78xx + FAT capacitors = ripple elimination??

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I was just thinking. The ripple performance of a 78xx (after bridge rectifier) is just so-so using the small datasheet-recommended input and output caps.

What if we just put GIANT caps on the input AND output sides of the 78xx? Would that effectively give ripple performance on par with the expensive fancy voltage regulator circuits? Are there any downsides to this? If ESR is a concern, what if multiple low-ESR caps in parallel were used?

Thanks!
 
One problem with using a huge cap after the reg is that on power-on,the cap looks like a short,and may trigger the regulators overload protection. Having said that,I've put some large caps (up to 10,000uf or so!) after the 78xx regulators before,and not had any real problems that I can remember. You will want to make sure that the regulator has a diode connected across the IN and OUT pins,reverse biased (cathode to input side),so that on power-off,the output side can can discharge into the input side. Most regulators don't like being reverse biased,which can happen if the input side cap discharges faster than the output side cap.
 
Why not try putting a shunt-type capacitance multiplier after the 78xx?

IIRC, the 78xx will become a 1A CCS if shorted, so there shouldn't be a problem with using large caps.

Beyond that, better performance can be had if you build a discrete regulator (I'll bet you can get better than the 78xx simply by building around a TL431, with no other ICs).

- keantoken
 
AX tech editor
Joined 2002
Paid Member
I was just thinking. The ripple performance of a 78xx (after bridge rectifier) is just so-so using the small datasheet-recommended input and output caps.

What if we just put GIANT caps on the input AND output sides of the 78xx? Would that effectively give ripple performance on par with the expensive fancy voltage regulator circuits? Are there any downsides to this? If ESR is a concern, what if multiple low-ESR caps in parallel were used?

Thanks!

One thing to remember is that ripple rejection for mains hum isn't an issue for amplifiers, as they generally have hugh ripple rejection ratios at lf. The (load induced) hf ripple is more a concern for amps, but that is less easy to get rid of with just more capacitance. Caps thend to become less effective at hf due to series ESL for instance.
BTW Increasing cap after the rectifier increases hf components in the ripple so that may actually work backwards.

So if your concern is not per se with mains hum, you'd better going to some hi performance regulator, with, the ultimate, at-the-load remote sensing.

jd
 
Thanks for the help so far. The 78xx's are for a DAC. So it's okay to put a 10,000uF output cap on without worrying about a short-circuit on powerup causing the thermal protection to kick in? I guess it would be okay if it was current-limited to 1A, I just wouldn't want it to shut off completely. Also, is the diode absolutely necessary in practice?
 
If your output cap is much larger than your input cap, it is a good idea to have a diode there. Probably don't need a huge one, just one that can withstand some surge current depending on your output cap.

It shouldn't turn off completely if it goes to 1A. These regs are used as CCS sources in tube amps and such; I don't think you have anything to worry about.

- keantoken
 
I was just thinking. The ripple performance of a 78xx (after bridge rectifier) is just so-so using the small datasheet-recommended input and output caps.

What if we just put GIANT caps on the input AND output sides of the 78xx? Would that effectively give ripple performance on par with the expensive fancy voltage regulator circuits? Are there any downsides to this? If ESR is a concern, what if multiple low-ESR caps in parallel were used?

Thanks!

In general you do better with mulit-stage filters then just making one stage bigger. The effect of the stages multiply. But making a big stage is additive.

You will see better ripple rejection if you split the large cap in two and connect them with a resistor. This makes a "pi" filter. It is very much better then the same amount of capacitance all in one cap. Of course the resistor drops the voltage. The is the price you pay but if the regulator is after the pi filter it does not matter much.

If you don't like the resisor's voltage drop you can get better performance with an inductor but at higher cost. If you need very clean DC you can use multi-section pi filters.

More info:
http://en.wikipedia.org/wiki/Capacitor-input_filter
http://www.engineeringpost.com/modules.php?op=modload&name=News&file=article&sid=5
 
Putting a big cap after the reg isn't a good idea.

Besides the "short on power up" issue, it will make ripple rejection worse, beacuse the error amp input is effectively shorted to ground.

Big cap on input, then let the regulator do its job :)

Cheers!
 
Thanks for the help so far. The 78xx's are for a DAC. So it's okay to put a 10,000uF output cap on without worrying about a short-circuit on powerup causing the thermal protection to kick in? I guess it would be okay if it was current-limited to 1A, I just wouldn't want it to shut off completely. Also, is the diode absolutely necessary in practice?

For a DAC, I assume Digital to Analog Converter with a small current load, a better option would be a Shunt regulator like the TL431 (Adj. Zener), which will have lower ripple than a Standard Linear regulator.

Putting a Big Cap doesn't really make that much since, because Big caps usually have higher ESR, Inductance and leakage. A small tantalum or MLCC that has a very low ESR will probably perform better. MLCC caps have lower ESR when the Max Working voltage is low and the capacitance is fairly small, so its a good idea to use MLCC caps that are slightly higher than your max working voltage. Parallelling several MLCC also helps as long as you keep the connections paths very short. Long traces add copper resistance which will increase ESR and possibly inductance depending on PCB layout. For a low current load of less tha 25 mA, a set of 0.1 uf MLCC and 1.0 uf tantalum, will provide better noise reduction than a signal 10,000 uf electriclytic cap.

If your voltage source is coming from a unfiltered Linear PS or a high ripple SMPS, then use a PI filter between the PS and and the Regulator. If you need ultra low ripple ( < 2 mv) then use a coupled inductor in your PI filter. I've used the couple inductor to remove ripple for charge sensitive amplifiers and other devices that require very very low input noise. The key to making a really good Coupled inductor is very tight coupling between the two windings. Either use a Bifalir winding or twist the windings together with a couple of twists per inch. You can also designed a tuned couple inductor if you PS noise is periodic (ie SMPS switching frequency).

http://www.hamill.co.uk/pdfs/ciabfbb_.pdf
http://www.analogzone.com/pwrt0712.pdf

Related DIYAudio Thread:
http://www.diyaudio.com/forums/showthread.php?threadid=118746
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.