TVS diode acting bizzare..

[Reza]

I've got a comparator whos inputs i wanted to protect from overvoltage,
so I used a 5.0V TVS diode, but I'm getting all sorts of bizzare
results.

First off, I'm powering the board off a 300mA transformer. There's a
TVS diode between the +5 and GND lines, but it seems like it's sucking
a lot more power -- I measure only .9V with the TVS diode in place, and
4.9V without the TVS diode.

I've also used them on data lines, but it breaks the PWM communications
when it's added.

I generated a 5V 50Hz square wave from a MCU and had the output line
connect to a TVS (which was connected to ground). I didn't see any
change in the output w/ and w/o the TVS on my scope. It seems like
it's behaving as it's supposed to (letting 5v through), but in both the
case where it's attached to the power line and where it's attached to
the data line, it's causing grief.

The TVS diode I chose was the SMAJ5.0(C)A, which is supposed to be a
bidirectional diode. They can't be installed backwards, can they? Any
thoughts?

Thanks,
Reza

 

[Ban]

The TVS diode I chose was the SMAJ5.0(C)A, which is supposed to be a
bidirectional diode. They can't be installed backwards, can they?
Any thoughts?

Thanks,
Reza

Look at the marking:
AE is unidirectional, WE is bidirectional
I do not think it is a good thing to have these on inputs, they are good for
power supplies. Mount them with the stripe to the +5V.
Be aware that there might be not much protection, the device will need 6.4
to 7.0 V to conduct 10mA.
On the inputs a dual schottky to the respective rails will be much better,
because then the internal diode of the comparator will not be activated. I
often use two cheap BAT66 (each has the diodes paralleled) for a really
sensitive input, all 4 diodes have 120pF together instead of 1.2n of the
SMAJ5.0A and the cost is 2x0.04.

 

[Jamie]

I've got a comparator whos inputs i wanted to protect from overvoltage,
so I used a 5.0V TVS diode, but I'm getting all sorts of bizzare
results.

First off, I'm powering the board off a 300mA transformer. There's a
TVS diode between the +5 and GND lines, but it seems like it's sucking
a lot more power -- I measure only .9V with the TVS diode in place, and
4.9V without the TVS diode.

I've also used them on data lines, but it breaks the PWM communications
when it's added.

I generated a 5V 50Hz square wave from a MCU and had the output line
connect to a TVS (which was connected to ground). I didn't see any
change in the output w/ and w/o the TVS on my scope. It seems like
it's behaving as it's supposed to (letting 5v through), but in both the
case where it's attached to the power line and where it's attached to
the data line, it's causing grief.

The TVS diode I chose was the SMAJ5.0(C)A, which is supposed to be a
bidirectional diode. They can't be installed backwards, can they? Any
thoughts?

Thanks,
Reza

you do know that most transformers after rectified produce more than the
voltage stamped on the unit?
if you were to pass this into a filter cap, i suppose it would try
to hold the TVS to a clamp.
then you have the holding current of the TVS.
http://www.worldproducts.com/pdfs/WPITVS.pdf
there is a nice link for you to look at.
my self, i would use a higher hold off voltage

 

[Reza]

The markings look like this...

D:42R
HE

And the stripe is near the GND on inputs. I'll grab some of the
BAT66's and swap them out to see if they behave better. Thanks.

Reza

 

[w_tom]

4.9 volts without how much ripple voltage added? A 5 volt
diode would not be enough margin. And what is the tolerance
of that diode? 10%?

If this is a Zener diode, you have it backwards.

 

[Pooh Bear]

The markings look like this...

D:42R
HE

And the stripe is near the GND on inputs. I'll grab some of the
BAT66's and swap them out to see if they behave better. Thanks.

Remember that TVSs are designed to clamp ( after a fashion ) power
supplies *not* signals.

Diodes to rail and ground are the solution for signals. You should use a
small value series current limiting resistor too ideally.

Graham