Current Source

Hi Antonio,

I found the following App notes in our website.  Hope it helps!

"MAKE A PRECISION CURRENT SOURCE OR CURRENT SINK"

"IMPLEMENTATION AND APPLICATIONS OF CURRENT SOURCES AND CURRENT RECEIVERS"

Hi Antonio,

I have some ideas but before I can recommend anything I need to know some more about your design constraints:

1) What power supplies do you have available, or is this flexible?
2) What kind of load are you driving? Is it reactive or purely resistive? Is the load grounded or can it be bridge tied?
3) By 40V aperture, I assume you mean the current source needs to have a compliance voltage in excess of 40V. Is this +/-40V?
4) What kind of bandwidth do you need?
5) What is your input sourced from and what is its range?

Hi Mayrim,

thanks for your hope.

The current generator I need has to be bidirectional and the output current has to be a sine waveform 2Arms (1mAdc max), so I think it is not so easy as shown in the appnotes you sent me.

Hi Zak,

thanks for your interest.

1) I think 40V/50V could be enough
2)The load can be also a time-variable load, in the sense that the load will be parzialized as resistive load/short, dinamically. About the configuration load grounded or other I don't have constraints, it depends on the best precision configuration. In this moment I don't know what is the best for precision, even if I'd prefer grounded.
3) Yes +/-40V of aperture, so I think the power supply has to be +/-40V or more
4)The max frequency of the sine current wave has to be 2kHz, even if the output current frequency will be 50Hz
5)I want to use a sine wave voltage max +/-10V, filtering and amplifying the output of a DDS (so I can change freq) and then using MDAC for attenuation. So when I use +/- 10V I have 2Arms in output. Just to say a VCCS (Voltage Controlled Current Source)

The current generator has to be bidirectional, 40V of aperture, 2Arms, 2kHz, and in particular offset <1mAdc.
I was thinking of using a sort of DC servo to cancel the offset for the power stage, and then another offset control using a DAC in the stage near DDS, but i don't know if it can work because i don't have experience about this high precision circuits.
For the final stage I would use an integrated chip configurated as a current generator, for example i was thinking of an Howland configuration, but i repeat, i'm not an expert.

Thanks for your help.

Perhaps you could consider using a high voltage power op amp (Apex Micro have quite a few) in an Improved Howland Current Source configuration?

regards

Steve

Hi AR,

 

I apologize for the delayed reply, please see attached for a concept design that should work for you: 

Bidirectional Current Source OPA454.TSC

Note: R7, R9, R10, and R12 are only in the schematic to help with convergence of the models and are not necessary in the actual design.

For a quick overview, in this circuit the power op amp drives a pair of external power transistors to boost the current (the OPA454 can only source up to 50mA). These transistors are class AB biased to minimize crossover distortion. The output current is sensed through Rset and fed back to the input amplifier to regulate the input to the OPA454. Having a gain of 5 on the OPA454 reduces the voltage that the OPA188 needs to source. The output current level is controlled by Rset because the INA117 is a high input impedance difference amplifier with a gain of 1 and thus the OPA454 will drive the transistors to source the current necessary for the voltage across Rset to match the input voltage of the OPA188. The primary source of error comes from the offset voltage of INA117, so the larger you can make Rset, the less current error this will contribute. Rset will have to be sized to handle the power dissipation and at 200mOhms you will want at least a 4 Watt resistor. To achieve full scale output range of +/-2.8A peak, your input needs to be +/-560mV, so you will need to include a precision divider at the input of the OPA188 if you are driving it with a +/-10V source.

You will need to carefully consider the power dissipation of the OPA454 and your power transistors and make sure to include proper heatsinking. If you are not familiar with this then we have a precision labs series on power and temperature that you can find here: training.ti.com/ti-precision-labs-op-amps-power-and-temperature

You mentioned that the load will change dynamically between a resistance and a short. This should be okay as long as the short condition is not held for any significant period of time, because under a load short condition the full supply voltage will be dropped across one of the transistors and if sourcing the maximum current this transistor will have to dissipate roughly 150 Watts of power. You can limit the power dissipation in the transistors by adding another power resistor in series with the transistor outputs, but this will limit your output swing and will also dissipate a considerable amount of power.

Hi Steve,

Using an Apex power op amp would definitely simplify the design as they have many devices that would not require you to add an additional transistor output stage, however this does come at a premium (many of these devices would be roughly 50x the cost of the OPA454). If cost is not an important factor than these devices would be a very good alternative.

I would not recommend using an improved howland current pump here though because it is very difficult and costly to get anywhere near the accuracy that AR is looking for with this configuration. Using 1% resistors results in a worst-case error of roughly 10% in addition to the fairly large offset error you will see from power op amps. The howlands are particularly useful if accuracy is not as much of a concern or if they are operated as part of a larger control loop, such as regulating temperature of a peltier.

Hi Zak,

thanks for your very great design solution; it is very clear what you mean and I like it.

The load is shorted by an active device controlled by a PWM. Can the output of the solution follow this load without problem and guarantees low DC offset (<1mAdc)?

Do you think that your solution is just enough and a loop that compensates automatically DC offset (something like DC servo or similar) is not necessary?

Anyway consider that in a previous stage, the +/-10V sine input (that has to be scaled by voltage divider in your solution) is a little trimmable in amplitude and offset by a micro for adjustment.

Hi,

The output maintains milliamp accuracy across most of the operating range, but as your load increases and the voltage at the input of the INA increases you will see some degradation in this accuracy. The error is dominated by the offset of the INA117 and the resultant current through Rset. The CMRR of the INA117 is around 70dB, which means every volt at the input results in an additional 316uV of offset. If you can find an integrated high input impedance difference amplifier with better offset and CMRR performance than the INA117 then I believe you could get to <1mA accuracy over the entire range. This was the best thing I could find in our portfolio for this application.

Alternatively, if you can float your load and handle a few hundred millivolts disturbance on your ground then you could use a low-side current sense configuration as shown below. This will be accurate within tens of microamps across the entire range.

Bidirectional Current Source OPA454 LS.TSC