Max6126 current source circuit transistor question ?

[Overspeed]

Hello

Following my post on the 6665 I have check the Max6126 , in the technical doc they propose a current source circuit simple

Question : what can be the best choise for the NPN transistor , will try to built a 1mA current source to replace my ref02 base on

Regards
OS

[Kleinstein]

For high precision a NPN transistor is not a good choice, as the base current is also going through the reference resistor. A change in the gain (e.g. with temperature) would give an error. I would consider a low threshold FET (e.g.  SK3557 or maybe LL mosfet). I have not checked the ref. chips output voltage range.

If it has to be an NPN use one with very high gain.

[EC8010]

Rather than attempting to convert a MAX6126 into a CCS (which works, but not terribly well), you could make a CCS from an op-amp (perhaps OP07) having a FET on its output and use the MAX6126 as the reference voltage to the CCS. A lot depends on whether you genuinely need the final circuit to be able to source a current into a load connected to 0V, or whether it just needs to sink a controlled current from a voltage source. After that, it's a question of the required voltage compliance.

[Overspeed]

For high precision a NPN transistor is not a good choice, as the base current is also going through the reference resistor. A change in the gain (e.g. with temperature) would give an error. I would consider a low threshold FET (e.g.  SK3557 or maybe LL mosfet). I have not checked the ref. chips output voltage range.

If it has to be an NPN use one with very high gain.

Hello

Thanks voltage will be 2.5V can be 3.00V as that also available3.000V but 2.5V also easier resistor set up 4x1 in // and as the source current is not from the MAX6126 that even perhaps possible to reach higher current ?

Regards
OS

[Kleinstein]

For a 1 mA current source already 2.5 V are relatively high. The 2.5 mW at the resistore can cause a noticible thermal effect. So the choice is more between the 2 V and 2.5 V version. 

I have not found any info on the voltage range for the force pin. Chances are it can go all the way (e.g. a bit under 200 mV drop out) to the supply and probably also reasonably low (like 1 or 1.5 V).  So both a BSS138 ( threshold < 1.5 V) or SK3557 (theshold > -1.5 V) should likely both work with a 5 V supply and 2.5 V ref. With the 2 V ref. voltage the JFET may need to be lower threshold (like -6 bin), like > -1 V.

If one wants significantly higher current, the choice of the resistor gets more critical and one may want a reduced voltage like 1 V.

I don't see a big issue in using the ref. force sense pins instead of a separate amplifier. One may still want provisions for a capacitor as direct higher frequency path, just in case. A current source may also need some RC series element at the output to ensure stability - this also applied to the version with the extra amplifier. The max6126 has a pin for a filter capacitor - so filtering could be done there, if needed.

A separate amplifier may be an advantage of more current sources are driven at the same time.

[Overspeed]

For a 1 mA current source already 2.5 V are relatively high. The 2.5 mW at the resistore can cause a noticible thermal effect. So the choice is more between the 2 V and 2.5 V version. 

I have not found any info on the voltage range for the force pin. Chances are it can go all the way (e.g. a bit under 200 mV drop out) to the supply and probably also reasonably low (like 1 or 1.5 V).  So both a BSS138 ( threshold < 1.5 V) or SK3557 (theshold > -1.5 V) should likely both work with a 5 V supply and 2.5 V ref. With the 2 V ref. voltage the JFET may need to be lower threshold (like -6 bin), like > -1 V.

If one wants significantly higher current, the choice of the resistor gets more critical and one may want a reduced voltage like 1 V.

I don't see a big issue in using the ref. force sense pins instead of a separate amplifier. One may still want provisions for a capacitor as direct higher frequency path, just in case. A current source may also need some RC series element at the output to ensure stability - this also applied to the version with the extra amplifier. The max6126 has a pin for a filter capacitor - so filtering could be done there, if needed.

A separate amplifier may be an advantage of more current sources are driven at the same time.

Hello

Thanks

The 2.5 V looks to be the best balanced solution  and following your advise So both a BSS138 ( threshold < 1.5 V) or SK3557 (theshold > -1.5 V) should likely both work with a 5 V supply and 2.5 V ref.

I check to locate some BSS138 to make a trial , great to able to make expirement with afordable components.

Regards
OS

[zlymex]

The circuit is simple but not precise. Apart from the Ib of the NPN transistor the sense pin OUTS also draw a considerable current.

[Overspeed]

The circuit is simple but not precise. Apart from the Ib of the NPN transistor the sense pin OUTS also draw a considerable current.

Hello

Thanks for your feedback

So the solution with voltage ref source + op amp + precision resistor is better ?

Regards
OS

[RoGeorge]

I'd use the MAX6126 as a voltage reference, followed by an opamp with compensated Ib for the BJT, like this:
https://www.eevblog.com/forum/projects/precision-current-sink-i-found-a-gem-of-a-circuit/
or maybe a FET instead of a BJT, depending on the desired performance.

What is the destination/use intent for your circuit?

[Kleinstein]

There are no specs for the input current at the sense pin. This could be an issue, but it could as well be rather small and only the variable part of the current would matter. With the relatively high noise level even with the nois filter capacitor I would not expect very much current at the input. In addition the ref. is low power and thus not much current to spare.

[zlymex]

The circuit is simple but not precise. Apart from the Ib of the NPN transistor the sense pin OUTS also draw a considerable current.

Hello

Thanks for your feedback

So the solution with voltage ref source + op amp + precision resistor is better ?

Regards
OS

I think so.
The Multisim model gives 38uA(sink) for the OUT_S pin. Also, the sense pin current of similar reference LTC6655 is very large at 2mA!

Edit: I've just measured the OUT_S pin current of one of my MAX6126_25 by HIOKI DT4282, 33.4uA, supply is 3.0V, and an 0.1uF is connected to the output which is 2.50013V by DMM7550.

And this 33.4uA current seems very stable against temperature, it changes less than 0.01uA when temp is increased by 35degC making TC<10ppm/K.

[David Hess]

I have not found any info on the voltage range for the force pin. Chances are it can go all the way (e.g. a bit under 200 mV drop out) to the supply and probably also reasonably low (like 1 or 1.5 V).

The datasheet says that the voltage range on the force output is limited only by the dropout voltage, so it goes practically to the positive supply.