Comparator stability - AD8561 versus LT1711

Hi gil_he:

Do the oscillations occur when the inputs are near zero cross?   That could actually be a noise problem rather than an oscillation, and might be fixed with a little more hysteresis.

The LT1711 is quite a bit faster than AD8561.  The LT1713 is more similar to the AD8561.  If you didn't want the extra speed of the LT1711, then you could use the LT1713 instead.   The higher bandwidth of the LT1711 will give you more total input noise.

PCI or breadboarding strip??

What value bypass caps?

How far away?

Put 10pF across the feedback resistor.

Harry

Hi GlenBrisebois:

The oscillations occur near the zero crossing (the circuit works that way).

My understanding is that when the input signal slew-rate is small enough, the passing through the linear region of the comparator is slow. As a result, a very little noise on the inputs and a very high gain of the comparator might cause oscillations. Is my understanding correct?
In addition, to my understanding, when the output slew-rate is high enough and the input hysteresis is too small, interlead capacitance between input and output pins (parasitic output to input feedback) might also cause oscillations.

I used a 20mV hysteresis and a 133mV hysteresis, with no improvement.

In addition, with a square input signal there were no oscillation effects. It happened only with a sine wave input signal with low enough frequencies (low enough slew-rate).

I also realize that the LT1713 has similar gain to the AD8561 and that might do the trick.

Why the higher bandwidth of the LT1711 will give more total input noise?

Thank you !

Gil

Hi Harry,

I stitched the device through a 1-2mm wires to pcb pads (that occupied the ad8561). Bypass caps (on the pcb) are 10uF on each rail. The pcb is a layout pcb board.

However, with a square input signal there were no oscillation effects. It happened only with a sine wave input signal with low enough frequencies (low enough slew-rate), so I assume the stitching is not the cause.

I put a 4.7pF and also a 1nF on the feedback resistor, with no effect.

Thank you !

Gil

Hi Glen,

I am testing now the LT1713. Same setup.

The oscillations are much "better".

With an input signal of 10Hz, feedback cap of 4.7pF, hsyteresis of 40-100mV (few scenarios), Iout of 10mA (also tested with 7mA):

Output /Q is stable. No oscillations.

However, output Q starts oscillating (at the beginning of the change) at frequencies of 5KHz and lower.

Why is this?

thank you

Gil

Hi Gil.

Not sure what is so magical about AD8561 compared to LT1711.   LT1711 seems at first glance to be a faster part.  

I believe Harry and others may have talked about supply bypass.   In addition to trying a cap across the hysteresis feedback resistor (which helps to couple the output into the input very quickly), you should make absolutely sure that your bypass capacitor is VERY close.  You can try adding 100nF also.  10mA or other output current turning on suddenly will generate a big L*dI/dt spike, so you need your bypass capacitor close.  

Note, the AD8561 datasheet shows Voh and Vol with only 3.2mA.  Less current.  Less twitchy behavior.

Sorry if this does not exactly answer your question...