Alternate Approach For A Reduced Noise Protection Circuit?

Go To Last Post
7 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello everyone!  First time posting in this forum . . . hopefully this is the right place!

 

We have an ASIC that is outputting a 0.5 - 4.5V signal into an ATTiny5 that is reading the ADC voltage and processing it into another signal.  One issue I'm facing is that the Vout line of the other ASIC is also the programming pulse line so we have a resistor and 5.1V zener diode protecting it (feeding directly into the ATTiny5).  The issue is increasing the voltage on the diode is causing a lot of noise and non linearity on the ADC input.  For instance, the current pulled across the resistor is less than 1 uA at 1.5 V, is 50 uA at 3.9V, and about 300 uA at 4.5V!  This obviously presents us with a pretty noisy (and non linear) issue.  

 

Does anyone have an alternate approach to this issue?  If I continue the route I'm headed, I can probably find a better suited zener for this application (we're currently using a MMBZ5231BW zener, not sure why this was picked as, unfortunately, no one ran the design by me until it was built up . . .) and a better resistor value (currently 1k due to cost/power/size) but I'd like to hear if anyone else has another potential approach to protecting our ATTiny5 from a 13.5V programming pulse.  

 

Unfortunately, the chip is full so no digital algorithms can be used to solve the noise issues so I'm looking for an analog alternative.  

 

Any help you guys can give me would be appreciated!  Thanks!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Is a physical link out of the question? Maybe one that MUST be removed to enable your programming connector to be inserted?

 

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Is this for a Prototype, to be used while developing the software (firmware), or for the finished product?

 

A simple Program / Run switch would work... until the human forgot to put it in the Program mode.

 

I shouldn't give suggestions without a while to think about them, but perhaps just the series R and then a schottky diode to the V+ rail.

The ADC signal should be < Vcc, and presumably the schottky diode with a 0.15 V Vf might conduct less than the soft-knee'd zener.

 

That said, I've not looked at the V-I curve for a low Fv schotty lately.

Just the first thought I had.

 

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you could route the 13 V programming voltage through the second half of a DPDT switch, coupled with disconnecting the uC, then the Program / Run switch would work.

If the human in the system forgot to flip the switch, there wouldn't be any programming voltage supplied.

 

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The old zener diode problem! It catches a few people out. I'd suggest a diode to rail. Some might suggest a schottky, but they are leaky and noisy. This may be an issue in your app, so consult the datasheet. If problems persist, consult a medical professional.

Last Edited: Thu. Oct 1, 2015 - 02:21 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

smiley

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:

Is this for a Prototype, to be used while developing the software (firmware), or for the finished product?

 

A simple Program / Run switch would work... until the human forgot to put it in the Program mode.

 

I shouldn't give suggestions without a while to think about them, but perhaps just the series R and then a schottky diode to the V+ rail.

The ADC signal should be < Vcc, and presumably the schottky diode with a 0.15 V Vf might conduct less than the soft-knee'd zener.

 

That said, I've not looked at the V-I curve for a low Fv schotty lately.

Just the first thought I had.

 

JC

 

Well, it will be for a final product at one point but for now, it's in the prototyping stage.  Unfortunately, from a manufacturing standpoint, it's not practical to remove the connection from the ASIC to the uC (which sucks because that would definitely be the easiest solution) and, since we're calibrating the coefficients of the ASIC based on the output of the uC, we would need them both connected and can't calibrate one and install the other at a later stage.  

 

I can actually think of a couple of ways to do it that involve adding a few more IC's but from a space and cost analysis, it seems a bit impractical.  I had actually considered the schottky to V+ rail yesterday but ran out of time to test it.  Guess it's time to do that today!  I know it adds some noise as well (I used something similar to protect the inputs of an ADC but removed it because it added noise) but I don't know how it matches up comparatively.  

 

Thanks for the suggestions, everyone!

Last Edited: Thu. Oct 1, 2015 - 01:56 PM