TLV3605: Adjustable (internal) Hysteresis

Ken

Thanks for your post. Oddly enough, the best and simplest would just be a DAC. We should add this information to the datasheet. Placing a resistor to ground on the pin is simplest method for setting a fixed amount of hysteresis. Our internal current source times the resistance creates a voltage that our circuit detects and sets the hysteresis. However, when we test and created the curve we swept voltage and measured current to calculate the relative resistance values. 
If interested, I can search for our voltage vs hysteresis data for you. Likewise, I’d be interested to learn more about your application if you can share. If needed, we can take this offline via email. Just let me know your preference. 
Chuck

Chuck, thanks for the clarification.  I had a guess of what was going on inside and it was totally wrong. Yes, your voltage versus hysteresis data would be useful if it exists for several temperatures over the range -20C to +40C.  My application needs a peak-sensing comparator. The approach I tried (in simulation) was to input a time-differentiated version of the pulse to the comparator (TLV3604) configured with external stuff to make it sense the zero-crossing of that waveform. There was chattering both on the initial threshold crossing and the subsequent zero-crossing. To avoid the loop delay that, presumably, is causing the chattering the same function of the "external stuff" could be achieved with internal hysteresis combined with an external offset.  That is why I stated looking at TLV3605 and having questions about the internal hysteresis pin. 

Cool application Ken, thanks for sharing.  I found this characterization plot that shows hysteresis voltage vs voltage applied to the VHYS pin.

Blue is -40, green is 25, orange is 85 and red is 125C.  I hope this helps in your design.

Chuck

Hello Ken,

The LE_HYST pin can be modeled as an internal 40k resistor to a 1.25V source.

The internal 40k resistor creates a voltage divider with the external R_HYST resistor and generates a voltage across the external resistor on the LE_HYST pin. That voltage is proportional to hysteresis.

The hysteresis input range is 0.8V to 1.25V. At 1.25V, hysteresis is at minimum. At 0 to 0.4V, the output is latched, and above 1.25V is shutdown.

The SPICE model has a table of resistor voltage to hysteresis voltage, based off Chucks graph at 25C.

Where: VR_HYST, V_HYST

0.4,0 (< Output latches)
0.5,0.0636
0.55,0.0636
0.6,0.0636
0.65,0.0636
0.7,0.0635
0.71,0.0636
0.72,0.0635
0.73,0.0636
0.74,0.0634
0.75,0.0635
0.76,0.0638
0.77,0.0637
0.78,0.0637
0.79,0.0637
0.8,0.0636
0.81,0.0636
0.82,0.0636
0.83,0.0636
0.84,0.0425
0.85,0.0411
0.86,0.0398
0.87,0.0386
0.88,0.0371
0.89,0.0359
0.9,0.0347
0.91,0.0334
0.92,0.032
0.93,0.0309
0.94,0.0296
1,0.0223
1.05,0.0164
1.1,0.0108
1.15,0.0056
1.2,0.0007
1.25,0 (> Shutdown)

The LE_HYST pin can accept either a resistor or a forced 0.8V to 1.25V external voltage to set
the hysteresis voltage up to 63mV. So you could theoretically drive the input with an ADC - either direct voltage or current against the 40k resistor.

Paul, thanks for the extra information.  This is really helpful.