Recent Posts

why not use a linear adjustable regulator as current source which will also safe some space in circuit see attach milliohm metter, in Elektor electronic a nice design was published with op-amp based current source I will search it and share

thanks guys for all these inspiring ideas.
will try them
but I have a work project that has to be closed in the next weeks and I won't have much time for the shorty project now.
Hope to have more time for this in july. stay tuned.

would be good to understand why the osc is occurring. If you want to change the current source, just do it with an opamp and transistor.

I totally agree that using an opamp based CS will be superior. However, that needs a complete pcb modification. I was thinking along the lines of minimal modification if there are already pcb's produced. The TL431 can be implemented without much trouble in place of the zener.

It will be useful to see some waveforms and to establish if the design performed correctly after the current source was modified the first time round.

would be good to understand why the osc is occurring. If you want to change the current source, just do it with an opamp and transistor. You can crush supply LF supply noise by nearly 120 dB. it will have superior performance in all regards. That is, if it was the problem to begin with-- which hasn't been established yet.

Here is a cheap modification for the current source that is worth a try maybe.

Replacing the zener with a classic TL431 may provide improved regulation with consuming about 50% of the current required for the existing regulator.

Below shows, the diagram and the result of imposing a 2 Vpp ripple on the  5 VDC for testing purposes. The new current source (AM1) shows a variation of only 0.02 mA while the old one (AM3) varies by a few mA's. The variation of the old source shows some non-linearity while the new one appears more symmetric.

These are just simulation results. I have not tested it in hardware.  

Looking at the used Opamp DS, the slew rate is rather low (the AD8628 https://www.analog.com/media/en/technical-documentation/data-sheets/AD8628_8629_8630.pdf opamp being an instrumentation opamp that's not surprising) which should make it rather insensitive, but it might help if you add some small capacitors in parallel with the negative reaction resistor (33k by value): a 10pF to 30pF might help but you can try with a higher value, into the nF range. Like @optikon said, the oscillation you see might be just an envelope of a higher frequency oscillation.
Other than that, given that you have placed the decoupling 0.1uF capacitor between VCC/VSS, the only thing that may be the cause is to have some capacitance connected where is not supposed to

If I were in your place I would disconnect the 1k resistor found on the negative input of the opamp and check if there are still oscillations. This way you can determine where the oscillations are caused. You can also remove the current source transistor (or whatever you have in the new schematic). Bottom line is that you have to isolate the exact location where those oscillations are generated. After that it is much easier to fix the problem by adding some capacitors.

I used a LM317 as an adjustable current source. After tweaking the correction factor in code and current adjust, I get repeatable readings.



****corrected schematic attached

10 - 20Hz is a strange frequency. Too low to be a loop oscillation directly or mains. But it might be an alias of a much higher frequency aliasing with your scope or whatever it is you are using to measure with.

If you have supply oscillations, you should be able to see them

1) measure your power supply pins. Do they also oscillate?

Do you have a common mode oscillation because the whole thing is floating, sitting on your desk or ESD mat ?

2) Measure your ground, is it also oscillating?

kripton2035  wrote

there is no LC or RC or ferrite bead on the +V line, but as the power supply is a 9v battery is it really necessary ?

no the initial schematic was a little different, there was no zener look at the preivious page of this thread, or the schematic on my web page for the project
the new current source is less dependant of the supply voltage, so it should have a better reproductibility of the measures.


I suggest you carefully lift the AD +ve supply pin, and add the RC

Then you can see if it fixes the problem

Zener diodes tend to produce white noise, it may be worth analyzing in this direction.

or better he used temperature compensated voltage reference like LM336DZ 3.6 volt