PI-Filter

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello again...

Originally posted by GRollins
> For those who don't remember a certain thread
> from last summer, Keith likes to argue more than > anything else in the world.

Heh!

> As for intuitive...PI filters seem quite
> reasonable to me; make perfect sense.

I never said they didn't. What I guess I'm trying to say is that a properly designed PI filter may be helpful in reducing hum and noise, while a poorly designed one may be less than useless. I'm trying to be helpful by providing a rederence for information.

Keith
 
Hi Wuffwaff,
In my Ampzilla amplifier it is very crowded too.
I managed to find a place on the heatsink as far away as possible from the transformer. As the inductors are ferrite cored the field is concentrated in the coil and they don't seem to pick up hum.
As you can see the inductance of the coils I use it much higher than yours with only slightly higher DC resistance.:)
 
As for shielding, it'll need to be something ferromagnetic. Iron, nickel, cobalt, or alloys of same (i.e. steel). Then ground it. If that doesn't do the trick, then there's nothing like distance to do the trick--i.e. a remote power supply. Yes, they're a pain in the rump sometimes, but they work.
I don't remember if it's been mentioned yet or not, but try reorienting the transformer this way and that. Transformers do not have uniform radiation patterns and sometimes you can fix things that way. Of course, the underlying assumption is that you've got room to move things around.
Someone--grataku(?)--had a similar problem last spring and ended up moving the offending transformer outboard to cure the problem. Bummer if you hadn't planned on doing it ahead of time, but I believe it got things going for him.

Grey
 
OK, useful information regarding the coils used in PI filters.....

Make an assumption. The coils are largely carrying DC, with an AC component that is due to two factors.
1. The ripple voltage
2. Current varying as the output level of the amplifier varies.

These will cause a varying magnetic field around the coils. If the inductance is low or a non-saturating iron based core is used, the size of this field is likely to also be low. Thus, if the coils are mounted as far as possible away from the input circuitry, then all should be hunky dory.

Due to the small size of the inductances, the cross coupling between the coils should be minimal, but to minimise this, mount the coils at right angles to one another.

Of far greater import are the values of the inductors and capacitors. Take the standard PI filter. The two capacitors and the inductance form a parallel resonant circuit. Under certain dynamic conditions, this circuit may add a ringing voltage to the D.C. output which will affect the amplifier. The requirement in solid state amplifiers for low resistances will exacerbate the problem as the Q or quality factor of the circuit will be high. The higher the Q, the higher the ringing voltage superimposed on the power supply voltage. The cure is to ensure that the inductance is sufficiently large to keep this resonance well below the amplifier's passband. Use the standard formula fo=1/(2 x pi x sqrt(LC)) where C is the value of the two capacitors in series (in farads) and L is the value of the inductance (in Henries). A rule of thumb is to have the resonant frequency down around 2Hz for a 20Hz to 20kHz passband.

This phenomenon may explain why some people perceive a change in the sound characteristics of amplifiers when they add inductances into the power supplies.

Hope this help.

regards, Keith
 
Hello Keith,

according to your formula this would mean between 0,3 and 3farad for an inductance of 2 to 20mH. This 0,3 to 3 farad would be the series value of the two capacitors!:eek:
Since I would need this 4 times I think the extra case I have to build for the power supply will be to large to fit my living room.:rolleyes:

How critical and under wich circumstances will this ringing occur?

william
 
wuffwaff said:
How critical and under wich circumstances will this ringing occur?

william [/B]

That's the problem. Much will depend on the physical characteristics of the components involved. In solid state amplifiers, we go for large wires, low resistors and low ESR values for the capacitors. As stated, these characteristics can cause a high ringing voltage.
When does the ringing occur? Well there are several theoretical causes. Sudden large changes in current requirements can cause a step in the current going through the inductance. This step may excite the ringing. Another possible cause is if the current is varying at the resonant frequency of the PI filter. Another thing to look at is the length of time the ringing lasts once it's started. It may only last for one or two cycles. It may last last for a considerable period . All this will depend on the damping factors, and they're affected by the resistances around the circuit.

When will ringing occur in reality? Who knows. It may be there and you can't observe it.

I remember looking at this some years ago and, in the end put it in the too hard basket. I decided that the physical construction of the power supply had a far greater effect on the system than the addition of a couple of inductors. If you're determined to use inductors, the L filter doesn't suffer from the same problem.

For those of you thinking about valve amplifiers, they too may suffer from the same problem but to a lesser extent. The filtering chokes used usually have high resistances (of the order of several 10s of ohms). There are also often resistors in series with the diodes and sometimes bleeder resistors across the capacitors. There are also sometimes load resistors across the power supplies in order to maintain a minimum load as some filters work best with a minimum DC load.

Then there is the standard RC filter used in many situations. In combination with lead inductance of wires and PC board tracks, these can also ring.

The design and implementation of power supplies is in no way simple. If you modify an already designed PSU, you really need to know what you're doing.

Enough of my blathering. Have a great New Year.
regards, Keith
 
Actually an oscilloscope and a square wave at the input will do a very nice job of detecting any problems with ringing.
In my experience, it's absolutely a non-issue. In theory, it can happen, but in practice I've never built a circuit that had a problem with a ringing power supply. The only thing I've ever had come up was motorboating if too many gain stages were run off the same rail, but that's a no-brainer to detect and cure. Oh, and oscillations with chips, but that's easy to cure also (...don't use chips, duh...just kidding).
When I first started keeping bees, I read over twenty books on the subject. Every one of those books had at least one chapter on diseases and pests that can effect your bees. Dozens of nasty possibilities: American foul brood, European foul brood, tracheal mites, varroa mites, chalk brood...the list goes on and on. It's intimidating. You know how many of those problems have ever cropped up here? Two. Varroa mites and the small African hive beetle, which is such a new problem that it wasn't even in the books. The moral of the story? Too much theory about problems that <i>might</i> happen will drag you down. The only cure for too much theory is to leaven it with hands-on experience. That way you learn what really is important.
The only ringing you'll get is if someone calls on the phone while you're soldering.

Grey
 
Hello Grey,

I´ve also come to this conclusion and I´ll just put it together and have a look at it with the scope and a nice square wave........
I always think it´s an advantage not to be an electronics engineer when building amps cause I never know what isn´t suppose to work, what can go wrong and what´s impossible to have any influence on soundquality. .:)

happy new year to all of you,

william
 
GRollins said:
Actually an oscilloscope and a square wave at the input will do a very nice job of detecting any problems with ringing.

When I first started keeping bees, I read over twenty books on the subject. Every one of those books had at least one chapter on diseases and pests that can effect your bees. Dozens of nasty possibilities: American foul brood, European foul brood, tracheal mites, varroa mites, chalk brood...the list goes on and on.
Grey

Good analogy Grey that supports just what I'm saying. Grey, as a result of all that study, you have the knowledge that you can use should such a thing happen. Like all theory it gives you the background knowledge for use in the real world. The real world has taught you what will happen in your world and only your world. In other circumstances you may see other bee pests. The theory has to prepare anyone anywhere for possible situations.

And, no, an oscilloscope and a square wave generator at the input will NOT tell you if the power supply is ringing. It is not simple to detect this. There are methods, but they're not simple. Me, I avoid the whole thing by using simple capacitive filters and amplifiers that are well designed and that have good PSRR.

regards, Keith

This discussion started off as someone who wanted to know about the use of inductors in power supplies for solid state amplifiers. My advice is to forget it. He may want to try his idea out. Fine, that's his decision. There let it lie.
 
William,
Go for it!
Some of the neatest things I've learned came about as accidents (yes, I've fried a few parts--who cares--we're not talking about living creatures that feel pain, just little pieces of metal and plastic). Theory is nice, but can paralyze you, because you're afraid to try anything, especially if you've never done it before. Theory says bumblebees can't fly, yet they do. Guess who's right, the bumblebee or the theorist. Poor l'il critter doesn't have enough brains to know that what he's doing is impossible, so he just goes and does it. Bravo, little bumble!
Should life deal you a wild card and your power supply rings...add more inductance, take some out, add a resistor, add some capacitance. <i>Play</i> with it. The operative word here is play. This is supposed to be fun. If you aren't having fun, take up bird watching or fishing or bowling...find something that does suit you.
Keith,
Next time I see Santa, I'll ask him to bring you a new oscilloscope next Christmas. Obviously, yours isn't up to the task.
In the meantime, I hear that there's a Burger King in Des Moines with a job opening.

Grey
 
William
I agree with Grollins, go for it.
High Q for the PI filter, sure. But wat is regarded as high?
Don't waste time on the type of discussion like the glass is half full, the glass is half empty.
Build the thing, and do experiments. If you like a sneak preview (simulation) you can use a very nice little program from j.d.duncan called Power Supply Designer II (PSUD II).

http://www.duncanamps.com/psud2/

I did build my powersupplys for an ALEPH, HIRAGA, MONSTRE and JLH's always using a PI filter. They work like a charm.
After building the supplys I checked with PSUD and the results are very similar to the actual thing.

regards
 
jam said:
Maybe Keith is saying that pi filtring works better on class A circuits where current draw is constant and not as well ia class B circuits.

Any comments Keith?

Cheers
Jam

Well, it's nearly midnight and the entry to a new year. Off to see the fireworks shortly.

In this discussion, I've been trying to say several things. Possibly summarise as follows:
1. Power supply design is not a simple straightforward task.
2. If the current is constant, then ringing won't be excited. Jam is right and it's one of the things I've been trying to say.
3. Give it a bash. Don't assume however that a change is an improvement. It may not be.
4. If you want to modify power supplies, look at it this way. Would you modify the circuitry of an amplifier? If the answer is no, then remember that the power supply is part of that amplifier and has design criteria with which you're mucking about.
5. If you want to design something, first learn the theory. It gives you the background that you can use to make an effective job of the design. It introduces all the compromises that must be mad and tells you the basis of what you're doing.
6. By all means experiment. But do it from a background of at least some knowledge. That way you may advance the art.

And that's my last word on the subject ( I can see Grey wiping his brow and saying "phew!").

'appy New Year.
Keith
 
I've been reading the thread so far with much interest. Got to try the PI-filter some day...

I have a question though that is a bit of a jump to side but related to the discussion.

I was reading the "Rectifier Application Handbook" the other day and started thinking about LC filters. If I understood the theory correctly there is critical value of L with this configuration beyond which there will be current flowing through every diode in bridge at all times. Then this should be very desirable as the usual capacitive filter cause the diodes to only pass sharp waveforms of current and thus give unwanted impurities.....

Has anyone played with that idea and have some real life experience. I quess this is also only applicable with quite high and constant current demand as smaller currents require too large values of inductance.
 
I have to agree with Keith this time. heh!

I have never had much luck with pi filters with class b or ab circuits due the changing current demands of these designs, at best they did not hurt the sound or helped little but most of the time they just screwed up the sound of the amp.

The pi filter works well with class a amplifier designs including preamps, tuners and dac's, the key being constant current demand.

Cheers,
Jam


P.S. Sorry Keith I will not be able to make your blood pressure rise this time, by the way how were the fireworks?
 
diyAudio Editor
Joined 2001
Paid Member
remote caps

My Son o Zen will have separate pi filter power supplies.
I was thinking of having the second bank of caps in the pi filter
on board the amp instead of with the rest of the power supply
I figure that this way any very short term fluctuatuations in the average demand on the supply won't have to travel through the umbilical. Also it might filter out any noise the umbilical picks up.
Probably I'll add some film caps there later if the thing turns out OK. Any comments on this? Are both banks supposed to be similar in Mfd or can one be a lot larger?
 
Mark,
Having the second leg of the PI onboard is an excellent idea. If you're going to have cap banks of different sizes, make the second one (the one closest to the circuitry) the larger one.
I'm a big proponent of film caps, but have to say that the SOZ will give you less bang for your buck in that department than others. Why? Because the current draw from a differential is nearly pure DC (assuming well-matched devices and a balanced signal). I'd say to bypass the electrolytics with...oh, I dunno...say, .5uF to 1uF of film caps and let it go. More than that won't hurt, of course, but you're not going to get all that much sonic benefit from it.

Grey
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.