current sensing a pow/supply stage!

[kkrafs]

Is it normal for a "classic" high side current sensor to
hit the roof and stay latched if the current sensor
output are shorted to GND?

(the design is just as the one in the LM158/358
(data sheet page 8, Current monitor to the right).

kkrafs

 

[bcarso]

Need more info. Whose datasheet? What is/are the rail voltage(s)? What's being shorted? What is the highside voltage you are sensing current near?

If it is the circuit on page 13 of the National Semiconductor datasheet, with a transistor and the voltage proportional to current being provided at the transistor emitter, then you will pull momentarily a large current if the output is shorted, but the op amp output should go to about Vbe above ground after that. However this may not be a stable state from the standpoint of oscillations and during the swings may well then destroy the transistor.

If you are exceeding the common-mode range of the 358 then all bets are off anyway (note the restriction to no more positive than 2V below the op amp V+).

 

[kkrafs]

Im sorry Becarso, i wrote completely wrong, im not shorting
the current sensor output, im shorting the regulator output!
R1 right hand side, supply to load point.

Geeee i can really screw things up!

>Need more info. Whose datasheet? What is/are the rail voltage(s)? >What's being shorted? What is the highside voltage you are sensing >current near?

National,30V,the regulators output, (im doing a uC controlled supply,
NPN pass, Op amp regulator loop) Voltages are between 0 to 30V DC.

>If it is the circuit on page 13 of the National Semiconductor datasheet, >with a transistor and the voltage proportional to current being provided >at the transistor emitter, then you will pull momentarily a large current
>if the output is shorted,

Wich output do you mean, the regulator
or the op amp current sensor circuit?

>but the op amp output should go to about Vbe above ground after >that. However this may not be a stable state from the standpoint of >oscillations and during the swings may well then destroy the transistor.

The sensor circuit are never shorted, the op amp output blasts to
+V ie, 28V or 32V (havent decided yet wich supply) but emitter has
1k ohm to GND so current are +V -Vce about 30mA +Vbe ies nearly
+30V on base! perhaps not so good, whatif i put in a ohm and a zener between opamp output and transistor base? The accuracy is not a very
important matter i have a lookup table in the MCU to map the current values anyway im just after a rough voltage representing the current.

This data sheet:
http://www.national.com/pf/LM/LM358.html
page 13 right most cirquit.

>If you are exceeding the common-mode range of the 358 then all bets >are off anyway (note the restriction to no more positive than 2V below >the op amp V+).

There is a note in the datat sheet that says:
:Note 7: The input common-mode voltage of either input signal voltage
:should not be allowed to go negative by more than 0.3V (at 25C). The
:upper end of the common-mode voltage range is Va b1.5V (at 25C),
:but either or both inputs can go to a 32V without damage, independent
f the magnitude of Va.

Wich sugests this old fart is one of these "top over supply" Opamps! No?
And if yes then i could supply the opamp with +5V and have my +30V
applied to it as common mode range to sense the current!

Is the differential current sensing a better approach then this?

Kkrafs

 

[bcarso]

"Wich sugests this old fart is one of these "top over supply" Opamps! No?
And if yes then i could supply the opamp with +5V and have my +30V
applied to it as common mode range to sense the current! "

No no no. The Note 7 is irrelevant to what you are doing---it applies to input voltages below the "ground" of the op amp (pin 4).

I'm talking about voltages close to the +V pin of the op amp. These cannot be too close to that pin or the op amp doesn't work properly. The app circuit doesn't say what the op amp's supply voltage is, just that the input to the current sensor has to be at least abut 1.5 to 2 V less than this.

So you can't sense current unless the op amp V+ is 1.5 or 2V more positive than where the current comes from. So, if your regulator is, say, an 18V one, you could sense the output current with the National circuit if the op amp is fed from the unregulated voltage and that voltage is about 20V or more.

But if you are sensing the current into the regulator, you have to provide a still higher voltage for the op amp for the circuit to work. It sounds like that is what you are doing if you are dealing with ~30V.

"The sensor circuit are never shorted, the op amp output blasts to
+V ie, 28V or 32V (havent decided yet wich supply) but emitter has
1k ohm to GND so current are +V -Vce about 30mA +Vbe ies nearly
+30V on base! perhaps not so good, whatif i put in a ohm and a zener between opamp output and transistor base? The accuracy is not a very
important matter i have a lookup table in the MCU to map the current values anyway im just after a rough voltage representing the current. "

No. That is not how the circuit works (when it is set up properly). The inputs of the op amp when the loop is able to close are at the same voltage. At zero load current, that voltage is the same as the input voltage that is the source of the current you are sensing. There is no current in the transistor and the base is at about zero.

When you now start to deliver current to your load, the op amp turns the transistor on---so that the collector current pulls the 100 ohm R/n.i. input downwards until it is at a voltage equal to the 0.1 ohm resistor/inv input junction voltage. Thus everything balances. Since there is a voltage gain from the base to the collector of 0.1, and the sense resistor is 0.1 ohm, the emitter voltage is numerically equal to the load current in amperes.

(this all assumes no base current, which is a good approximation at the ~1% accuracy level)

So---bottom line: feed the op amp from a voltage higher by 2 volts than the left hand side of the current sensing resistor and the circuit will work.

 

[Svart]

:? Bcarso beat me to the punch again...

 

[PRR]

> Wich sugests this old fart is one of these "top over supply" Opamps!

Look at the schematic bottom of page 20.

The inputs are two PNP Darlingtons under a 6uA current source. At a glance, we know that if the input puns rise closer than 1.2V to the V+ pin, a Darlington will shut off, it will fail to operate as expected. In fact the current source needs voltage to work, hence the V+ minus 1.5V or 2V Input CM Range spec in Electrical Characteristics. (FWIW, you can take inputs a few tenths of a volt negative of V-, and it will work.)

The +32V spec in Note 7 says it should not be damaged, but again the functional (works like an opamp) range is V+ minus ~1.5V.

Therefore, the suggested circuit "Current Monitor" will not work unless the opamp is fed at least 2V more than the voltage at "VL".

It may work better to sense the low-side current.

You could rig a 4-resistor differential amp to sense a high-side resistor far above the chip supply.

There are automotive current monitors intended to sense a high-side that may be higher than the chip supply voltage.

 

[CJ]

What are we sensing?
AC?
DC?
How much current?

 

[kkrafs]

This has turned into the completely wrong direction!!
Im not arguing against what you folks say, im not argueing
about what happends at +V im interested what happends
at GND!! If we rewind to the start i asked why the circuit blast's
its output to +V -2V if load side shorted to GND.
(minus input on opamp).

Im sorry if i contributed to the goofing up of my
own question but i had previously been awake
for the latest 20 hours!!. :?

And no i dont use stuff that make you awake for longer periods!
I dont smoke the brass i eat it! :green: ......Mooouuu!! :grin:

Small experiment:
If minus input on Opamp shorted it act as
a comparator and trows output to +V -2V. :thumb:

kkrafs

------------------------------------------------------------------------------


[quote author="PRR"]> Wich sugests this old fart is one of these "top over supply" Opamps!

Look at the schematic bottom of page 20.

The inputs are two PNP Darlingtons under a 6uA current source. At a glance, we know that if the input puns rise closer than 1.2V to the V+ pin, a Darlington will shut off, it will fail to operate as expected. In fact the current source needs voltage to work, hence the V+ minus 1.5V or 2V Input CM Range spec in Electrical Characteristics. (FWIW, you can take inputs a few tenths of a volt negative of V-, and it will work.)

The +32V spec in Note 7 says it should not be damaged, but again the functional (works like an opamp) range is V+ minus ~1.5V.

Therefore, the suggested circuit "Current Monitor" will not work unless the opamp is fed at least 2V more than the voltage at "VL".

It may work better to sense the low-side current.

You could rig a 4-resistor differential amp to sense a high-side resistor far above the chip supply.

There are automotive current monitors intended to sense a high-side that may be higher than the chip supply voltage.[/quote]

 

[CJ]

threads always meander.
started out pretty vague, and got worse.
what are we trying to do?
audio related?

 

[kkrafs]

[quote author="CJ"]threads always meander.
started out pretty vague, and got worse.
what are we trying to do?
audio related?[/quote]

Well, trying to sense the current trough a shunt resistor
converting this to a voltage that can be sampled by a
uC ADC as to create a variable current limiting function
in a programmable PSU who are goin to be used to
feed power to audio circutry! :grin:

Here is an artickle that point to interesting subjects:
http://www.edn.com/article/CA402131.html

I just love hidden features in opamps! :green:

Kkrafs

 

[bcarso]

Last attempt.

If I understand you, you are shorting the output of the power supply to ground, and asking if the current sensor latches high when you do that.

The answer is probably YES, if the op amp is able to swing high enough to forward bias the base-collector junction of the transistor---thus making a negative feedback loop positive. And it will be if there is current limiting already upstream of the 0.1 ohm shunt resistor, and the op amp supply is not collapsing with the upstream voltage. If however the short pulls everything down then it's a hard call as to what happens.

If you want to prevent the latchup behavior it should work to put a diode in series with the transistor collector.

I would advise against using the uC in a loop with the sensor and ADC. It will likely not be stable (the system will oscillate) without a lot of work.

 

[CJ]

We sell a hall effect dc sensor. the split core model snaps over the dc line, no shunt resistor needed.
totally isolated.
puts out either 4-20 ma or 0-5, 0 to 10 volts dc, depending on model.

powered by either 24 ac/dc or 120 ac.
don't order any becuase they are a pain in the .....

I mean, wonderful product! Buy many!

but usually used in high current apps, 5 amps full scale and up.

(mag field is weak, noise is not)

i could probably pull one out of the bone pile and tweak it for low amps, but i.........hmmmm...............


.
.
.
.
got any cool iron laying around?

:razz:

 

[kkrafs]

[quote author="bcarso"]Last attempt.[/quote]
Dearest bcarso i just love you (on a technical level that is)
for putting in all this wrestling with my wierd questions :grin: :thumb:

If I understand you, you are shorting the
output of the power supply to ground,

Yes! :grin:

and asking if the current sensor latches high when you do that.

No! :grin:

:The answer is probably YES, if the op amp is able to swing high enough
: to forward bias the base-collector junction of the transistor

It is! :grin:

:thus making a negative feedback loop positive.

Yes it does. :grin:

:And it will be if there is
:current limiting already upstream of the 0.1 ohm shunt resistor, and the
p amp supply is not collapsing with the upstream voltage.

The power supply is not collapsing, the regulator loop
including current sens circuit have its own supply chain.

:If however the short pulls everything down then it's ahard call as
:to what happens. If you want to prevent the latchup behavior it
:should work to put a diode in series with the transistor collector.

I dont want to protect it from the latchchup i want it! :grin:
I just wonder if its normal behavior from this type of cirquit
with a LM158 as sensor amp! :grin:

:I would advise against using the uC in a loop with the sensor and ADC.
:It will likely not be stable (the system will oscillate) without a lot of work.

I just did some basic testing works allright, no spurious oscillation
during switched loads from 20mA to 1A. i have implemented some routines that deals with heavy switching of loads and some oscillations.

Kkrafs

 

[kkrafs]

[quote author="CJ"]We sell a hall effect dc sensor. the split core model snaps over the dc line, no shunt resistor needed.
totally isolated.
puts out either 4-20 ma or 0-5, 0 to 10 volts dc, depending on model.

I had some heavy toughts about that one, there is some new hall based
current sensors IC's out there but i asume they are a bit on the
expencive side. :green:

kkrafs

 

[kkrafs]

[quote author="CJ"]What are we sensing?
AC?
DC?
How much current?[/quote]

Did you relize that you just said ACDC?

http://www.motorhead.ru/int13ec.htm

A rater insane looking lemmy! :green:

kkrafs