offset issue in improved Howland current pump using LF356n

Hi Tejas,

I assume the LMP8601 is placed in series with the LED line somewhere. How much offset are you seeing? Does the offset go away if you remove the LMP8601 and just measure the voltage across RL to measure current?

The LMP8601 has several microamps of bias current, that will be in parallel with the LEDs. The current will look resistive (~300k x 2), and even goes negative when the LED terminal voltage is under 2V.

See Figure 3 or 4 in the LMP8601 datasheet. Remember the graph is per input, so the actual current drawn from the LED line will be twice what the graph shows (two input bias currents combined).

Regards,

HI Paul,


1.I have placed LMP8601 as a highside current sense amplifier. I have placed the LED's only to visually demonstrate the change in current direction. I am using a sine wave generator as Vin.

2. I have connected the offset pin of LMP8601 to 5V. I will remove the LMP8601 and measure for the offset. I will test and update it as soon as possible.

3.What I mean when saying offset is , the Voltage Va in my circuit when measured using a oscilloscope is with a positive offset. As a result the LED1 lights up during the positve cycle and LED2 doesnt light up for lower amplitude sine wave. But for higher amplitude it turs on but its intensity is lesser compared to LED1. So, I suspect that the offset may be produced by LF356.

4. I failed to understand your statement, "The LMP8601 has several microamps of bias current, that will be in parallel with the LEDs. The current will look resistive (~300k x 2), and even goes negative when the LED terminal voltage is under 2V " . Can you please explain it?

Regards,
Tejas S J

Hi Tejas,

I re-read your initial post and missed the fact that you had grounded pins 1 and 5. DO NOT ground these pins! These pins are for manually trimming the offset and connecting them to ground can partially disable the input stage.

Instead, there should be a 20-50k pot placed between the pins with the wiper connected to V+, as shown in Figure 55 of the LF356 datasheet. If you do not desire trimming, leave pins 1 and 5 OPEN with NOTHING connected to them. Note that these are sensitive nodes and should be kept as short as possible and close to the amp to prevent noise pickup.

In order for the LMP8601 to be able to measure below ground, without getting into the gory details, the LMP8601 input stage has a circuit that will force reverse current into the internal resistors when the common mode is less than +2V (the internal bias point of the LMP8601). The bias current crosses zero around 2V then goes negative. This is shown in Figures 4 and 5 (bias current goes through zero at 2V). Since 2V is around the LED voltage, you may be seeing the 8601 input current switching directions on the positive swing.

The LMP860x is designed to measure from low impedance sources, like power supplies capable of 1 to 1000's of amps of current, where a few micro amps of current change would not make a difference. Your circuit may be sensitive enough to see these changes. Just keep Figure 4 in mind when measuring the current, as the LMP8601 inputs are a parallel ~100k load to the LED.

Does the circuit work into a resistive load? Is the output swing proportional into a resistive load?

Are the LED's identical? Is one damaged? Even briefly applying over-current to an LED can cause it to permanently "dim" (less light for the same current). What if you swap the LED's - does the behavior change?

BTW: I assuming you are reading AN-1515 by Bob Pease:

www.ti.com/.../snoa474a.pdf

Regards,