I am trying to understand the constant current source at figure below. I couldn't understand the purpose of R6. The circuit just works as fine without R6. What is the purpose of R6?
R6 in your design only reduces the base current and effective hFE to the load, so delete it. In theory at max slew rate for low amplitude < +/-0.7V R6 could might improve small signal slew rates, but only marginally.
The bigger question in my mind is what is R5 for that separates +FB from -FB? In theory if +FB gain is greater or -FB gain or "over unity @ 0 deg." you have failed Barkhausen's stability criteria.
So you should attenuate your +FB loop with some value like R/2 value on Vin+ to gnd for margin. This will also reduce the ratio from V to I.
Here I put voltage signals on both inputs and replaced R5 with a fuse.
I used 10M just to plot the bottom trace then show the output current into 1 Ohm above that the output voltage and finally at top the average power into the PNP
R5 is the shunt used for current measurement and is part of the so called "improved Howland current source".
R3 and R4 are the paths to measure the voltage across R5.
The snipped article (which was missing a courtesy link) was not from professional designers at ADI but just a user who helped another in the ADI chat forum. Although the forum is moderated by ADI.
You may have missed my comment that it is unnecessary. All the resistors may be used to scale gain and V to I conversion.
Furthermore it reduces maximum current. That is why I replaced it with a fastblow fuse which should not affect the accuracy.
if you say R5 is unnecessary ....
and if you say the schematic is not from a professional designer....
Then you probably don´t trust the TI application notes either.
@OP:
All basically say the same as the "non professional designer". And they explain the function of an improved Howland current source without current booster.
And I agree with them.
if you say R5 is unnecessary ....
and if you say the schematic is not from a professional designer....
Then you probably don´t trust the TI application notes either.
@OP:
All basically say the same as the "non professional designer". And they explain the function of an improved Howland current source without current booster.
And I agree with them.
I'm a bit surprized about the confused discussion. Abstracting from the current booster which has no direct relation to current source topology, the 5 resistor or "improved" Howland source implementation can be found quite often. The fifth resistor gives an additional degree of freedom to set output current, gain and input impedance independently. The basic 4 resistor topology connects non-inverting OP input to load, in other words shorts R4 in post #1 circuit. R5 is still necessary.
The considerations about stability problems of 5 resistor topology are unsubstantiated. You can show that overall loop gain is negative.
Actually I overlooked the former output R5 as the current sense resistor and may be made as small as necessary to reduce losses which using voltage gain trades off BW.
My reference to current boost comes from the trannies with voltage gain by ratiometric balancing of resistors with any modest reactive load impedance is balanced on the sense side by ratiometrics values. This is the same configuration as the above "improved" Howland Current source with my high power version provides gain from V to I , i.e. current boost. Sorry if I wasn't clear enough. (CPU heatsinks needed for this Class B design but high GBW of Op Amp eliminated the need for bias diodes.)
I'm a bit surprized about the confused discussion. Abstracting from the current booster which has no direct relation to current source topology, the 5 resistor or "improved" Howland source implementation can be found quite often. The fifth resistor gives an additional degree of freedom to set output current, gain and input impedance independently. The basic 4 resistor topology connects non-inverting OP input to load, in other words shorts R4 in post #1 circuit. R5 is still necessary.
The considerations about stability problems of 5 resistor topology are unsubstantiated. You can show that overall loop gain is negative.