Why 7805 is better than LM1085?

[tiroth]

I really don't think 1000uF is unreasonable after the reg in certain applications. The size of the capacitor will obviously have complex effects, but if you are optimizing for low noise (analog) or your load variation is above 100kHz (high speed digital logic) it can be helpful to make it large. You just have to read the datasheets and see what effect the size, and the ESR, have on the regulator. And sometimes you need to build it and see.

P.S. a certain well-liked designer advocates large downstream capacitance for analog circuits

[Fred Dieckmann]

Or maybe not..... large value capacitors have larger series inductances and thus potentially higher impedances at high frequencies. I have found that caps in the 50uF to 220uF range work better on digital circuits. Look at the impedance vs. frequency graphs to be sure.


Read your data sheets.
Fred

[SY]

Jan, thanks for the reply. I've used a lot of your ideas and circuits over the years.

Your point about not verifying observations is key; it does indeed lead to a lot of chasing of one's own tail. I noticed people on other threads exulting over how much their sound improved when going from a fast silicon rectifier (very low Z) to a tube rectifier (very high Z). Or even worse, a mercury vapor tube, a horrific source of HF noise (I last used 866s in my high-power 8005-based transmitter and the PS alone screwed up every TV set in the house).

So what am I to make of this? That a power supply should have a high and nonlinear source Z? People claim that "sounds" better. Or a low source Z? Other people claim, just as fervently, that THIS "sounds" better.

[Pjotr]

Patwen, a sine is not a very well idea, although it is possible. When you use a square wave pulse load, you are looking at the step response of the system in accordance with control theory. It is also easy and cheap to build with a few components. I don’t have it hand but if you want to experiment with, I can post a schematic of it. It is not a solution for the problems, but an analysing tool.

Fred, you are quite right to my opinion. Larger capacitors have larger inductance. In digital circuitry: If you use such large capacitors in combination with a 100 nF ceramic hf capacitor close to it, the inductance together with ceramic cap causes hf ringing on the supply lines. It is good practice in such cases to use 1 to 5 ohm resistors in series with the electrolytic cap to dampen the ringing. If you need a large capacitor, its is better to use a number of smaller ones in parallel to keep inductance low in such cases.

Capacitors on the output of a regulator are needed to keep the impedance low at higher frequencies. That is to say, from the point where the loop gain of the regulator drops to a too low value. Large values aren’t needed indeed. Large capacitors are something for after the bridge rectifier and for power amps and even then …

SY, what is good and what is bad? As always in electronics, every thing is connected to everything. Just as a large chain. But every bit counts, some bits count more and some bits count less. But let us start with proper engineering, it is difficult enough. Without it there is no floor to stand on and to be creative. Or is this just what you meant?

[janneman]

Originally posted by SY
[snip]So what am I to make of this?[snip]

I dunno SY.
Some time ago there was this guy starting a thread, asking a question, and got at least a dozen answers. Some of those were very valuable, clearly well thought-out and reasoned. Anybody in his right mind would be absolutely delighted with the concentrated knowledge and experience he got for free.
Not this guy. He just started a new thread, complaining that on the other one they gave him so many different answers, and pleaded if someone now please give him the RIGHT answer.
Go figure. It made me feel depressed.

Jan Didden

[Fred Dieckmann]

the rectifiers low impedance is only during the conduction periode when charging the filter caps. The Impedance seen by the input is more a funtion of the input capacitor seen by the regulator. There seems to be some evidence that limiting the peak current charging the filter caps can have benificial effects on diode generated noise and transformer saturation during peak cuurent from the rectifier bridge. The is also a widely held believe that too large a value of filter caps can cause problems for the same reasons. Every body raise thier hands that thinks audio power supply design is a slam dunk........

For inquiring minds:

http://www.gensemi.com/appnotespdf/quik108.pdf

http://www.microsemi.com/micnotes/302.pdf

http://131.109.59.51/images/pdf/Calc...m_Snubbers.pdf

http://www.diyvideo.com/forums/showt...ight=rectifier


Absolute power corrupts absolutely,
Fred


P.S. patwen

The LT1085 requires minmum load current of 10mA and probably more ouput capacitance for good perfomance. You have also told us nothing about input and adjustment terminal bypass capacitors that you used (I hope)

[SY]

But things like diode noise and transformer saturation are well-understood, quite measurable with that nasty ol' scope, and are factors only insofar as they cause the rails to be something other than constant DC under the conditions of the operation of the load circuitry. And the worst solid state diode I ever used was orders of magnitude quieter than an 866.

PS for audio a slam-dunk? No. Straightforward? Yes.

[tiroth]

Originally posted by Fred Dieckmann
I have found that caps in the 50uF to 220uF range work better on digital circuits. Look at the impedance vs. frequency graphs to be sure.

Read your data sheets.
Fred

Certainly--it all depends on the parts and the application. As far as HF ringing--hard to determine without testing. Francios' investigations with paralleled caps seem to indicate that we can sometimes get away with these designs that look like complex resonances waiting to happen.

One thing that hasn't been discussed much is regulator noise...some of those parts generate quite a lot of it. Does a few mV of low-frequency noise matter in a digital or high PSRR design? It shouldn't. Does it matter in a low-PSRR or input stage? Maybe. Those smaller caps won't necessarily get rid of it. (don't trust 'em, but check out those 7805 datasheets--a lot of variation with noise depending on the manufacturer--I've seen 1% quoted!)

Any thoughts Fred?

[dice45]

Fred,

There seems to be some evidence that limiting the peak current charging the filter caps can have benificial effects on diode generated noise and transformer saturation during peak current from the rectifier bridge. The is also a widely held believe that too large a value of filter caps can cause problems for the same reasons.

oh yes. For instance using a small power resistor at the input of the PS's CRC filter just squishing the cap loading peak a bit more wide and flat reduces filter ripple measurably and makes a world of a difference sonically. Tried out fr several occasions also something as weird as a PS for a loudspeaker field coil.

SY,
second your suggestion to "listen" to the music on the PS rail. And to response of square wave pulse load like Pjotr suggested. I look at them at the scope too.
Very telling, to look at the scope how the PS responds to varying loads.

All,
Feeding the PS output voltage influenced by such signals to a spectrum analyzer sounds like a very good idea to me, just, my analyzer stops at 60kHz, audio only. Have to try that out, mayeven feel urged to buy that RF analyzer combo at the local surplus shop.

[Pjotr]

Noise on digital PS lines doesn’t usually matter much indeed. But when it comes to AD and DA converters it has a large influence on jitter. It will shift the threshold levels of digital circuitry accordingly and introduces extra jitter.