power supply damping in psu d.

[Ianm2]

not much has been alluded to in this, but being a resistor, choke, and capacitor 'network', filter, etc., power supplies ring, oscillate, etc.

So I thought I would set the ball rolling,

using the diyhifi supply billie psu ( this amplifier is so good, its practically a textbook design, everything is pretty near perfect, although still room to change bits) for starters, here are the specs.:

mains tx is 415 volts, 80 ohms sec. resistance, use a 5u4 rectifier.

input resistor of 82 ohms.( you can't really alter this too much as it drops or raises voltage, but its good to play to see the effects, making it higher damps the ringing more, but you lose volts, it also delays the ramp up time before the psu 'fills up' with volts.

CLC: values, 100 uf, 80 ohm dcr choke, 8 henries, 100 uf cap.

so its a RC, LC filter with a current load of around 80 milliamps.

also notice something rather odd about this amp, in its prior incarnation, the series resistor, the mains tx secondary, the choke, and in fact the output transformer primary(thin wire, more resistance, less current, more volts than secondary), all have just about the same dc resistance, of around 80 ohms, design or accident??)

now upon startup, we get a bit of 'bounce'.

I wonder what would happen to the sound if we try to make an optimally damped power supply?

There are a few things we can tweak independantly and depandantly viz:

the series resistor ( transformer secondary is fixed)

more resistance gives more damping.

change the henries/dc R of the choke, remember this is largely fixed by the maker.

change the capacitors, better results are had with a lower input capacitor, and higher output one.

It seems as in some other designs on the web, failry good results are had with 20uf input, and 50-100uf output.

You can start 'gently' and reduce by say 20 uf each time, or simply 1/2 each time. It looks better with 50uf on the input at startup.

You want to tick the boxes to the left, either Volts in I-1 which is the current source, or volts in C2, both of which our output 300b valve will 'see' (ignoring the output tx)

I haven't the time to fiddle, but hopefully others will play a little and come up with some conclusions/answers.

Measuring resistances of chokes/trannies is easy, just a dc ohms measurement across the terminals...

Finally, anyone know how I drop in the extra rectifiers section, did it b4 but a different machine running, adn can't recall how I did it? help...

There is a schematic I have drawn, please scuse the triode as a rectifier, the prog. doens't have a diode!!

And I have managed to get an image of this on psu D, note the 'bounces' on the lhs, as of the improper damping.

[Ianm2]

here is one with better psu damping, note the changes in values...

Also, this means you can use a quality cap, perhaps motor run or oil or a solen/ansar/obliggato/ICW/LCR polyprop., which would be more ideal

I hope this will help get a few newbies into psu designer, and help me learn more, too!! I don't know that much about it.

As an addenum, if we add another LC filter stage, and make the caps. all 100uf, our 'bounce' still comes back, so we can start of from lhs to rhs with 15uf, then goto 30uf, then goto 100uf, and our bounce again vanishes.

The ideal, I believe is a fast power supply that charges up quickly, with little voltage ripple, and no mountains or spiky peaks that give oscillation. (ie a straight line up and across)

This is all very one dimensional, I haven't fathomed current or anything else yet.

Thing is, if you come to design from scratch, once newbies get the hang of it, you need to start at your power valve, the output, and work out the dc volts you want on it, and the dc or idle current, and operating point (that's load lines) and work back thro the psu, that's how designs work in reality. Take it back from where you want to be.

alt. you can just select a mains tranny, and cobble the thing together, and it will work fine, if you are careful, and adjust your output, be it a preamp or power, to suit the delievered voltages.

As a more advanced guess, if we replace the constant current source, the resistive load will be the dc load of the output transformer primary, 80ish ohms, followed by the DC RESISTANCE OF THE 300B valve at the operating point, which is 470 volts minus 70ish volts divide by 70 milliamps, ( circa 6000 ohms?)
ohms law, plus the cathode resistor of 1k ohms????

i.e everything the end of the psu sees to ground, we go via

1 the output tx primary,

2 the 300b tube,

3 and the cathode resistor.

[NickG]

Measuring resistances of chokes/trannies is easy, just a dc ohms measurement across the terminals...

But what isn't easy and makes the accurate simulation a bit more of a problem, is the actual inductance of the chokes when passing current, you have to either measure with a DC supply, or hope the manufacturer has given th ereal value.

[Ianm2]

no interest? I hoped this would be a sort of primer to get you going in psu d, are people trying it out, or is it just not of interest? It does take a bit to get it going.

[IslandPink]

"As an addenum, if we add another LC filter stage, and make the caps. all 100uf, our 'bounce' still comes back, so we can start of from lhs to rhs with 15uf, then goto 30uf, then goto 100uf, and our bounce again vanishes.

The ideal, I believe is a fast power supply that charges up quickly, with little voltage ripple, and no mountains or spiky peaks that give oscillation. (ie a straight line up and across)"

You're on the right road . I've heard the differences between indifferent & good power supply design - most recently when Dave Dove brought his amp round to compare . Get the rise time quick and direct ( no ringing ) and you get benefits in bass tone , lack of bass booming (lumpy bass ) , timing & flow . Two-L stage with lower DCR is the best way to go , but there are optimum cap values , DCR from trans secondary and DCR for choke for even CLC type supplies .
Now check out how to use the 'stepped current' option on a current load .

Mark

[Tweeker]

How to calculate LC filter Q:

Q=(1/RDC)*[sqrt(L/C)]

The damping effect of the load may be calculated and added to RDC via:

L/(C*Rload)

A Q of .5 gives critical damping and is optimal so far as ringing is concerned. A Q >.707 will cause an LF peak in the response. A Q > 2 is quite likely to be trouble and cause some ringing. Look at the voltage across L in Duncans for this, in addition to output voltage. Math courtesy Morgan Jones Valve Amplifiers.

There are a few ways of dealing with too high a Q, add more C, add more DCR (and possibly even more L at the same time), snubbers and/or zobel networks.

LF resonance and filter cutoff (-3dB) frequency is given by:

1/{2pi*[sqrt(L*C)]}

[Tweeker]

In a hifi or any SE power amplifier (PP instrument amps are another story) I favor a choke input supply with a single LC filter stage. Use a choke preferably of 2x+ critical value with the least DCR that I can afford to damp with the filter capacitance. This will typically give both very low ripple and low output impedance.

For richly biased amps with capacitor input supplies I favor: small C - L - big C.

[Tweeker]

Damping Ringing in LC Circuits

CLC Power Supply Ringing

Rectifiers and Transformers

Tube Power Supply Design Notes