Hi,

I found the attached howland current source from a book that has the equation vin / R. I tested it using a dc voltage source and it produced a close current to the equation. I then attached a rectified AC signal to it from my circuit and the expected current is way off.

Does anyone why?

 

This circuit is not a howland current source. But as for the question.
Your input AC signal create ac current on the outputs. But your DC ammeter show average value.

 

Not sure what you mean. I have attached the ac signal before it is rectified and smoothed with a LPF.

 

If so, there must be something wrong with your simulation.
Because the diagram is correct.

 

Do you mean the current source is wrong or the smoothed rectified signal?

 

The current source looks good. Also post the multisim file.

 

There would be a 'problem' with this circuit when the input voltage goes below 0V.

 

In fact there will be problems even before the input even approaches 0V. U9A output will be limited to say 7.5V. This means the input cannot be less than ~0.9V for correct operation.

 

here is my circuit in two parts. The third part is the current source.

 

Your current source needs some re-thinking for the reasons already stated. I'm not sure what you are attempting to achieve. Perhaps you might enlighten us.

 

Also I don't think the 4066 CMOS would like a 30V [±15V] supply.

 

The circuit will attach to strain gauges positioned on a cantilever beam.

I have used a wheatstone bridge at the start to replicate small changes in bridge voltage. The two bridge voltages are input into the instrumentation amplifier through two buffers then a subtractor. The gain is 21.

The amplified signal passes through the phase sensitive rectifier in part 2 to the LPF filter where it is smoothed. The CMOS 4066 switches between the top and bottom path and is driven by LM311N comparators square wave that is derived from the AC excitation source.

The smoothed rectifed voltage should pass through the howland current source and produce a linear 4-20mA current.

The Howland from a book is attached

 

OK I get the front end [pre-current source] concept now.

Why didn't you use a typical precision full-wave rectifier approach instead of the other approach with the comparator / analog switch idea? Seems to me to be over complicating matters. Interesting idea though it be.

 

In fact there will be problems even before the input even approaches 0V. U9A output will be limited to say 7.5V. This means the input cannot be less than ~0.9V for correct operation.

In my simulation in multisim the circuit work good. But we can always add a LED between U9A output and Q2 base.

 

The smoothed rectifed voltage should pass through the howland current source and produce a linear 4-20mA current.

The Howland from a book is attached

Do you understand the design limitation with respect to the op-amp output voltage swing I mentioned earlier?

 

I used the Phase Sensitive rectifier because it can output a negative rectified signal so it can work when the beam is bent upwards and downwards, also to calibrate the offset.

Sorry the CMOS has a 6V supply, the comparator has 15V

Not really sure how voltage supplies are decided.

Can you see why the howland current source isnt outputing the correct current? Do you think it has something to do with the supply of VSS and the Op-amps?

 

not really tnk, I have dont really know about the practical side of electronics, all I know is the theory. So what were you saying?

 

Thanks for that jony130, I have a look a while and get back to you.

 

So try to add the LED between U9A output and Q2 base.
And see see if this helps.

 

I can see how adding the LED would give you some extra headroom on the op-amp output. Eventually one still encounters a problem as the input goes towards zero volts.

One thing puzzles me with respect to the planned circuit operation. I see now the requirement for the phase-sensitive rectifier is to switch the rectified output polarity depending on the beam direction of deflection. As I see it, the current source will only accept a unipolar input. If the beam deflection produces a negative output at the phase-sensitive rectifier then this will be problematic as an input for the current source circuit.

At this stage the current source will simply mimic the current set by the input control into the output PNP transistor. Presumably you would have to provide an offset to the input control to set the minimum current to 4mA. I would guess the 4mA level would define the maximum beam negative deflection condition and the 20mA level would define the maximum positive beam deflection condition. Or do you actually want a bi-polar 4-20mA output?