Post Go back to editing

Very simple project ran smooth virtually, but real prototype does not.

Category: Hardware
Product Number: LT1819
Software Version: LTspice XVII(x64) (17.0.35.0)

I have a task to get simple lever shifter from 3.3V to 10V, exactly, for feeding the MOSFET driver from ESP32 microcontroller. It is double op-amp LT1819 used as comparator between fixed 2.7V and incoming signal from MC and logic inverter based on next op-amp. The SPICE netlist is as follows:

V1 N001 0 10 Rser=0.1
V2 N006 0 3.3 Rser=1
V3 N002 0 PULSE(0 3.3 0 5ns 5ns 1ms 2ms 10) Rser=1000
R1 N004 0 5000 tol=1 pwr=0.1
XU1 N005 N002 N003 N001 0 LT1818
XU3 N005 N003 N004 N001 0 LT1818
R2 N005 N006 1k
R3 N005 0 9.1k
.tran 5
.lib LTC2.LIB
.backanno
.end

1000 Ohm parasitic serial resistance in V3 (the virtual port of MC) eliminates the usual serial resistor placing. The simulation gives perfect result. But op-amps, bought from an reliable supplier (digikey.com) warm up quickly and does not act as expected.

  • I did same project simulation in NI Multisim 14.3 and surprisingly got equal to real situation result. 

    It means that the comparator design has error and could not run as expected by my side originally. Who can explain, why the LTSpice simulate as was the idea and goal, but the real model and Multisim are showing the project's being non-workable? 

  • Hi Vahe:  The LT1819 should not be used as a comparator.  Here is text from the data sheet;  "Sustained differential inputs, however, will result in excessive power dissipation and therefore this device should not be used as a comparator."

    Depending on gate capacitance and speed required, I would suggest using a comparator for this task.  An LT1713 would do well, as it can be supplied with 10V.  Or if speed is not required, an LT1716 might do it (very weak pullup).

  • Dear Glen, thank you, thank you and once again, thank you.

    You have solved a head destroying puzzle for me.

  • Hello Glen. I did another mistake and have immediately ordered (non-cancelable) at arrow.com which was the only Co. that has stock. As it seems, the LT1713 being good choice at first instance, requires 2 supply rails. Can I leave the ground floating and connect the "Minus" rail to ground, making single supply +10V? If floating GND is not a way, could be the synthetic ground, build up from serially connected capacitors, from rail-to-rail play role of ground just for comparator chip? Next to this chip will be connected a mighty driver-DEIC420. It is out of production since IXYS acquisition by LittleFuse, but has not an equivalent alternative for given project. The minimal threshold of high level is 7V for DEIC420 input port.

  • Vahe, the LT1713 is rated for single-supply operation.

    • Rail-to-Rail Complementary Outputs (TTL/CMOS Compatible)
    • Specified at 2.7V, 5V and ±5V Supplies

    A quick look at the datasheet shows an example on the first page.