[sdiy] Problem Solved! was: voltage ladder with hysteresis

[mark verbos]

Hi There,

I changed the resistors on the input opamp and ran 1M resistors from  
the outputs of the comparators back to the summing node of the input  
OP Amp. I had to add a 30 k resistor to +15 to compensate for the  
offset from all those feedback resistors, but now it's workin' like a  
champ. The shift for each comparator's hysteresis is now the same  
amount for each threshold and the transitions are nice and crisp. Is  
this technique new or am I just now discovering some old news?

Thanks to those who helped...

Mark






On May 13, 2007, at 7:49 AM, Steve Lenham wrote:

> ----- Original Message ----- From: "mark verbos"  
> <mverbos at earthlink.net>
>
>> Hi there,
>>
>> I built a ladder of 16 comparators on a sequencer. The CV goes   
>> through and inverting stage and thresholds are set with a string  
>> of  resistors from ground to -15 volts, going to the non-inverting  
>> input.  I found the transition points were blurry, and it needed  
>> some  hysteresis. I added 1M resistors from the comparator outputs  
>> to the +  inputs and this works fine on the lower stages, but I  
>> find that when  I sweep the input with a saw wave, now the lower  
>> stages go slow and  it moves faster as it goes to the right. All  
>> the way on the highest  stages wouldn't even switch right until I  
>> upped the feedback resistor  to 2M. Is there some standard way to  
>> figure this problem out?  Should  I just put a 1M on stage 1 and a  
>> 2M on stage 16 and divide equally  between?
>
> Hi Mark,
>
> You are using inverting comparators, so your hysteresis resistors  
> are all injecting current into the string of reference resistors,  
> but at different points. Therefore the effect that they will have  
> on the reference voltage at that particular node on the string will  
> vary. Imagine the hysteresis resistor and the "bottom" part of the  
> reference string as a potential divider between the output of the  
> comparator and ground.
>
> For the "lower" stages, the resistance to ground is fairly low, so  
> the comparator output voltage will be divided down significantly  
> and the change to the reference voltage will be small.
>
> However, the higher up the chain you go, the higher the resistance  
> to ground, with the voltage being divided down to a smaller degree  
> and the effect on the reference voltage being correspondingly  
> larger. The highest stages will require increasingly large input  
> voltages in order to turn on (remember that as well as making the  
> comparator stay on more easily once its reference has been  
> exceeded, hysteresis also makes it slightly harder to turn on when  
> it is off).
>
> The calculations for your circuit are complicated by the fact that  
> several comparators can be on at the same time and so the effect of  
> all 16 hysteresis resistors needs to be taken into account at all  
> times.
>
> I would suggest that you look at using non-inverting comparators  
> (Google for it), where the reference string feeds the inverting  
> input whule the input and hysteresis feedback are applied to the  
> non-inverting input. This isolates the hysteresis for each stage,  
> allowing you to easily set the same amount of hysteresis for each  
> with no interaction between stages and no need to consider  
> different hysteresis resistor values for each stage.
>
> HTH,
>
> Steve L.