current limit problems

[IanM]

hi,

I am limiting the current in a cct to ~60mA by using a 10ohm resistor in
the load current path to turn on a pnp transistor which turns off a mosfet.
It works but I cannot afford the voltage drop across the 10ohm resistor.
Normal max operating current is 30mA and the 300mV drop is causing problems.
Any solutions have to be upower as the cct spends most of its life drawing
about 24uA.

any ideas?

thanks

IanM

 

[D from BC]

hi,

I am limiting the current in a cct to ~60mA by using a 10ohm resistor in
the load current path to turn on a pnp transistor which turns off a mosfet.
It works but I cannot afford the voltage drop across the 10ohm resistor.
Normal max operating current is 30mA and the 300mV drop is causing problems.
Any solutions have to be upower as the cct spends most of its life drawing
about 24uA.

any ideas?

thanks

IanM

Maybe use one of the many various mosfet driver IC's..


D from BC

 

[me]

hi,

I am limiting the current in a cct to ~60mA by using a 10ohm resistor
in the load current path to turn on a pnp transistor which turns off a
mosfet. It works but I cannot afford the voltage drop across the 10ohm
resistor. Normal max operating current is 30mA and the 300mV drop is
causing problems. Any solutions have to be upower as the cct spends
most of its life drawing about 24uA.

any ideas?

thanks

IanM

The best bet would be to eliminate any conditions in the circuit that would
cause it draw over ~60 mA and do away with the current limiting. Without
more information it would be hard to give a better answer.

 

[Fred Bloggs]

I am limiting the current in a cct to ~60mA by using a 10ohm
resistor in the load current path to turn on a pnp transistor which
turns off a mosfet.

Is this a series pass MOSFET? If so then a current limit does not turn
it 'off', it should be reducing the gate drive to maintain 60mA.

It works but I cannot afford the voltage drop
across the 10ohm resistor. Normal max operating current is 30mA and
the 300mV drop is causing problems. Any solutions have to be upower
as the cct spends most of its life drawing about 24uA.

You obviously need to reduce the current sensing resistor and amplify
the voltage drop to a working magnitude. It will be easier to use an
amplifier with an input range that includes the positive rail, and input
offset will be a consideration if you are to avoid trimming.

 

[Jan Panteltje]

hi,

I am limiting the current in a cct to ~60mA by using a 10ohm resistor in
the load current path to turn on a pnp transistor which turns off a mosfet.
It works but I cannot afford the voltage drop across the 10ohm resistor.
Normal max operating current is 30mA and the 300mV drop is causing problems.
Any solutions have to be upower as the cct spends most of its life drawing
about 24uA.

any ideas?

use a 1 Ohm resistor, followed by a 10x CMOS opamp?
Or use a germanium transistor, it would trigger at about 150mV.

 

[Hammy]

Is this a series pass MOSFET? If so then a current limit does not turn
it 'off', it should be reducing the gate drive to maintain 60mA.


You obviously need to reduce the current sensing resistor and amplify
the voltage drop to a working magnitude. It will be easier to use an
amplifier with an input range that includes the positive rail, and input
offset will be a consideration if you are to avoid trimming.

Why dont you look at a HIGH-SIDE CURRENT MONITOR IC like the Zetex
ZXCT1009 (CHEAP and small SOT-23) you have an increased error with a
reduced sense voltage less then 100mV,but worth looking at.Sense your
current on the input of your regulator . You can easily rig a OVP,as
well as UVP.I'm using one for this purpose.

 

[Phil Allison]

"Jan Panteltje"

Or use a germanium transistor, it would trigger at about 150mV.

** I was frightened to even suggest that kind of " horse and buggy " idea.

Glad you did - but.





........ Phil

 

[Jan Panteltje]

"Jan Panteltje"



** I was frightened to even suggest that kind of " horse and buggy " idea.

Glad you did - but.

I used a ge transistor in something similar.
The Ic temp sensitivity (Icb leakage) caused it to shut the thing
down at about 75 degrees C, which was good, as it needed temp protection

 

[Phil Allison]

"Jan Panteltje"
"Phil Allison"

I used a ge transistor in something similar.
The Ic temp sensitivity (Icb leakage) caused it to shut the thing
down at about 75 degrees C, which was good, as it needed temp protection

** I know that is simply no joke - as the very famous 1970s " Phase
Linear" 700 watt and 400 watt audio power amplifiers use Ge devices as
output Ic sensors in their ( VI) current limiting circuits.

Types used were the TO5 pack 2N1304 and 2N1305.




........ Phil

 

[Winfield Hill]

I am limiting the current in a cct to ~60mA by using a
10ohm resistor in the load current path to turn on a pnp
transistor which turns off a mosfet.
It works but I cannot afford the voltage drop across the
10ohm resistor. Normal max operating current is 30mA and
the 300mV drop is causing problems. Any solutions have to
be upower as the cct spends most of its life drawing
about 24uA.

Ian, there's a simple scheme using two resistors that can
gain you a factor of two reduction in the sense-resistor
voltage drop. Namely, you add a resistor in series with
the PNP transistor's base, and tie another resistor from
the base to ground. This reduces the CL threshold roughly
by the resistor ratio times the input voltage. You should
confine yourself to a factor two because of the PNP's Vbe
tempco and part-part uncertainty. You also have consider
the PNP's base-current needs, as it shuts off the MOSFET.

Vs Rs 4.7 60mA CL, 140mV loss at 30mA
---+-----/\/\----+---, ,----------------
| | _|__|_
e 20k | ,--- Vbe - Vx - R1 Ic/beta
b -+- R1 --' | I_cl = ---------------------
c | | Rs
| chose R2 |
| | | where Vx = (Vs-Vbe) R1/(R1 + R2)
+---- | ----------'
| | 0.65 - 0.32 - 0.05
| GND I_cl = ------------------- = 60mA
0.25 mA 4.7

I chose R1 to drop 50mV, max, by assuming a 0.25 mA
MOSFET gate-drive at cutoff, and a minimum PNP beta
of 100. You chose R2, based on your supply voltage,
to develop I = 0.32/R1 = 16uA through R1 (is that low
enough micro-power?) to get a 320mV drop across it.
For example, for 15V you'd chose R1 = 875k.

There's a slightly more complicated scheme using five
resistors and a 2nd PNP transistor that allows you a
factor of three or four reduction. This scheme, which
I first saw used by Bob Widlar in one of his IC designs
35 years ago, obtains the reference voltage for the
CL-reduction from another PNP wired as a Vbe multiplier,
rather than using the power supply. This allows you
to cancel out the BJT's Vbe tempco and part-to-part
uncertainty issues. However, if you're going that far,
I'd consider one of the other suggestions made here
instead, such as the Zetex ZXCT1009 or a small opamp.

I'll bet if Tony Williams was here, he'd have something
interesting to say.

 

[Jamie]

hi,

I am limiting the current in a cct to ~60mA by using a 10ohm resistor in
the load current path to turn on a pnp transistor which turns off a mosfet.
It works but I cannot afford the voltage drop across the 10ohm resistor.
Normal max operating current is 30mA and the 300mV drop is causing problems.
Any solutions have to be upower as the cct spends most of its life drawing
about 24uA.

any ideas?

thanks

IanM

Hmm.
In the old days, using an incandescent bulb worked
as current increased, so didn't the R.
I suppose one could use a CLD (current limiting diode).
they are actually FETS back to back internally but maybe
a worth to look into.

In any case, it sounds like a good candidate for a signal
amplifier.

 

[Fred Bloggs]

I'll bet if Tony Williams was here, he'd have something
interesting to say.

Yes, without a doubt. It's impossible to digest the fact that we'll
never be hearing from him again.

 

[Winfield]

Yes, without a doubt. It's impossible to digest the fact
that we'll never be hearing from him again.

Another sobering fact to digest is that one of these
days it'll be us.

 

[IanM]

Ian, there's a simple scheme using two resistors that can
gain you a factor of two reduction in the sense-resistor
voltage drop. Namely, you add a resistor in series with
the PNP transistor's base, and tie another resistor from
the base to ground. This reduces the CL threshold roughly
by the resistor ratio times the input voltage. You should
confine yourself to a factor two because of the PNP's Vbe
tempco and part-part uncertainty. You also have consider
the PNP's base-current needs, as it shuts off the MOSFET.

Vs Rs 4.7 60mA CL, 140mV loss at 30mA
---+-----/\/\----+---, ,----------------
| | _|__|_
e 20k | ,--- Vbe - Vx - R1 Ic/beta
b -+- R1 --' | I_cl = ---------------------
c | | Rs
| chose R2 |
| | | where Vx = (Vs-Vbe) R1/(R1 + R2)
+---- | ----------'
| | 0.65 - 0.32 - 0.05
| GND I_cl = ------------------- = 60mA
0.25 mA 4.7

I chose R1 to drop 50mV, max, by assuming a 0.25 mA
MOSFET gate-drive at cutoff, and a minimum PNP beta
of 100. You chose R2, based on your supply voltage,
to develop I = 0.32/R1 = 16uA through R1 (is that low
enough micro-power?) to get a 320mV drop across it.
For example, for 15V you'd chose R1 = 875k.

I think I saw this in your book Win, AofE is at work so can't verify it at
the moment.
Thanks for this. It looks like a solution to me.

There's a slightly more complicated scheme using five
resistors and a 2nd PNP transistor that allows you a
factor of three or four reduction. This scheme, which
I first saw used by Bob Widlar in one of his IC designs
35 years ago, obtains the reference voltage for the
CL-reduction from another PNP wired as a Vbe multiplier,
rather than using the power supply. This allows you
to cancel out the BJT's Vbe tempco and part-to-part
uncertainty issues. However, if you're going that far,
I'd consider one of the other suggestions made here
instead, such as the Zetex ZXCT1009 or a small opamp.

I'll bet if Tony Williams was here, he'd have something
interesting to say.

I spent a good while researching the sci.electronics groups before I posted
this question and Tony Williams posts came up a lot.

 

[JosephKK]

Winfield [email protected] posted to sci.electronics.design:

Another sobering fact to digest is that one of these
days it'll be us.

So twit Horowitz and get the next edition of AoE out.

 

[Winfield Hill]

Winfield [email protected] posted to sci.electronics.design:



So twit Horowitz and get the next edition of AoE out.

Hey, we spent over half the days working on it so far...
Got any good capacitor-dieletric tempco data you'd like
to see us include in our plots? Other capacitor stuff?

 

[ChairmanOfTheBored]

Why dont you look at a HIGH-SIDE CURRENT MONITOR IC like the Zetex
ZXCT1009 (CHEAP and small SOT-23) you have an increased error with a
reduced sense voltage less then 100mV,but worth looking at.Sense your
current on the input of your regulator . You can easily rig a OVP,as
well as UVP.I'm using one for this purpose.

Zetex makes good stuff.

 

[ChairmanOfTheBored]

Another sobering fact to digest is that one of these
days it'll be us.

Everybody has to go sometime. After seeing enough deaths in one's
family, it almost becomes welcomed.

 

[John Larkin]

Hey, we spent over half the days working on it so far...
Got any good capacitor-dieletric tempco data you'd like
to see us include in our plots? Other capacitor stuff?

Are you going to venture into the great Bypassing/ PowerPlanes/
ReturnCurrent debate? I might have some interesting TDR stuff.

John

 

[JosephKK]

Winfield Hill [email protected] posted to sci.electronics.design:

Hey, we spent over half the days working on it so far...
Got any good capacitor-dieletric tempco data you'd like
to see us include in our plots? Other capacitor stuff?

I do not have any to hand any more. Maybe i can get some from
previous co-workers, that make not pan out or take some time however.
I was part of a test group space qualifying MLC capacitors about 20
years ago.