0
\$\begingroup\$

Greetings fellow EE's:

Referring to the classic LM78xx series, I notice there is no specification on the maximum amount of ripple voltage these devices can handle at their input terminal.

So for instance a 7812 with 16VDC fed to the input. Let's say you have 15-20% of voltage ripple. It's probably okay, and yes it will be attenuated through the regulator, but is there a rule of thumb or maximum amount for best performance?

Your help very much appreciated.

A...

\$\endgroup\$
1
  • \$\begingroup\$ For "best performance" the maximum amount of ripple is 0. For a realistic answer, check the line regulation and ripple rejection spec in your datasheet. and make sure that your input voltage never drops below the minimum input voltage spec at the bottom of the ripples. \$\endgroup\$
    – brhans
    Commented Oct 16, 2018 at 17:14

2 Answers 2

Reset to default
1
\$\begingroup\$

The LM7805 datasheet (https://www.sparkfun.com/datasheets/Components/LM7805.pdf) provides a very clear spec for ripple rejection for VI = 8 V to 18 V, f = 120 Hz, which is minimally 62dB.

If there is ripple near those parameters on the input, so long as none of the other parameters are exceeded, output ripple will be at least 62dB smaller.

If you need something more exact, or smaller, you need to find a better datasheet, or more likely, a better part.

The ST version of the LM7805 (https://www.st.com/resource/en/datasheet/l78.pdf) has a similar spec, "Supply Voltage Rejection", of 61dB, and they also supply a graph that suggests you should be able to rely on that behavior up to around 10kHz.

As for "maximum" input ripple, I'd just assume that you can expect the above behavior so long as you respect the maximum (and minimum) input voltage specified.

\$\endgroup\$
4
  • \$\begingroup\$ Scott: Thanks. I am aware of that spec, but I am not sure it answers the Q. Ripple rejection I believe is the ripple in vs ripple out of the regulator. I'd like to know if there is a maximum amount that can be fed into the device. \$\endgroup\$
    – Aram
    Commented Oct 16, 2018 at 17:20
  • 1
    \$\begingroup\$ One normally feeds full rectified AC into a 78xx type regulator to get DC out, with caps on input & output to help with smoothing. At 60Hz AC, that is where the 120Hz above comes from. Input voltage needs to be 1.5 to 2V above the desired output. If the input is going from Vpeak down to 0, then the caps have to make up the current for the fall and then rise of the AC. Small load, you can get by with caps and lower input voltage. Large load, you need bigger caps and higher Vpeak to be able to have enough margin to regulate down to the desired output. So like most electronics - it depends! \$\endgroup\$
    – CrossRoads
    Commented Oct 16, 2018 at 17:29
  • 3
    \$\begingroup\$ @Aram Look to the dropout voltage required for the bottom peak of the allowed ripple, and look to max input for the top peak of allowed ripple. For a 7805 regulator this range is approx. +7V to +35V. Any ripple within this range is reduced by loop gain. \$\endgroup\$
    – glen_geek
    Commented Oct 16, 2018 at 17:34
  • \$\begingroup\$ For almost any linear regulator, the ability to reject ripple depends upon the idling-current of the regulator, because better regulation requires more stages and those need MORE current. Then, at high frequencies, the current soars because all those internal junction parasitics must be quickly charged and discharged. For low-power regulators, the ability to reject high frequency ripple tends to approach ZERO above 100KHz or 1MHz. Iinput ripple blasts right through to output. And these frequencies are exactly where ripple and ringing of the prior switching-regulator will be generated. \$\endgroup\$
    – analogsystemsrf
    Commented Oct 17, 2018 at 6:03
0
\$\begingroup\$

As long as input voltage variation (in a ripple or other form) doesn't exceed the max specification and doesn't go below (Vout + Vdrop), you should be fine. The dropout voltage Vdrop obviously depends on the load, so Vdrop_max would be a safe margin. And of course the frequency of ripples cannot/shouldn't exceed overall loop bandwidth of a regulator.

\$\endgroup\$
0

Not the answer you're looking for? Browse other questions tagged or ask your own question.