LOG114: Precision problem with LOG114 ...

Hello Uber,

Do you have the thermal die pad on the underside of the LOG114 soldered to a corresponding pad on the PC board that is connected to V-?

What are you using as you current source for I1, the input current?

How are you measuring the offset of A1 and A2?

Can you provide a complete schematic for your LOG114 applications circuit?

Regards, Thomas
Precision Amplifiers Applications Engineering

Hello Thomas,

thanks for the interest … below the answers.

Do you have the thermal die pad on the underside of the LOG114 soldered to a corresponding pad on the PC board that is connected to V-?

The Exposed thermal die pad is connected to V- (-5V) … as indicated on page 22 of the datasheet.

What are you using as you current source for I1, the input current?

I1 is generated by a 5K resistor connected to one side at pin 4 (log114) and the other to a linear voltage generator (range 10mV to 4500mV for 2uA – 900uA)

How are you measuring the offset of A1 and A2?

With an HP Precision Multimeter ... the same result with the oscilloscope ...

Can you provide a complete schematic for your LOG114 applications circuit?

I am arranging for it ...

Thank you for your cooperation.

Regards, Uber.

Hello Thomas,

to complete the previous message, attach the partial schematic of the circuit section involved.

I have just tested several prototypes and have the same problem: if the current I1 exceeds 150uA, the offset on the input (referring to VCMIN) exceeds the 4mV specification and is carried at 30mV with 900uA.

I'm waiting for clarification.

Thank you for your cooperation.

Regards, Uber.

Hello Uber,

Thank you for the additonal information regarding your LOG114 circuit. I have studied your circuit and ran some simulations to check the performance. Fundamentally, I am not finding any issue with your design at this time.

"I have just tested several prototypes and have the same problem: if the current I1 exceeds 150uA, the offset on the input (referring to VCMIN) exceeds the 4mV specification and is carried at 30mV with 900uA." Previously you mentioned that you measured the offset with a precision DMM and oscilloscope and had the same results. Where exactly in the circuit are you connecting the meters in the circuit? What are the input impedances of the DMM and oscilloscope? Has every board you have tested behaved the same way?

Regards, Thomas
Precision Amplifiers Applications Engineering

Hello Thomas,

below the answers:

Previously you mentioned that you measured the offset with a precision DMM and oscilloscope and had the same results. Where exactly in the circuit are you connecting the meters in the circuit?

Directly on pins 4 and 5 of the log114 amplifier.

What are the input impedances of the DMM and oscilloscope?

DMM HP34401A with input impedance of 10Mohm.

Oscilloscope Probe TEK P3010 with input impedance of 10Mohm (13pF).

Has every board you have tested behaved the same way?

Same problem on 5 board ... but I have 7 more to test.

Regards, Uber.

Hello Uber,

I am not sure if you can accurately measure the voltage offset of A1 by connecting a DMM or DSO directly across the LOG114 I1 and VCM_IN inputs. My concern is the noise injection into I1 input and the possibility of causing the LOGAMP to become unstable. The transimpedance gain may be very high and the noise will be amplified masking the true output level.

For those reasons that is not the way that the logging amplifier's offset is measured. The input offset A1 and A2, is measured by applying matched common-mode currents to each input I1 and I2. Ideally, using VLOGOUT  =  0.375 log (I1/I2); if I1 = I2, then VLOGOUT = 0 V. Any deviations from 0 V are associated with offsets. The common-mode input current leves are manipulated to not only derive the input offset of A1 and A2, but also A3. The voltage offset of A4 and A5 are measured using the same techniques applied to other op amps.

Originally, you mentioned "By raising the current I1 from 2uA to 900uA on the VLOGOUT output there is an increasing offset, compared to the theoretical value, ranging from 26mV to 90mV (with the growth of I1)." Putting the the offset change aside for now (disconnect the meters from the inputs), what is the issue with the VLOGOUT output over the specified input current range? Do you have a plot of VLOGOUT vs. I1?

Regards, Thomas

Precision Amplifiers Applications Engineering

Hello Thomas,

below the answers:

I am not sure if you can accurately measure the voltage offset of A1 by connecting a DMM or DSO directly across the LOG114 I1 and VCM_IN inputs. My concern is the noise injection into I1 input and the possibility of causing the LOGAMP to become unstable. The transimpedance gain may be very high and the noise will be amplified masking the true output level.

I too immediately thought of a noise problem on the input of the amplifier but two tests performed make me think that perhaps this is not the reason:
1) The oscilloscope measurement does not indicate a oscillation  or high frequency component but a fixed offset voltage (DC)
2) By keeping the output voltage (VLOGOUT) under control by putting or removing the DMM on the input of the amplifier the output does not move ...

For those reasons that is not the way that the logging amplifier's offset is measured. The input offset A1 and A2, is measured by applying matched common-mode currents to each input I1 and I2. Ideally, using VLOGOUT  =  0.375 log (I1/I2); if I1 = I2, then VLOGOUT = 0 V. Any deviations from 0 V are associated with offsets. The common-mode input current leves are manipulated to not only derive the input offset of A1 and A2, but also A3. The voltage offfset of A4 and A5 are measured using the same techniques applied to other op amps.

With I1 = I2 the offset on the VLOGOUT is 45-50mV ... it seems to me at specification limit ...

Originally, you mentioned "By raising the current I1 from 2uA to 900uA on the VLOGOUT output there is an increasing offset, compared to the theoretical value, ranging from 26mV to 90mV (with the growth of I1)." Putting the the offset change aside for now (disconnect the meters from the inputs), what is the issue with the VLOGOUT output over the specified input current range? Do you have a plot of VLOGOUT vs. I1?

Attached.

data.pdf

Regards, Uber

Hi Uber,

Thank you much for the latest information. The graph of VLOGOUT vs. I1 nicely illustrates the issue you are observing with the LOG114 boards. Is there any possibility that you could run the graph again with a log current scale on the horizontal axis? That would give us a better idea of what the error looks like at the low end of the current range.

We are going to have to start to look for potential causes that lie outside of the ordinary. An issue that user's of precision amplifiers sometimes get suprised by is a PC board's cleanliness. More than once in recent times we have encountered customer circuit electrical problems that were eventually traced to PC board flux contamination. Residual solder flux left on a PC board after cleaning can support leakage currents that disrupt normal, precise functionality.

We clean our PC test boards using an ultrasonic cleaner and run them through the cycle not once, but twice. If you have that capability to clean boards I would suggest giving that a try. Even if it doesn't solve the problem, that is one more thing that we can check off the list.

Meanwhile, I continuing to look for any other cause.

Regards, Thomas
Precision Amplifiers Applications Engineering