TLV3701: TLV3701QDBVRQ1: Defective batch?

Hello Richard,

They should have the same pin out and behave identically, however there is a very big difference in the speed speed between the two (40ns vs 40us).

Is this a steady DC signal, or AC p-p input signal? The top theoretical speed for the nanopower TLV3701 is about 10kHz.

If the signal is >10kHz, the 3701 just cannot keep up. If around 10kHz, then some 3701's may be able to keep-up and some cannot due to device to device variation.

I have factored this The signal is essentially DC, there no AC and signal bandwidth is limited by RC circuit at the input.
The TLV3201AIDBVT was used because I have it on hand to verify the circuit function which proven correct, but it basically slow moving DC. Nothing much to do with speed.

Hi Richard,

What is the load? Is there a leftover pull-up resistor? The micropower TLV3701 has much less output drive (<2mA at 5V) and may not be able to sink/source enough current.

Can you provide a schematic?

DAC-VPAD provide 2.5V threshold.

AN8_VGEN3 connect to buffer op-amp as where it bring HV down to within 5V range. However, the circuit was not fitted at the time of the testing. The 10Meg DVM attached to this pin and measured 0.2V.

I use scope probe to check O/P as well as I/P. They all DC or low freq nature. 

With -VE = 2.5V and +VE = 0.2V, the O/P is HIGH which should be LOW.

The O/P attached to GPIO of the MCU which is input. 

All signals are low frequency, 

Under the same circuit, with TLV3201 which it worked fine. I need to add hysteresis circuit in next revision. This also proves the MCU GPIO is input. 

Hello Richard,

I just want to get this correct...

DAC-VPAD is set to 2.5V, which will apply 1.5V to pin 4 through the divider.

VGEN03 varies from 0 to 5V, depending on the monitored supply through a x7.2 divider. I was unclear about your statement "he circuit was not fitted at the time of the testing". How was the voltage set during your test?

Did you bypass the buffer and drive directly from the divider? What is the impedance of the divider? Would paralleling 10Mohm from the meter impedance drop the input voltage?Can you share a schematic of the input stage?

Note that some bench meters switch between Gig-ohm to 10M impedance inputs between the 2V and 20V ranges. This can lead to some bizarre results if the source is high impedance. Granted...the behavior should be the same between the two devices. Did anything else change? Does the output toggle if you touch the input terminals?

In any case, the output should go high when the monitored supply goes above 10.8V.

Confirm that you do see the 3701 toggle low, and that the output is not "jammed" high?

That is strange...

We have not had any reports of any devices that are wired 'backwards' or reversed. They would not have passed final test. The layout of the die and bond-wires makes reversing the inputs "accidentally" VERY difficult. Normally we see 'reversals" when the inputs are taken well below the GND pin (>1V).

The TLV3701 input is a rail to 'greater-than-rail' input (up to 5V above V+). The input bias current will reverse (current flows in to the input) and increase to nA's when the inputs are above V+. Is there any chance or condition where the inputs could exceed the V+ supply?

Are you sure the ground pin is connected?? Measure continuity at the "shoulder" of the pin, not the solder pad. It almost sounds like the ground pin is floating. Do you see any cracks around the pins?

Is the "high" and "low" output voltages within a few mV of the supply/ground?

Did all the 3701's do this? Or just a few?

I pointed out earlier that the circuit (idenical setup as TLV3701) was tested with TLV3201 which worked fine.

With this kind of questions which is not helping much and decided to scrap TLV3701 and move to other device from Microchip and Maxim which they do nanapower comparator. I order them for evaluation.

I do not have time to investigate (unlike in 1980s/1990s) and leave this open issue to you to sort out or ignore.

BTW: I am professional engineer for 30 years and not dumb with these comparator. You can do your test yourself if you really concerned about it rather than wasting time passing the buck to me via forum.

R.