OPA547EVM: improved howland circuit

Hi Ahmed,

In retrospect, it would have been more useful if the OPA547/548 Improved Howland Current Pump example had been arranged for a much higher output current level. Nonetheless, the concepts are correct even at the very low current level used in the Figure 10 example.

Can you provide us your circuit schematic showing component values, supplies, etc.? Also, show the load information and provide the needed output voltage compliance range that the pump is supposed to work across.

Regards, Thomas
Precision Amplifiers Applications Engineering

Dear Kai

yes i did for RF+ i used 10k as i did not have a 9.09k at that moment but i dont think this is the critical point?

for power supply i varied from +/- 20V to +/- 30V wtihtout load must used 30V as you indeed get the 32V expected but it should have been with the load. i dont get it. attached the circuit without load and the traces. first trace is without load and second 20k.

Thanks

Ahmed

Dear Thomas

thanks for your feedback, i exactly reproduced the schematic fig 10. i attached a drawing with the values. for power supply i varied from +/- 20V to +/- 30V without load. without load i do get the 33.2V but drops half with 20k and 0 with 20ohm which i target.

best Regards

Ahmed

Hi Ahmed,

RF+ must exactly be 9.09k, not 10k. That's the cause of your problems:

ahmed.TSC

By the way, don't omit the power supply decoupling caps. Put them close to the OPA548. And shorten the cables length and component wires.

Kai

Hi Ahmed,

do you know this appnote?

snoa474a.pdf

Kai

Dear Kai

Thank you so much indeed the 9.09Ko is much more appropriate. i ve read the technical note i was wondering if reducing the resistance would help to get some signal if i use a 20-30ohms load since Iout is inversely proportional to R1.

Regards

Ahmed

Hi Ahmed,

have you thought about using a standard MOSFET current sink instead of the improved Howland current pump?

Kai

Hi Kai

no i did not know this configuration but i am not sure it give the flexibility of a Howland circuit? I need an AC current flowing through my load this is why i went for the Howland but i havent check if anything better.

i need to find a way to apply it to small loads, i tried with large parallel load but does not help.

Thanks

Ahmed

Hi Kai

My load is rather small it s a MEMS micro heater i need to heat with AC of less that 100 ohms typically 20-30 ohms. I need to pass low frequency < 10 Hz AC current varying from 10mA to say a maximum of  500 mA but if i get up to 150-200 mA i am happy with that. i try to put serial or parallel resistors to the load but it s not helping.

Thanks

Ahmed

Hi Ahmed,

you mean like that?

ahmed1.TSC

Kai

Hi Kai

thanks so much , if i read properly by injecting 4 V at about 7 Hz you get 200 mA on a 30 ohms load, correct? that sounds great , i am planning to vary current by steps of 20mA starting from 10 mA i wonder if the voltage source can go low enough. by the way i am trying to download the demo software to tune by myself, any how i ll try experimentally on Monday morning and give you feedback.

thanks again

Ahmed

Hi Kai

I managed to open your file and run simulations. with an input of 250mV amplitude and without DC offset i manage to get 13 mA amplitude and up to 250 mA with 5V input at 1Hz. i will try experimentally Monday and give you feedback. out of curiosity how did you work out the resistance values? i looked at the application note 1515A and tried experimentally figure 5 page 6 but was far from expected.

many thanks again

Ahmed

Hi Kai

Correct, i want pure AC through the load to generate second harmonic. I tried on simulation without DC it seems working. I ll test the circuit asap.

Thanks

Ahmed

Hi Kai

I tested your solution it works i just used 20 ohm as load to be closer to my device. it seems fine but the voltages necessary are higher than simulation by large.i can reach  240 mA at 11V  while for 4V as in the simulation i just reach 100 mA. The OpAmp heats up a lot and i had to add a heat sink to prevent thermal shutdown.one thing bothers me is the parasitic on the output signal just appeared today. overall it seems ok this way i can put that on a PCB. any recommendation welcome.

Thanks

Hi Kai

yes i put 100nF decoupling caps to power supply, components are as close as possible to op Amp but i iwill try to shorten some wires and add a capacitance in parallel to R2.

Thanks

Ahmed

Hi Kai

the output is much cleaner with a capacitance but surprisingly there was no effect when in parallel to R2 so i just decoupled the output and it seems decent to me, there is still some spikes but i feel it s the function generator that is poor, what you think.?

Thanks

Hi Ahmed,

it's difficult to get better results from a breadboard design. You would need a proper PCB to improve the results. You should even put the PCB into a shielding metal enclosure and use shielded cables.

To make the Howland current pump circuit perfect one would have to know all the parasitic components (inductance, capacitance) of the MEMS heater and the whole setup (cabling!). With this data it would easily be possible to properly compensate the OPA548. Do you have an equivalent circuit model of the MEMS heater?

Tim Green has made some remarkable stability analyses on the Howland current pump. You might want to search for them in this forum.

Kai

Hi Kai

the heater is a simple metal serpentin of about 20ohm nothing special so i guess can be model by R and L. I will sold all that on a PCB and have a go i just need to dissipate AC to another device the current design allow me up to 200 mA i am happy with that, i ll have a look at your reference to improve the setup. Thank you so much for your time and expertise really appreciated it.

Best Regards

Ahmed

Hi Kai

 the source is meant to supply on homemade microdevices my load resistance may vary upon nano fabrication. Do you think i could use potentiometers instead of fixed load to tune each time the set of resistances?

Thanks

Ahmed