Hello, I am trying to design a precise VCCS on a +/- 5v supply, which is then used as a reference current in a wilson current mirror with the output leg connected to a voltage charge pump, allowing for higher impedance loads than with the Howland Current pump alone.

The maximum desired current is 2.5 ma, although I could live with a lower maximum current of no less than 1.5 ma. I am using a TLE2426 to split the rails with a virtual ground. The opamp is TL074. Charge pump is using 22 uF electrolytics, the frequency for Sig and SigInv is 1.8 khz, the transistors are 2n3904 and 2n3906.

The circuit works perfectly when both legs of the mirror are connected to 5v vdd. As soon as I connect the output leg supply to output of the charge pump, the following occurs (Vref for Howland pump set to 0). 15v at charge pump falls to 6v, current flows through Rs at about .2 ma, current flows backwards though output leg at -40 ma, the multivibrator output peak to peak voltage drops from 5v to 1 or 2v.

As I decrease Vref, current flows through Rs as normal, and current continues to flow backwards through the output leg at a constant -40 ma.

Is there a better way to do this without increasing the supply voltage of the entire circuit? Can I then mirror this design with an NPN mirror to sink current from the negative rail as well when changing Vref to a positive value?

Thanks in advance

 

It's not clear what you are trying to achieve.
What's the maximum plus and minus constant current that you want, and what's the maximum "higher impedance load" you want to drive with that current?

 

I am trying to get a variable current of 0 - 2.5ma in either direction. The highest impedance would be 6000 ohms. Really I’m trying to mirror my current that is going through the sense resistor Rs to the other side of the Wilson mirror but while using a higher supply voltage. I’m not sure if the error is from me misunderstanding the fundamental design of the current mirror or with the voltage pump.

 

According to Ohm's law, the most you can get with a 10V supply is 10/6k = 1.67mA.
Why do you think you can get more?

A current-mirrors maximum current is limited by the supply voltage, that same as the Howland circuit.
2.5Vma will require a supply of at least 2.5mA * 6k = 15V.

 

According to Ohm's law, the most you can get with a 10V supply is 10/6k = 1.67mA.
Why do you think you can get more?

A current-mirrors maximum current is limited by the supply voltage, that same as the Howland circuit.
2.5Vma will require a supply of at least 2.5mA * 6k = 15V.

So even with a voltage charge pump to 15v, I won’t be able to source 2.5 ma? Also, did you see the attached pdf with the circuit on it? Thank you for helping I do appreciate it a lot.

 

So even with a voltage charge pump to 15v, I won’t be able to source 2.5 ma?

Yes, 15V across the 6kΩ resistor would give 2.5mA.
But for ± current you would need to apply ±15V.

did you see the attached pdf with the circuit on it? T

Yes, but I saw no 15V charge pump.

 

Sorry I think my drawing is a bit unclear. Is this any better? I have the charge pump input and output highlighted, the multivibrator circuit next to it is providing the 1.8khz timing signals. I added the lower mirror section as it is my idea for getting reverse current. The impedance of the load in that direction will always be less than 2000 ohms so I'm pretty sure I’ll be okay on compliance voltage, but I haven’t yet tested it to find out if it will work.

 

Please take a look at the basics of a 4-transistor Wilson Current Mirror, shown nicely at https://circuitdigest.com/tutorial/...ilson-and-widlar-current-mirroring-techniques. From that you have A) your input transistor wired wrong, B) the supply/ground side emitters at different voltages, but they need to be connected together and at the same potential, and C) a basic misunderstanding of the current mirror input voltage - it is not flexible but rather locked at 2 Vbe from the supply/ground. As such it must be current fed as the current mirror determines its input voltage. Consequently the input is unipolar, not bipolar as you desire. (Additionally the Vbes need to be matched and the transistors held isothermal, a virtual impossibility with discrete parts.)

Perhaps a better solution to your goals is an OTA - Operational Transconductance Amplifier. The CA3080 OTA has a current mode output in which the current is the product of a (unipolar) bias current and a differential input voltage. Take a look at SBOA117A from that Texas semiconductor company for other more interesting OTA devices.

 

So why not just run the Howland op amp at a higher voltage to get the current you need?

If you can float the load, you could use two Howland circuits connected with the source connected to one of the inverted inputs to give a bridge type output and double the load voltage for a given supply voltage.