no compromise 845 amp. Introduction After many projects built with 2A3 and 300B I have decided to try the 845 tubes.
My delay in starting with this tube has been the fear of using a high power supply voltage.
Playing with high voltage is dangerous and any test with the amp. switched on must be done with great attention.
Also, if you switch-off the amp. to change something before to put your hand inside it is necessary to test the voltage on power supply capacitors.
As we will see later I have inserted some resistors to discharge these capacitors after the switch off.
Only once have I got a 700V shock during this test and it wasn’t a good experience.
With this introduction the consequence is that I don’t reccomend this project to any people with low experience with tube amplifier.
The main advantage of using the 845 is the high power that can be obtained with only one output tube.
In fact this amplifier can give a class A1 power about 15.6W and more than 20W in A2.
For the less expert, who I hope are reading this article only for curiosity, I should explain that the class A1 means that the output tube works always in the negative grid area.
If the tube goes in the positive grid area this will start to get current from the grid and the driver must provide this.
Schematic All my amplifiers have enough sensibility to be driven directly from a CD player.
This because I don’t think that the active pre-amp. is a good thing in a system and I suggest to use it only if you have vinyl player.
My idea is to remove any component not necessary on the signal flux from the DAC to the loudspeakers.
Some of you I hope, have read my article about DAC improvement where all the output stages are skipped using one transformer.
For this recison I am using 3 stages all with a good voltage gain and the driver of the 845 must also provide enough current to have a reserve of power to allow the 845 to go in class A2.
The choice of the first stage has been very easy because the 6SL7, tested many times, gives a good sound and enough voltage gain (about 50).
The decision on the second stage has been more complicated and my idea was to not use an interstage capacitor between the driver and the 845 grid; because in this position the value of the capacity should be high, 2uF minimum, to allow good a low frequency cut-off also when the output tube goes in class A2.
In fact in class A2 region the 845 grid as mentioned before get current and so presents an input resistance that can be easy calculated with Rgrid = Vgrid / Igrid where Vgrid could be 100V and Igrid = 20mA so Rgrid = 5Kohm that means a capacitor of 3uF must be used to have 10Hz(-3db) calculated with C = 1/ (2 * pi * Rgrid * Ft).
To avoid the interstage capacitor the solution are only two: direct coupled or interstage transformers.
The directly coupled is complex and it needs a second power supply with high sound quality characteristics like the mains.
Some people could think of using, in these cases, the negative power supply of bias but this is always made with solid state diodes instead of vacuum diodes and also with resistor cells instead of inductors.
The use of the transformer in this point avoids any problem and only the economic aspect could be compromise but in any case if you want a good capacitor like the Jensen copper film paper in oil with the high capacity value requested the price will still be high.
We must not forget in any case that we are building a no compromise amp. so the slight difference in cost cannot be important.
For this purpose the interstage transformer used is the LL1671 designed by Lundahl with a low resistance and good output swing.
Instead of other commercial products any Lundahl transformer could be asked for the desidered gap given many possible of choice of tubes.
Always for the less expert I explain that the gap is a tiny air-space at the core of the transformers used in single ended design where the core is cross by DC current to prevent saturation of the core with consequent creation of distortion.
Of course changing the gap changes also the inductance value and more gap means more DC current but less inductance.
Possible drive tubes with a enough current to drive an 845 could be 2A3, 300B or pentodes like EL34, KT88, 6550 used in triode connection.
I have excluded the pentodes that are less linear and the 2A3 because at 50mA the inductance of the LL1671 was too low to be driven by this tube (too high low frequency cut-off).
The 300B is an ideal choice with an internal resistance of about 700ohm and with the inductance of the LL1671/50mA transformer (the version with a gap for a 50mA current) about 21H we have an ft = 700 / (2 * pi * 21H) = 5,3Hz (-3db), also, if the internal resistance of 300B is higher than 1000ohm, the low frequency cut-off will be good.
Buying a 300B might seen like simple but this is not true because there are lot of differences in the sound, probably more than any other tube, a 300B Svetlana is so different from a 300B Golden Dragon that it sound like another tube.
My ideal 300B until now is the 4300B Golden Dragon, which is very linear, without distortion and with a good sound throughout the frequency range.
It is an expensive tube but a good investment also for other projects, I only consider the KR300BXLS to be better, but this is another kind type of tube.
Also the output transformer is a Lundahl product and the model to use is the LL1620/90mA with the PCB_B that helps to connect the 8 secondaries to create the output terminals.
A strong parameter that I consider in my design is the damping factor that depends on the output resistance of the amplifier and this is too often forgotten.
The damping factor determines the good frequency response and the good bass frequency reproduction without tailing.
The LL1620/90mA with PBC_B has a turn ratio of 38 and with the internal resistance of the 845 of about 1700ohm means an output resistance of 1700 / (38 * 38) = 1.2ohm that gives a sure result driving any loudspeakers.
The Power supply In this amplifier I have tried to integrate the maximum sonic performance of power supply design.
In fact both the anodic power supplies are without input capacitor (after diodes) to prevent the generation of current peak on diodes.
The diodes in this situation are subjected to only by a constant current equal to the necessity of the audio circuit without any spurious peaks.
The vacuum diodes are the 6D22 Sveltana that are an indirect heating type allowing a low voltage rise.
In a no compromise design like this the capacitors can be only paper in oil type and in fact I have used the excallent Jensen double metallized paper impregnated with mineral oil.
These very big size capacitors give the best sound I have heard and I consider these superior to ELNA Cerafine and other old stock paper in oil.
The power supply to fix the bias is quite simple but there are some tricks like the high current used to be sure of keeping the bias point fixed also when the grid of output tube start to get current going in to class A2.
There is a resistor (first 820ohm) after the diodes bridge to reduce the generation of current peak and another after to create a cell to reduce the ripple.
To set the 845 bias point (about 120V) the very good Spectrol precision wire would multi-turn potentiometers model 534 has been used.
The components About the tubes and transformers to use I have written before in the other sections but also these other components are important to get the best sonic performances:
About the passive components types, for me there isn’t much choice: Caddock resistor (MK132 0.75W) on anode, grid and cathode.
The interstage capacitors after the first stage should be Jensen Copper film paper in oil and must be quite far from the heat of the tubes.
About wire forget the Teflon! Use only stranded tinned copper wire with pvc insulator.
A very good quality stranded tinned copper wire could be buy directly from E-Z-HOOK that carries an extensive line of fine stranded and extra flexible wire.
I have used two toroidal transformers got from Studio-M and created to my specification for this project by Antrim.
They are now installed in my amp. but I don’t like these because they produce a lot of heat and some noise.
It’s incredible that we can find good audio transformers but factories are not able to produce a good power supply transformer.
The only good one was that made for an OTL on my strong specifications, giving to the factory the wire section, the number of turns and the core to use.
All the products, commercial and custom, have high electromagnetic flux and the transformer begins to be just a noisy stove.
A good power supply transformer should have a flux of 30% less than that suggested on the core’s datasheets so we must go from 17000 to 13000 or less for the toroidals.
Also the first two LL1638 on the 900V give some noise but this is because they should work with maximum 300V each and so must support 450V (half of the 900V).
The solution here could be to add another LL1638 in series with the others because is not easy to find an inductance that works in these conditions.
Startup This instruction must be followed the first time to prevent damage of the amplifier.
The power supply section cannot be switched on without the output tubes because the inductive power supply gives more output voltage than the requested when there is no load.
The formula to calculate the output voltage of this kind of power supply is V = (Vac * 1.41 * 2) / pi = (1000 * 1.41 * 2) / 3.14 = 900V but if there is no load the output will be V = Vac * 1.41 = 1410V which can destroy the capacitors.
So, these are the correct steps:
power on the amplifier without any tubes (not even the 6D22)
check the voltage on the pins of vacuum diode sockets 1000V and 400V
check the voltage on the socket pin relative to the grid of 300B, it should be about 70V
check the voltage on the socket pin relative to the grid of 845, it should be about 120V