TINA/Spice/INA133: Tips to carry out stability analysis for improved howland current source

Hugh,

Since your question focuses on the design of the circuit, I will move your inquiry to the Precision forum,
That forum is monitored by the product experts, who will be better able to help.

Regards,
John

Hi Hugh,

the transient analysis shows the following result:

Hhm, this does not look very stable...

hugh.TSC

Kai

There are two key presentation I attach for you to study.  The "Design and Apps Current Sources" contains an important section on the Improved Howland Current pump, especially on driving inductive loads. The "Dual FB Beta_plus and Beta_minus" gives an alternate approach to analyzing loop stability for the Howland.  The INA133 the way you are trying to use it is not a good idea. Replace it with an op amp because we will need to be able to use compensation around the op amp . Read through attached and rebuild your concept circuit and re-post and I will help you with the stability.

8231.Design and Apps Current Sources.ppt

4048.Dual FB Beta_plus and Beta_minus RevD.pptx

Kai, I know its not very stable, and had looked at the transient response, but I didn't know how to go about fixing it.

Thanks very much tim, I think that info was very useful.

I have switched over to using another OPA192 to close the loop instead of the INA133.

I have then run through the referenced material and I would just like to confirm I am understanding the compensation correctly.

The inductive load is effectively adding a pole to the loop gain plot, causing the phase margin to drop below 45, causing overshoot and/or instability. By adding the RC network to the op-amp + terminal we are adding an additional zero to cancel out the phase shift of the pole from the inductive load. Therefore to ensure stability I need to insert the zero at around the frequency of our coil with the lowest resonant frequency.

The Ro value is in place because voltage source we are using to split things is ideal, and Ro is just being used to emulate losses in the output stage of the op-amp.

Does this all sound about right?

With the compensation added I can get the following open loop response plots.

Assuming I have simulated this correctly now only the purely restive case is on the edge of acceptable.

Transient response also looks better, it looks like there are some effects going on with recovery times when the driver stage is saturating but I suspect I can clean that up with a resistor and some zenners between stages.

As a side issue is there any way to get the post processing curves to work with stepped parameters instead of having to do a bunch of manual cleanup on each plot.

Hi Hugh,

when you have opened TINA-TI click onto the tab "Analysis". Then click onto "Select Control Object". Click with the left mouse key onto the component you want to be stepped. A window will open. Finally click on the button "Select..." and enter the desired parameter settings.

Kai

Hi Hugh again,

by the way, what speaks against the use of OPA548 in your Howland Current Source?

e2e.ti.com/.../673984

Kai

Hi Kia

I have used the parameter stepping before I was just asking after a way to have it so if I define a curve is post processing i.e. if I define Aol = VO(s)/Vbeta(s) and simulate a single step, if I then enable parameter stepping my VO(s) turns into VO[1](s), VO[2](s) etc and the extra curves error out. I was after a smart way to have it create the Aol, and 1/B curves for each of my parameter steps.

I didn't select the OPA548 due to its large offset voltage (would cause up to 0.2mA offset in our use case). I also didn't want to use it as a buffer as I wanted to avoid having to deal with the extra phase shift. It is also much pricier than the buffer I used in my circuit.