General Power Supply Questions

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

I'm working on a McIntosh C24 pre amp for me.
The power supply is 250V with 20 ma.
It is at the bottom of the attached schematic.

I want to add capacitance and Pi filters to really lower the noise floor.

So I was going to add and replace the existing caps.
1. Change the diodes to the Mur60s.
2. Place film caps around them at .01uf 600V
3. Remove the c43A/C43B dual ceramic.
4. Replace C44 with a CLC filter of
330uf/400V cap + small choke + 330Uf/400V cap
5. R97 W Metal oxide
6. C45/A R99 C45B combo w/ 120Uf/450v cap Metal Oxide, 120uf/450 cap
7. R98 a metal oxide same.

Then I have some other questions regarding feeding the Base of Q102
and the c45C along with the 75V zeener.

Can I make a little Pi filter of C45 with CLC
and add a lot of capacitance? Will it affect the zeener?
or is it better to just add a single cap but with many uf?
200uf, 470uf, 1000uf, 2000uf, or 10,000 uf?
Do the zeeners care?

Then athere is the c45d and c46.
better to add more ufs there? Also
8. Make another Pi filter at c46? at the 10V
supply. I think I have some 470uf, - 20,000uf caps.

I have experience with tube amps but not with
transistors and what makes them sound goood
nor do I have experience with zeners and other
basic SS stuff.

I have a lot of caps that I can use along with bypassing
them with films or PIOs.

I want a rock solid power supply so that when I listen to it first,
then start to lower the niose floor and change the coupling
and other caps along with new resistors and transistors for it.

But I want to listen to it after making the P/S rock solid.

Any thoughts?

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
The thing which can make or break a power supply is correct grounding. Far more important than fancy components or overkill design, assuming the basic design is sound. The circuit diagram you posted does not show the grounding arrangements so it is impossible to tell if they are correct.

What are your ripple calculations? Where does the existing circuit need improvement? You need to understand its current operation in order to determine how to improve it.
 
The thing which can make or break a power supply is correct grounding. Far more important than fancy components or overkill design, assuming the basic design is sound. The circuit diagram you posted does not show the grounding arrangements so it is impossible to tell if they are correct.

What are your ripple calculations? Where does the existing circuit need improvement? You need to understand its current operation in order to determine how to improve it.

Here is a pic of the Power Supply Caps.

An externally hosted image should be here but it was not working when we last tested it.


Then, here is the PS board and low level transistgors which feed the
preamp circuit.

The power supply feeds from the right side on the board where the thermally heat
sink is located.

An externally hosted image should be here but it was not working when we last tested it.



Here is the overhead fiew of the preamp.

An externally hosted image should be here but it was not working when we last tested it.


Ripple TBD.

Cheers,

Sync
 
Last edited:
Are you sure that the noise is ONLY generated by power supply, and NOT by other source, like tubes/transistor, shooting noise in resistors, bad shieldings, etc.? If not sure, you will work uselessly.

No, I'm not. But the caps were bad and physically leaky. So as long as
I'm in here, I might as well make the P/S rock solid.

The grounding scheme isn't so go as it is that little tab on the multisection
cap. Thinking of making a grounding bus bar from solid Romex 14 awg wire.

Trying to do a little spice model of it, but now realize that I have to
define all the componants into the superspice program. I was just
looking for something simple to model it but I guess it would be
simpler to just rebuild the circuit.

However I did find some other model schematics where McIntosh did
upgrade the power supply. They are shown here.

C26 Power Supply Schematic:

An externally hosted image should be here but it was not working when we last tested it.



C28 Power Supply Schematic:

An externally hosted image should be here but it was not working when we last tested it.
 
Ok, but I think that given you want to manipulate the circuit and do lots of mods, I would start by eliminate the simple pass transistor, and make regulated both supplies, using IC´s or discrete components. And respect to the grounding scheme, surely your idea is good. But, i first would try to get working the equipment as is, and then do all mods.
 
Q302 may be deleted, and directly replaced by a LM317 if you want exactly 14V, or a easiest LM7815.
D304 in the 100V supply may be wired in series with a TL431, and then you must add the feed back resistors, plus (eventually) a cap between anode and ref pins if it oscillates. The zener voltage must be lowered, by the fact that TL431 only supports 36V maximum, so putting a zener of 24V less than the original, the resting 24V will be easily managed by TL431. And this way, the PS becomes lower noise and lower output impedance than actually. Only my suggestions, may be hundreds of different opinions.

Good luck.
 
Ripple Calc at the first filter:

V = It/C

60 Hz
time = 0.015s

20mA from schematic/Power Supply

470uf cap = .000470

It = 0.015 * .020
------------------------ = .64V (Vripple)
.000470

V ripple
330uf cap = 0.91V
200uf cap = 1.5V
100uf cap = 3.0V
68 uf cap = 4.41v
47 uf cap = 6.38v
22 uf cap = 13.64v

cheers,
 
Thinking along these lines...

inout_ps_layout_web - My Photo Gallery

First sentence is true. Second sentence is false, as it guarantees the very thing you should be avoiding by injecting charging pulses into the star ground.

The return path for the first cap should not go to the star point but to the bridge/transformer. The return path for the second cap can go to the star point. Then link the ground sides of the two caps together. The result is that the star ground only gets a little smooth ripple current into it instead of big narrow charging pulses.

I am puzzled why people keep making this mistake. There must be a website somewhere which encourages it.
]

That said in my little scheme of things I would do the following:

from the PS with parts on hand put in a 330uf 400V cap,
reference that back to CT AC ground, the run the a little coilcraft 12Uh/3.8A inducter series connected to 68uf/450 then s.sK in series with a 1000/200V cap and the rest of the schematic.

To be continued...
 
Larger capacitors will result in less ripple but will also mean higher (and of shorter duration) current pulses through the diodes and transformer.

Less duration / higher levels of current pulses mean higher harmonic content.

You would be reducing delta V issues for delta I issues.

If there is headroom in the regulator then use it.

One method to reduce noise in rectified supplies is to introduce series resistance in the diode path. This will increase the "filtering" the regulator has to work with but will reduce the harmonics of 120 Hz that are being generated.

Less ripple also will result in greater power dissipated in the regulation device
 
Larger capacitors will result in less ripple but will also mean higher (and of shorter duration) current pulses through the diodes and transformer.

Less duration / higher levels of current pulses mean higher harmonic content.

You would be reducing delta V issues for delta I issues.

If there is headroom in the regulator then use it.

One method to reduce noise in rectified supplies is to introduce series resistance in the diode path. This will increase the "filtering" the regulator has to work with but will reduce the harmonics of 120 Hz that are being generated.

Less ripple also will result in greater power dissipated in the regulation device

DUG, Okay, so if I go with the MUR850s, they are
plenty fast and it will have plenty of head room.
Now I understand why the UF diodes sound better
than slow diodes--Less duration, high current= harmonic content.

As DF96 menntioned, then with the first filter's ground,
I'll connect that to the bridge center tap.
I recall a buddy of mind always connecting
everything to a star ground but he always
had 120 Hz problems in his builds.

Thinking about some things I've read from Dave Funk,
discussing grounding and how you want each functional
section to be its own little grounding island connected via a small
wire to the other islands to reduce noise.

What if all these islands are connected through a 100 ohm reference?

Maybe it is simpler to place the AC Ground reference at the bridge
center tap at 100 ohms and leave the regulated grounds in their
own planes at chassis potential?

I'll create a 12-14 AWG bus wire for the remaining power
filter grounds and tie that to the first CT ground though a
smallish 22 - 26 AWG wire with a100 ohm resistor, lthen
ifting that supply above the CT ground also.

Just thinking out loud here.

Thinking then I can implement the C-L-C-R-C-R-C filtering.

Then when I get to the different voltages supplies and use the
componants that Osvaldo suggested.

Osvaldo,

I'm thinkg that for the 14 supply to use what you recommend,
that is the LM317 or LM7815.

Then on the 100V supply to the using the zener in series to the
TL431...(24V less in the zener). I think that was a 75 V zener.
I guess I would have to go with a zener 40 volts less?

Becaseu the TL431 only supports being fed 36Vs and it will
make up the difference of the zener?

I what I think I hear and understnad.

I do undestand that I'd have a much less noisy power supply.

Osvaldo, I've never done this before so I hope you don't
mind helping out on occasian.

Gentlemen, thank you all, if iyu have any other comments
thoughts along the way of If I'm not thinking correctly
about a comment, please speak out to I get my thinking
correctly.

Cheers,
 
Last edited:
After the Power Supply

Then once I get the power supply issued solved and built
and tested, I'll then move to the two componant boards
in the pre amp.

The McIntosh C24 has two boards that shape the tone for the
preamp and also have some very old sounding SS transistors
driving it. I am told by those whove gone before me to replace
these transistors with modern ones as material sciency has come
a long way since 1964.

They couldn't believe that materials science has made these premaps
even better than new.

I want one...

I have one...

I want it to sound great but not like early 60s Japanes gear.

cheers,


Spike
 
DUG,

Yep always running before walking that is why I fall down, but I get back up
and keep trying.

I'll hold off on the 100 ohm ground reference.

Instead, I'll include a couple of pages that I found specifically
what you discussed about the delta V delta I changes. And,
going through the work a little bit more closely I definitely
think I was about to over cap the pre amp.

Reconsidering what is happing within the preamp I bet substantial
gains can be had by installing a full wave bridge, instead of the
half wave CT.

I've got the little MDA2501 FW Bridges that I can mount where the
two diode tabs are located. This should provide plenty of harmonic
conten enhancing the performance of the preamp all the way through.

So I should have ample headroom for the CLC supplies.

I'll just take it a little slower and walk my way through this.

Interestingly, KOC discusses using a reverse mounted zener
at the CT of the Full Wave rectifier making the AC mains reverence
negative in value.


The references: The Ultimate Tone volume 2: System Approach to Stage
Sound Nirvana.
Kevin O'Connor, Power Press Publishing, London, Canada.
1997.

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
There is a minor mistake in that. He adds XC and XL in the denominator when calculating ripple reduction. He ought to subtract, as they have opposite signs. In reality it is usually good enough to ignore XC in the denominator, as component values may only be known to 20% anyway. Of course, when an author gets simple stuff wrong it raises questions about more complex stuff.

I assume those two pages are not a copyright infringment? The mods may remove them if they are.
 
Nope, it shouldn't be a probem as two pages from a multi hundred page work.
I think you can get more on amazon.... but this is for knowldedg, not for hire and
I"m summarixing the authors work.

APA's guideline is 3 pps of journal article more
of published work. The key is not to claim it as your own and there is full
citation by me of who the author is.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.