Platenspeler.Com |
 |
Platenspeler.Com |
|
After building 4 phono pre-amps over the last two years, I decided it was time to build a power supply for my last project Loekie that wasn't just another straigthforward power supply.
On the internet, there are several places where one can find a power supply design for a phono- preamplifier. However, most are as straightforward as my own, consisting of a transformer, some diodes, a few resistors and capacitors for pi-filters and maybe a choke.
But, especially with phono preamps, a ripple in the B+ can ruin the sound, and humm introduced by components or layout is difficult to fight. Therefore, I want to build a power supply with voltage regulation or a current source that takes care of these problems. Also, I want to hear whether a regulated power supply sounds better or maybe less musical than a non-regulated one.
The following is what I'm looking for in a new power supply for my Loekie phono equaliser:
That's it for the moment, I will add to this list when necessary
There are without doubt numerous ways to build a regulated power supply for a tube amp. Let me summarize below what I think to be usefull ideas:
This is the most simple design for a phono equaliser. Since such a small amp draws only 1.5mA for each channel, there is almost no need to make any kind of regulation.
But, since I want to do something special for the next power supply, I'll try to squeeze some extra components in the small housing. Because the amplifier uses either a SRPP design (Tweety) or a mu-Follower (Loekie) I had to lift the heater supply. R4/R5 is a voltage divider which brings the heater supply to V=R5/(R4+R5)*Vb2 = 100/570*280=50 Volts. This makes the Cathode-Heater voltage of the SRPP tube less critical (170 Volts max). For AC, the capacitor C8 makes both grounds equal.
A lot of regulated power supplies rely on components that are put in serial with the load (= the amplifier). These can be FET's, tubes or other components. In his book Mr. zur Linde describes a design for a regulated power supply based on two halves of a ECC81. Because of the serial design, the tubes need to be able to supply enough current to the amplifier itself. Fortunately, an ECC81 is capable of delivering 9mA which is sufficient for both channels of my phono amps.
A picture is shown above. Simulations in EWB show that this design should work fine. The power supply is regulated between 250-300Volts and ripple rejection to about 1mV.
There is a major disadvantage in using this setup: The Cathode of the lower tube is stabilised on 94Volts and the upper halve runs with the Cathode on B+. Since I want the B+ to be in the order of 275 Volts for my Loekie pre-amp, this means that I will have to lift the heater level with 150 Volts in order to protect the Cathode/Heater insulation. However, since I use a SRPP in Tweety and a mu-Follower in Loekie 150Volts is just too much for the phono amp itself and it will mean that I have to use two different sets of heater supplies for the power supply and the amp. This is just too much of a good thing, especially for the small boxes I use for Tweety, Granny and Loekie.
A solution based on parallel load can be found both on the site of TriodeDick and AudioNote. Both make use of a tube to get a clean B+. Both version basically use the same solution called a shunt-regulator to get a clean voltage supply. TriodeDick calls it the ripplekiller and also mentiones the idea to be available for decades.
The schematics of this design are as follows (see site of Dick):
In our case, the design might need a little modification in order to be suitable for higher B+ voltage as used in SRPP and mu-follower designs (an ECC81 won't survive more than 300V over a long time).
Peter van Willenswaard wrote an article for AudioNote where he decribes modifications made to an Audio Innovations L-2 amplifier. In the article he focusses strongly on the power supply section which is built around the same idea as the ripplekiller of Dick. Only in this design an OA2 is used to get a Cathode reference on 150 Volts.
Since I cannot link with pictures directly to the AudioNote site, I copied the figure to my own site (© AudioNote)
When extending the ripplekiller of TriodeDick I found myself more or less designing the same power supply as described by Peter, after all the purpose of OA2 is to lift the cathode of the tube with 150 Volts.
Again, since I do not want to lift the heater supply too much, so that I can use the same heater supply for both the power supply and the amp, I decided that a lift of around 100 Volts would be just what I needed in order to (1) not damage the tube and (2) have a nice heater supply. As a variation of the OA2 I decided to use Zener diodes in order to get a cathode reference for our tube of about 95V. (I might even use a Neon bulb instead).
Well the picture above descibes my thoughts at a certain moment in time. When simulating this design I found that it stabilises well around 280Volts and ripples introduced are eliminated will in the tube stage as long as Capacitor C5 is at least 100uF. In cases where power supply and amplifier are built in separate cases (like mine) capacitor C6 may be placed in the amp itself.
The design differs in that of AudioNote that we use a capacitor (C4) instead of a resistor network to load the grid of the ECC83. This means that DC variations won't be eliminated, but AC, and that's what worries us most, will do just fine and we do not have to worry about the correct grid voltage since R3 and R4 take care of that (relative to the values of the Zener diodes).
Capacitor C6 should be of a high-quality type such as the Black Gate 350V/22uF. Expensive but worth it. The trimming Pot P1 must have the smallest value possible, yet be able to maximise the effect. Simulations show that a value of 2k would probably even better, please use a 1W version.
In order to test the design I did several tests in the simulator to see the effect of changing component values and how critical the design is for these values.
....
<>