide about 3.3 mA at the
lowest supply voltage minus the Q3 base emitter drop and the
Q1 collector to emitter drop. Lets say that is 6.5 - 0.7
-0.3 = 5.5 volts.

So R3 is about 5.5/3.3 mA =1.67 k. Lets say you use a 1.6k
5% standard value.

Unfortunately, at the fresh battery voltage, R3 supplies the
base of Q1 with more current than is needed, and this hurts
the efficiency of the supply a little. Replacing Q1 and R3
with a switchable current regulator would reduce this effect.

 

[Tom Bruhns]

First of all, thanks for all the comments, it really helps out and I
appreciate everyone's time.

The regulator is modeled after an idea that I read about a few years
ago. That circuit could only source 20mA or so and required a higher
supply voltage, somewhere around 12V I believe. Using the same concept,
I've created a regulator that can be driven from a 9V battery (6.5V to
8.xV), is fairly efficient (95% or so simulated) that provides 5.49V
with <1.5% ripple. I'm trying to replace an LM7805 in a microprocessor
application that requires up to 85mA and do it on the cheap The
reason I don't want to use the linear regulator is that the application
is portable and the LM7805 is wasteful, especially with higher
voltages. Plus, the dropout is unacceptable for a 9V battery in my
application.

I am including a couple of links to both the entire schematic for my
regulator as well as a plot of the voltage at the Q3 collector-L1
junction.

http://www.jamescleland.com/content/5VReg/circuit.JPG
http://www.jamescleland.com/content/5VReg/emitter-L.JPG

Description:
R1 turns on Q1, which turns on Q3, causing voltage to build across L1.
Eventually D1, a 5.1V zener, breaks down and turns on Q2 which turns
OFF Q1 and, subsequently, Q3. The inductor rings through D1 and D2,
with C2 holding the output voltage at approximately 5.49V. Voltage
across D1 drops, Q2 turns off and the whole cycle starts over again.
Q3's switching frequency is about 50KHz.

I vary the load between 55 and 220 Ohms, simulating between 25 and
100mA load at 5.49V. So far, it's fairly stable, but I haven't
prototyped it yet.

In addition to what John Popelish wrote in reply to this, I'd add that
it could help you in getting it to switch a bit more crisply to put a
large value of resistance between Q3's collector and Q1's base. I'd
suggest perhaps 100k ohms, or a bit more. Also, R5 can be eliminated;
it does practically nothing. There's already ample path from the base
of Q2 to ground through R4 and R6.

Cheers,
Tom

 

[Tam/WB2TT]

First of all, thanks for all the comments, it really helps out and I
appreciate everyone's time.

The regulator is modeled after an idea that I read about a few years
ago. That circuit could only source 20mA or so and required a higher
supply voltage, somewhere around 12V I believe. Using the same concept,
I've created a regulator that can be driven from a 9V battery (6.5V to
8.xV), is fairly efficient (95% or so simulated) that provides 5.49V
with <1.5% ripple. I'm trying to replace an LM7805 in a microprocessor
application that requires up to 85mA and do it on the cheap The
reason I don't want to use the linear regulator is that the application
is portable and the LM7805 is wasteful, especially with higher
voltages. Plus, the dropout is unacceptable for a 9V battery in my
application.

I am including a couple of links to both the entire schematic for my
regulator as well as a plot of the voltage at the Q3 collector-L1
junction.

http://www.jamescleland.com/content/5VReg/circuit.JPG
http://www.jamescleland.com/content/5VReg/emitter-L.JPG

So, the 9K does not go to ground as you showed, but is a pulse input. I have
two suggestions:

Look at replacing the PNP with a P channel FET. This is going to save a lot
of base current since you are concerned about power. You need a FET that is
fully ON at 6V. You don't have to limit the gate voltage to limit the drain
current; that is limited by the inductor and the switching frequency. What
you are proposing to build is called a buck regulator. You can buy chips to
do that from Digikey and Mouser, made by outfits like National, Maxim, and
TI.

Or, forget about the switching regulator and look at LOW DROPOUT series
regulators. The efficiency would be very close to V(out)/V(in). Probably
less efficient than the switcher with a new battery, and more efficient with
an old battery. The 7805 puts out 5V, and requires a minimum input of 7V;
7.5 on a bad day (This is not low dropout)..

I don't know what you mean by 9V battery, but if it is what I think, did you
ever calculate how long it would last with a 60 - 85 ma load?

Tam

 

[Vans]

Hi,

I'm sure this is a repeat question, but searching hasn't revealed an
answer to my specific issue. If there's a thread that addresses this
already, please direct me to it. I'm not an EE, so forgive me if the
answer is simple


I have chosen a BC327 to use in a low power switching regulator and I'm
trying to bias the base for -0.7V and 100mA of collector current. The
emitter voltage will be anywhere from 6.5-8.9V. Assuming an Hfe of 150,
the base current should be about 660uA. I'm calculating my voltage
divider using 6.5V, anything more should only 'overbias' the base. I've
calculated the R2 resistance to be about 10K ((6.5-0.7)/.00066 =
9.66K). I used this to calculate R1 to divide 6.5V into 5.8V and 0.7V.
This came out to be about 90K (6.5*9666/0.7 = 89.755K). Is this the
correct way to apply the math in the circuit below if I want base
voltage of -.7V at 660uA?

6.5-8.9V
O----------+--------------+
| |
\ |
/ R1 |
\ 90k |
/ |
\ |
| /
| <-660uA |V_
+-----------| BC327
| |\
\ \
/ R2 |
\ 9K |
/ | 100mA ->
\ +-----------
|
|
|
_____
___
_