EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Ditiris on November 24, 2024, 04:36:03 am
-
Can someone help me build a BJT current sense amplifier? Specs would be 10-16V common mode, 1.2uV to up, as I plan on running a 50A circuit across 0.1mR sense resistor. So, it would be nice to be able to adjust from 600 gain to 100 or possibly less. I have tried to do this with a Wilson and Widlar current mirror but I can't get the lower two octaves or so. The reason I want to build this discretely is primarily cost - I can't afford very expensive amplifiers that provide the kind of response I want. I would like to be able to clamp/turn off my NMOSFETs in <100 ns with this circuit. Something like the INA199 has too much delay, and more responsive op-amps are too expensive. Any help is much appreciated, thank you!
-
you should post your circuit so far because there is no need to start from scratch on this.
making a basic discrete op-amp is like a college class or two, and thats for one that works in a simulator where you can adjust the transistor specs.
And I noticed you picked a zero drift amplifier. Their not known for the speed.
You should post which amplifier you found that WORKS. then you have a spec.I am looking at current sense amp prices and I don't think your going to get too much of a 'value' even if you make a complex discrete circuit lol
What price point are you trying to beat, exactly?
-
also what the hell is lower octave? i dont see anything about frequency
there is alot of literature about making mirrors have faster AC response. they are deep
-
you should post your circuit so far
Please see attached for the circuit taken from an academic paper. You can see the low-end is not linear and is an issue.
You should post which amplifier you found that WORKS
No amplifier works. The response is either accurate and too slow like the INA199, or fast enough but inaccurate/too expensive, like the LT1194, or some undesirable combination thereof. That's why I'd like to use the BJT amplifier. I may have to take the approach of using a circuit like the one attached for the threshold current to turn off the NMOSFET(s) and an amplifier like the INA199 for the accurate current measurement. Cost is an issue as I need 20-40 current measurement circuits for what is supposed to be a fairly cheap board.
also what the hell is lower octave? i dont see anything about frequency
This is something I lifted from the academic paper - they refer to decades and octaves for the higher currents where the voltage response is linear - in particular the upper 2.5 decades.
there is alot of literature about making mirrors have faster AC response. they are deep
There is no issue making the BJTs respond quickly - there is an issue with the lower end of the response as you can see from the attached plot. The authors of course saw this as well, but did not care about the lower end of the response. I need somewhat accurate +/-5% maximum, ideally more like +/-1% at the lower currents as well.
-
Forgot to say I fixed Vmeas in the circuit so ignore that. The load runs from 0 to 100A. I would set the gain to either be Vf of a BJT or V_h_min of a CPLD to turn off the NMOSFET.
-
Also, most of these circuits are 10-16V via the NMOSFET, but some will be 48V. Currents will range from 5A to 50A+. So I really need to be able to set the gain via resistors.
-
you said 100ns that is 700us its 5000 times too slow ?
I assure you getting 100ns good response from 8 transistors is going to be diabolically hard
-
also, try to make that circuit in real life, not in the simulator, and see if you get results anywhere near that, to see if its even remotely viable.
-
I see - the step response is terrible.
What would you suggest for a suitable current sense amplifier topology?
-
Unless your trying to beat a $40 part, nothing really makes sense to develop
especially I am thinking this has to do with 50 amps. thats alot of power. some avant garde amplifier might not be the thing you wanna trust to determine anything about 50 amps
making a discrete amplifier that will be stable with parts variations in transistors might be way more then you bargained for, if its more then like a few.
-
I am not sure that kind of dynamic range is compatible with faster than 100 nanosecond response. At the least, gain will need to be distributed through multiple stages, with the fast bipolar level shifter operating at a lower gain, followed by a current feedback amplifier.
Parasitic capacitance will limit resistances to lower values. 300 kilohms is not going to work.
The bipolar stage PNPs can be stabilized for good precision with a slow operational amplifier, but I do not have any examples to link.
-
adding a real op amp makes it much more realistic to accomplish but I still think that if this is price related, its going to be alot of effort to get minute savings.
I am kind of thinking that this amplifier could make a bit of a stir, there is alot of modern devices that want cheaper way to measure current.
-
I hope you realize that cheap transistor pairs have no better than 2mV Vbe offset specs, if you take care to keep them at equal temperature, power dissipation, collector voltage, etc.
50A through 0.1mΩ is 5mV. And you want it to be accurate over a few decades of current?
Also, are you sure you need amplifiers accurate over a range of currents and not each one amplifier accurate at one particular current? Maybe different shunt resistors for different current thresholds?