Impose a current through diodes

[HRicochet]

Hello to all,

I'm new on this forum and I permit myself to post a message on this forum because I have a small question in electronics. In fact, I would impose a digitally (SPI 3 wire) controllable current (ranging from 1mA to 300mA with a 2mA scale) through a circuit consisting of 20 diodes in series (Vf=1.3V so 1.3V * 20 = 26V total).

I would like that the power is idependant of my montage (ie: the current is the same as I connect 20 diodes or 25).

For this, I thought to make an "howland source" montage controlled by a voltage DAC. But in that case, the output of the amplifier must be able to deliver a 300mA current under 26V. Do you know an amplifier reference that is able to do this?

Do you think this is a good idea? or do you see other way to do?

Thank you very much for your help

HRicochet

 

[FvM]

For the intended unipolar voltage range, an OP controlled transistor current source (PNP or PMOSFET) seems appropriate,

 

[HRicochet]

For the intended unipolar voltage range, an OP controlled transistor current source (PNP or PMOSFET) seems appropriate,

Thank you FvM for your quick answer, so you mean a montage like the following one (even if it is a NPN on the link) ? :



So to conclude, this montage just takes an OP, a transistor passes 300mA maximum and a VCC>26V ?

Indeed, it seem less difficult to design than the "howland source" and it will need less components. Thank you again.

HRicochet

 

[FvM]

If you can work with a ground connected source, this circuit is the best solution.

 

[HRicochet]

Thank you very much FvM for your precious help, have a nice week end!

 

[crutschow]

.........................so you mean a montage like the following one (even if it is a NPN on the link) ? :



So to conclude, this montage just takes an OP, a transistor passes 300mA maximum and a VCC>26V ?

Indeed, it seem less difficult to design than the "howland source" and it will need less components. Thank you again.

You can use an N-MOSFET in place of the NPN BJT if you like. The MOSFET gives a slightly more accurate current source since there is no base current flowing through the RSENSE resistor.

The main reason for using a Howland source is if the load has to be grounded. In your case the load can be floating so the referenced circuit will work.

 

[HRicochet]

thank you crutschow for these detailed explanations

 

[HRicochet]

Hi all!

I come to you because I have another question. I tested the assembly previously mentioned in this thread and I noticed a little feature.

Indeed, I wish to impose a current from 0mA to 50mA (voltage controlled) through a diode array. This assembly "OPAMP CURRENT SINK" works very well and has the advantage of being simple to implement. However, when the control voltage is 0V, there is a current which flows in the circuit spite of everything (about 5mA :-().

To prevent this, I thought of using an N-MOSFET between the resistor R3 and GND which will act as voltage controlled switch (low Ron << R3). Thus, when Vgs = 0, no current will flow through the diodes.

https://obrazki.elektroda.pl/8995606700_1373702382.png

I do not had the opportunity to test this structure in practice. Do you see any problems? and do you think it might do the trick?

Thanks a lot,

HRicochet

 

[FvM]

A better solution would be an OP with an input voltage range including the negative rail. Many recent OPs have this feature, but not LT1001.

 

[HRicochet]

Thank you FvM,

I will see in your direction. But for information, do you think the N-MOSFET would do the job ?

 

[Borber]

Reason why output current can't go to zero is output minimum negative voltage of OPA. At single +15V supply it is about +2.5V. Add a 3.3V zener diode between OPA and base of transistor and 1k resistor between base and ground.

 

[HRicochet]

Thank you Borber,

I could experiment with SPICE and it works . However, in my design, I really need to cut the load circuit and that's why I considered the N-MOSFET. Can you tell me if this solution can work with the OPAM CURRENT SINK as I show in my previous link? I know that I will loose precision a bit with Ron but it is not really important for my application.

Thank you all

HRicochet

 

[FvM]

I don't see any purpose of using two output transistors instead of one. There are various ways to force a single output transistor to full off state, without using a second transistor.

My assumption is, that you primarly intend a linear current control down to (nearly) zero. This absolutely demands for a respective OP input voltage range. OP outout range, as mentioned by Borber, can be an additional problem. But a usual rail-to-rail or single-supply OP would be able to turn the transistor off without additional means.

Even under this conditions, there can be a small residual output current due to OP offset voltage. If it's a problem, you can superimpose a small offset to the circuit, or add a switch input.