• State TI Thinks Resolved
  • Locked Locked
  • Replies 13 replies
  • Answers 1 answer
  • Subscribers 47 subscribers
  • Views 3596 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TLV1701: Strange behavior of output of open-collector comparator - sourcing current when output should be Hi-Z

Aleksandar Lazarov81
Prodigy 220 points
Part Number: TLV1701
Other Parts Discussed in Thread: TLV1702

Hi Everyone,

I have a particular weird situation on my bench. I'm using the TLV1701 to detect if a resistor is present on two pins or not. This signal is going to a microcontroller, which will need to do something with this information.

At one particular instance in time, the TLV1701 has these potentials around it:

Vcc = 20V
Vss = GND
V+ = 13.4V (comes from a divider - 2/3 of 20V)
V- = 10V (comes from a divider - 1/2 of 20V)
Vpull-up = 3.3V
Rpull-up = 220k

The initial thing that I observed was that the output voltage, on pin 4 was 3.4V. This seemed odd as I don't have any 3.4V rails on my system (I double checked the uC supply rail). The next thing that I did was to put the Rpull-up (220k) resistor to ground. Having the same potential levels around the TLV1701, the output voltage on pin 4 is 165mV, which with the 220k resistor to ground, ends up to being ~700nA of output source current.

The datasheet specifies a typical value of output leakage current of around 70nA, but the value that I'm getting is an order of magnitude bigger. Also I suspect that the datasheet value is leakage current sunken by the comparator, and not sourced.

Has anyone seen this behavior, and maybe hint to a cause on what might be causing this ? I don't like when my open collector comparators source current on the output :-)

PS: The idea is that it will do a level shift from the 20+V world to the 3.3V uC world voltages in my design.

  • 0
    TI__Intellectual 1740 points
    Hi Aleksandar,

    Did you verify if you were seeing the same behavior on another device? Just want to eliminate it being a damaged part before we continue debugging.

    Regards,
    Jaskaran
  • 0 in reply to Jaskaran Atwal
    Prodigy 220 points

    Hello Jaskaran,

    Yeah, I did try it with another device, and the behavior was completely the same. Now I'm going to try it with another device, on a board on it's own, and a TLV1702, just to see if I get the same problem.

    Cheers,
    Aleks

  • 0
    TI__Guru 63366 points

    Hello Aleksandar,

    Can you provide a schematic of the circuit? Particularly the values of the supply divider resistors?

    I think you maybe seeing the effects of the pull-up resistor paralleling the divider.

    Remember that resistor dividers are not regulated and any parallel resistance paths will change the tap values (Mr. Kirchhoff is haunting you). When the output is "low", you now have a parallel 220k resistor from 3.3V to the V- tap, "lifting" up on the V- tap of the divider. This will shift the comparator V- (ground) up.

    Also beware at these current levels, be sure to include the parallel load of the DMM or scope probe (usually 10Mohm) during the measurement.

    Comparators can have slight changes in the supply current depending on the output state, a few percent change is typical. So the quiescent current drawn is not consistent. This can also change the tap voltages.

    Is the uC sharing the V- ground? Or the system GND ground? What is the voltage on the V+ tap when you measure the 3.4V?

  • 0 in reply to Paul Grohe
    Prodigy 220 points

    Hi Paul!

    Just to eliminate my surrounding circuitry, and have a closer look at the comparator, i have the following circuit on my bench:

    The +20V, +9V and +8V come from a Delta bench power supply (common GND). I'm measuring the output of the comparator with a scope, with a x10 probe (so 10MEG loading impedance). The value that I'm getting on the scope is around 156mV DC. This is the same as in the previous experiments... If I reverse the potentials on the input (V+ to 8V and V- to 9V), the output drops like a rock to around 20mV (which would indicate a sourcing leakage current of 90nA - I can live with that).

    I tried even with getting the input potentials as close as 10mV one from another, the V+ being the higher one, and the output is still stuck to 156mV.

    I'm going to test a TLV1702 now, just to verify that I didn't get DOA part :-)

    Cheers!
    Aleks

  • 0 in reply to Aleksandar Lazarov81
    Prodigy 220 points

    Hi all,

    After testing the same scenario with one comparator from the TLV1702 (while keeping the other one's output to ground - configured the inputs so that V- > V+), the same behavior can be observed. When the output voltage should be high-Z, I get around 160mV on a 220k resistor pulling it to ground. I have done the measurements with both a scope, with x10 probe, and a FLUKE 87V, the results are the same.

    I'm starting to suspect that I've got ether a bad batch, or this is the behavior of the chip (which I can't believe that it is the case).

    Any inputs will be highly appreciated!

    Cheers,
    Aleks

  • 0 in reply to Aleksandar Lazarov81
    TI__Intellectual 1740 points
    Hi Aleks,

    After taking a look at your schematic I believe the issue is with your "pull-up" resistor. The TLV1701/1702 are open collector devices and require a pull-up resistor to allow the output to transition high. Right now your "pull-up" resistor is GND and not allowing your output to go above ground even when you expect it to transition high. Instead connect the pull up resistor to a positive voltage (any voltage up to 36V above the negative power supply). This should allow you to see the high and low output transitions.

    Regards,
    Jaskaran
  • 0 in reply to Jaskaran Atwal
    Prodigy 220 points
    Dear Jaskaran,

    Thanks for the answer. My investigation (as mentioned in my first post) began after a strange behavior was seen with a 220k resistor and Vpull-up of 3.3V. In this scenario, when the V+ was higher then V-, the output of the comparator was around 3.45 volts. This can not be true if the TLV1701 is truly just open collector.

    If I do the same math, it follows that there is a voltage drop on the pull-up resistor of around 150mV, and current flowing _out_ of the TLV1701 when it's output should be HIGH.

    Bringing the pull-up down to GND just confirmed my assumption that there is current that is sourced by the comparator. This weird behavior is what brought me here :-)

    Cheers!
    Aleks
  • 0 in reply to Aleksandar Lazarov81
    TI__Intellectual 1740 points
    Hi Aleks,

    Sorry about that. Now I understand why you used the pull up resistor to ground. I am going to take the TLV1701 to the bench and see if I am able to replicate the leakage current values given in the datasheet and the ones you have provided with your test circuit. I should have a response for you by the end of the week.

    Regards,
    Jaskaran
  • 0 in reply to Jaskaran Atwal
    TI__Intellectual 1740 points
    Hi Aleks,

    I apologize for the delay in response. I was able to verify the leakage current observations you were seeing with this device. We are currently exploring if this is in fact expected behavior.

    Regards,
    Jas
  • 0 in reply to Jaskaran Atwal
    Prodigy 220 points

    Hi Jas,

    Is there any update on the matter?

    Cheers!
    Aleks

  • 0 in reply to Aleksandar Lazarov81
    TI__Intellectual 1740 points
    Hi Aleks,

    After working with design team, we believe this behavior may be due to a clamp in the ESD protection that can leak current from the supplies into the output in certain conditions. We are still investigating further. What are your requirements for the comparator? We can try to recommend another part in the meantime if you would like.

    Regards,
    Jaskaran
  • 0 in reply to Jaskaran Atwal
    Prodigy 220 points
    Hi Jaskaran,

    Thanks for the update. At the moment I'm rethinking if the powering scheme should be changed in my design, so that the supply voltage is in the bull-park of the pull-up voltage on the output node.

    If this doesn't happen, my initial requirements are, maximum input voltage >32V, supply current as low as possible (battery environment). I don't really care about Vos, as I will deal with large signal swings, so couple of mV (or tens of them) don't really make any difference.

    Cheers!
    Aleks
  • 0 in reply to Aleksandar Lazarov81
    TI__Guru 63366 points
    Hello Aleksandar,

    After reviewing the TLV1701 design, this is expected behavior. Your devices are not "damaged".

    Without getting into internal proprietary information, there is an ESD clamp between the output and the V+ supply pin. This clamp needs to allow large voltage differences between the V+ and output, so it is not just a simple diode.

    Depending on the voltage differential between the output "high" voltage and V+ supply voltage, there can be as much as 1uA output leakage current due to the clamp.

    The normal usage condition of the output is either pulled-up with a fairly low valued resistor (<100k) to >3V, or being sunk by the output transistor (or another device on the output bus).

    The output was not designed to be pulled-down to zero by a high resistance while the output is "high" (not saturated). Doing this places a large voltage differential between the output and V+, which causes the leakage current to flow. Normally this current is sunk by the output device (or other devices) and is not seen.

    The 70nA typical datasheet leakage spec is for Vpullup equal to Vsupply - a minimal difference. You were placing a 20V-3.3V=16.7V difference between V+ and output, which would result in a much higher leakage current. The 700nA is not unreasonable.

    So in your design, you can either decrease the value of the pull-up resistor to minimize the effect, or, add a second high value pull-down resistor to bleed away the leakage and bring the VOL voltage to within spec.

    We are planning to have this characterized, and add a graph to a future update to the datasheet. No timeline as of yet.

    If the input voltages do not exceed 18V, there are other devices that can be used that do not have this output leakage, but they do not have the same pinout..