How to design a current reference?

[IC_Lee]

The precision is less than 5% at any condition , including the variation
of process , power suply , temperature and ±20％ resistance.

Thanks a lot!

 

[IanP]

Do you mean current source which can be used as calibration/reference?
What is the current range : 4-20mA, 0-10A....??? and max voltage???

 

[rtpq]

Can you use external component like resistor? If so, you can use band-gap voltage (3~5% variation for all case) and a external resistor (1% variation).

 

[rtpq]

For implementation, you can use an opamp+current mirror to realize.

 

[cetc1525]

I think to use a high PSRR opamp&a cascode mirror.

 

[nickooeda]

10%-power supply,temp from -40-->120,Vth vary about 20%,proccess ......my advice is bandgap+current buffer

 

[Question]

ues bandgap cricuit and convert voltage to current

 

[IC_Lee]

Do you mean current source which can be used as calibration/reference?
What is the current range : 4-20mA, 0-10A....??? and max voltage???

Yes, the current source is use as reference.the current range is less than 1mA.
the power supply voltage is 2.5/3.3/5.

 

[IC_Lee]

Can you use external component like resistor? If so, you can use band-gap voltage (3~5% variation for all case) and a external resistor (1% variation).

I can't use external resistor ,so I should consider the ±20% variation of resistor and its temperature coefficient .

 

[IC_Lee]

ues bandgap cricuit and convert voltage to current

At present , I'm trying this way. But I meet some problem . Could you give me some detailed schematic?

 

[IC_Lee]

Thanks again for everyone .

 

[IC_Lee]

ues bandgap cricuit and convert voltage to current

At present , I'm trying the way. But I meet some problem . Could you give me some detailed schematic.

 

[IanP]

For 5V you can try the attached circuit.
For 2.5, ...3V...well..the only one solution that comes up is using simple FET transistor with Vgsoff close to 0V and a resistor connected between source and gate...this cicuit should work with voltages close to 1V++
I can draw it if you are interested...

 

[IC_Lee]

ues bandgap cricuit and convert voltage to current

At present , I'm trying the way. But I meet some problem . Could you give me some detailed schematic.

 

[IC_Lee]

For 5V you can try the attached circuit.
For 2.5, ...3V...well..the only one solution that comes up is using simple FET transistor with Vgsoff close to 0V and a resistor connected between source and gate...this cicuit should work with voltages close to 1V++
I can draw it if you are interested...

You're so kind .Thank you very much.
I had an urge to know it.

 

[IC_Lee]

For 5V you can try the attached circuit.
For 2.5, ...3V...well..the only one solution that comes up is using simple FET transistor with Vgsoff close to 0V and a resistor connected between source and gate...this cicuit should work with voltages close to 1V++
I can draw it if you are interested...

The pnp transistor in this circuit can't be implemented in CMOS process .

The current will change according to ±20％ variation of the resistance Re .

 

[IanP]

You can replace this N-P-N transistor with a N-Channel FET..
Attached is the other drawing..