Ideas for a nice sounding power supply

[Bardamu]

Hello,
I am eduard from the Netherlands . a long time audiophile but lacking skills to predict the outcome of connecting a power transformer, a rectifier tube , some chokes and some capacitors. So i just have to connect and listen.
There are however some things that are usually confirmed by things i built so far or to which i will stick to just for safety.
I will write them down randomly.
Untill now no matter if it is a tube design with low current or a solid state that takes 1A choke input with the right numbers of henry and a proper load resistor always is a good start.
Differences between chokes used for choke input are big. Not only comparing one brand to another but also going up in Henry . There are brands where you can choose for a smaller airgap so more H but less current capabilty but same DCR and the bigger H will sound better.
A transformer with a lower current rating and a choke with more H will be better than a bigger transformer with a smaller choke.
I have been building choke input supplies without calculations ( just did the load resistor and critical inductance) and start looking at PSUD some months ago.
I have been told that behaviour during start up isnt that important( unless you overload the rectifier) BUT the important thing to look for is how the last stage reacts to a stepped load.
It appears that using high henry chokes sometimes forces you to use more mF to stop strange behaviour. One would think that using more Henry ( i start doing this to be able to just need a very tiny amount of current for the load of the resistor compared to the circuit load) would make it possible to use smaller mF caps.
In the supply there are caps whose main task is to filter and there are the ones that act as a reservoir. Some people say in order for the supply to be quick the caps must be low value. So quick discharge and recharge. But if the last stage of the supply has serious mF the amount of energy available is bigger so the recharging could be a bit slower or not?
I start collecting some caps that i am going to use in a 620 volts dc that needs to supply maximum 70 mA to a stereo DRD with Vt25A tubes that will just be used for 600 cycles/hertz ( 12 db filter at line level)
I have Eyuda caps containing oil, Epcos MKV, some CSI 10mF ( which are made for the military but i know nothing about them) and some Russian 50mF
caps which are similar to the k75-10/k75-24 which are on the way to me. I have send one Eyuda to a friend and he tried them as the reservoir cap with different by pass caps so far K75 gave the best results.
It could well be possible that as the first cap after the input choke another cap could be better.
At first i wanted to use an LCLC supply and after the last C give each channel its own LC stage. After rereading Jean Hiraga articles from the eighties where he wrote if that makes a difference than the original supply wasnt good enough. His advice then was to add more caps ( it was for a phono preamp) not improve current capability but to reduce ESR ( of course they are related)
If i would go for LCLCLC i could go different caps for each stage. The Epcos and The Russian are both a mix of paper and some plastic foil. The MKV contain like a spoon of oil. The K75 are more like impregnated with oil not like a bathtub.
Any ideas or suggestions are welcome.
Thanks in advance and greetings, Eduard

[Paul Birkeland]

There are a very small handful of people online (fewer than 5) who fret endlessly about step response and drone on and on about how the power supply needs to be rock solid when simulated.

The natural solution to satisfying these simulations would be to shunt regulate the high voltage power supply, but this particular group of people will denounce that for a variety of unspecific reasons.  While they will use step response to justify how "good sounding" their amplifiers are, they are somehow also able to denounce an amplifier with perfect step response as being flawed.  A shunt regulator would have no step provided the regulator shunted more current than the step, which is not at all hard to do. 

This same group of people also tend to suggest much lower operating current than the book values for "long life" of output tubes, but they don't adjust the output transformer primary impedance.  This will create a large difference between the quiescent plate current and the average plate current, which will either require a stiffly regulated power supply or a power supply that would otherwise be designed for a class AB push-pull amp.  This also decreases output power substantially and nudges up THD. 

None of this has had any influence over our kit designs.  We find the low DCR power transformers espoused to be rather magnetically noisy, we design our output stages to avoid the step in the first place, and we pay attention to properly driving output stage Miller capacitance. 

If this path of DIY interests you, you should certainly not let me stop you from pursuing it. 

(You are also correct that the power supply requirements for a phono preamp are significantly different than those for a power amp.)

[Bardamu]

Hello Paul and all the others as well,
I have been building choke input supplies for 30 years now. Not a new one every year but just 3 or 4 that were sometimes adjusted by using other choke brands and/or values. Sometimes caps would be changed too.
The first choke input was for a French phono amp named Anzai and it was without any active regulation.
What io have build later was always used with a shunt regulation designed by the late Allen Wright. He always wrote that he likes his regulation used in combination with a supply where the choke is bigger than the transformer.
One would think that the shunt will be the main element that influences the sound but what is done as a kind of '' prerugulation '' also has great impact.

Some time ago i bought a phonostage build by a Dutch designer who uses choke input plus a tube shunt supply. STILL have to connect it because my turntable is on the modification table.
I asked him to make the same supply for my tube line preamp.

I have built long time without doing any measurements ( only the ones to keep the parts from being overloaded). Also no simulations.
Sometimes you end up using caps made 35 years ago and they sound better than caps with better specs made last year.
BUT when i was twenty i could buy a bicycle which would cost like a one month salary and cycle more than 100000 kilometers with it without any big repairs. after 100000 km the shop would say get a new one regularly. In the end i have done that after 14000 kilometers i was glad it got stolen. In the first two years i had more repairs than with the other one during its whole life.
Lots of the caps nowadays focus on things that might look nice but are not that important at all.
It seems that the Russian caps that appear on ebay are sometimes really good. I have bought many so called audiophile caps in the past. Some where nice but most were mediocre.
There are people who tell you to use the higher voltage models. I dont know if tecnically you can use one with the seize of your thumb why use one that can take 3 times the voltage with the seize of a coke can?

YES, it is true with an ultrapath cap you will be moving the cap choice problem from one place to the other place.
With the ultrapath your power supply caps need to be good because with bad filtering the ultrapath cap will worsen the problem.
In my circuit can i just skip that cap and introduce the cap at the cathode?  Just connect a cap across the 649 ohm resistor. With 600 hertz 12 db input filter could it be smaller? Which value for full range and 600 hertz up and what about working voltage.?
Greetings, Eduard.
P.s i know Jean Hiraga  use a small cap in the position of the ultrapath cap and also use the cathode cap. I thought it was someting like 10 or 20 mF

[Paul Birkeland]

My recollection is that Allen did a lot of class AB push-pull amps, so the regulation afforded by a choke input filter becomes more helpful/relevant.

For your circuit, the cathode bypass cap is a shelving type filter, so I wouldn't skimp on the value there.  You could certainly have two line level inputs, with one going through a passive high pass crossover to take some of the LF content out.

[Bardamu]

Hello,
The idea about filter between line amp and power amp came from France who did copy it from some audio magazine from the USA.  If i remember well it was also discussed in Sound Practices.
It needed a line stage with low output impedance ( why i have one with Tango line stage output transformer) and power amps with high input impedance.
My TAD2001 dont like low frquences so they like to see the 600 hertz be cut at 12db.

In all my projects so far adding choke input gave me better results.

It could take some time before i finish the amp that is going to drive the bass bins so i could use the VT25A for the complete loudspeakers.

Greetings, Eduard