Platenspeler.Com |
 |
Platenspeler.Com |
|
Tube amplifier El Dorado
Page 1: Designing an EL-84 amp from scratch (draft)
Last updated: Monday March 8, 2010
This project started after I bought myself a cheap Chinese EL-84 amp from Suppo Audio. The amp was cheap and did offer some opportunities for modification, but I figured out that after a certain amount of "improvements" the only thing sensible to do was start from scratch and build my own amplifier with the same tubes and the same basic setup but with better components and a slightly improved design.
So why building an amplifier based on EL-84 tubes? I do not know to be honest. I've been impressed with the El-Cheapo amp and I'm curious whether there is still room for improvements. On the other hand working with a new tube or so would be nice as well and the output power of EL84 is not enough for most of my speakers.
Because I always want to have some spare tubes for my working DIY amps, I do have a few 6N1P tubes available for this project. And looking at the rest of the components the total price of components will not break the bank either. So the main investment for this project needs to be made in output transformers.
Well let's assume for the moment we use "standard" 8k output transformers (with UL tap), and we bias the tubes around 30mA each at 275 Volts B+.
In my timezone, shipping iron by post is highly expensive and in general not a good idea unless you work for TNT post. And whether you ship in side Europe or import from other countries often doesn't make much of a difference anyway. So I'm thinking to buy a pair of Edcor output transformers which unfortunately need to come from the USA. Getting transformers directly from China is something that I do not trust at the moment: quality control etc. is often sacrified in favor or making money fast.
Ths following Edcor output transformer seems to be fit for the job:
There are some options for the splitter in in a EL84(-like) PP amp. The circuit will in most cases be quite simple, as the EL84 tubes are not too difficult to work with and as we want a simple yet good amp we look for such a setup.
One of the most used input/splitter configurations is the concertina (in Europe called Kathodyn). In this setup, the first tube halve amplifies the signal and the second halve of the tube is the splitter.
The advantage of this circuit is the very simple setup, where the second tube halve is always correctly setup. The disadvantage is that the output impedance of the top output is much (!) higher than the output impedance of the lower part (cathode follower). The amplification of the second stage is close to 1, the amplification of the input tube V1b is calculated as follows (common cathode output): A = - u * Ra / (Ra + Ri + (u+1) * Rk) = 33 * (47k + 4400 + (34 * 47k )) = 33 * ( 34 * 47k + 4400) = 0.94
For the V1b cathode follower output the fomula is: A = u * Rk / (Ri + Ra + (u+1) * Rk) and as Ra == Rk for the splitter, the outcome is the same as above: 0.94
The amplification of V1a input stage is calculated as: A = - 33 * (220k + 4400 + (33 + 1) * 2k ) = - 24.83 times (about 27.9 dB)
The Suppo amp when I bought it did have this phase splitter setup, and it worked very well.
The other type of phase splitter is the well known long-tail type.
Gain of the long tail is calculated withthe following formula: A = u * Ra / (Ra + Ri + (u + 1) * Rk), if Rk is bypassed with a large capacitor use 0 value for Rk. However in this case the formula does not really help as we are auto biasing the tube with a solid state component. And especially not a dynamic one acting as a constant current souce of around 10 mA per channel (20 mA in total per tube).
Comparing this to the "normal" setup of the Common Cathode Amplifier with a cathode resistor we would need -4 V for biasing the 6N1P to 10mA (at 250V), so normally we would have needed a resistor of 4 / 0.010 = 400 Ohms. Calculating with such value for a replacement of the CCS, our amplification is about: 33 * 22k / (22k + 4k4 + 34 * 400 ) = 19 times.
More gain is possible, but will require a far larger value for the plate resistors resulting in a lower current also.
One of the advantages of the long tail setup is that it's quite easy to use a CCS setup since both tubes are supposed to have the same bias. However, the concertina doesn't even need this as long as the output impedance and grid voltage stays withing acceptable limits. For the EL84 concertina looks quite good. For a EL34 I might choose a long tail setup.
And I like to use ca CCS cathode bias of 30mA per tube in order to keep the bias set correctly at all times.
Assuming 30mA bias per tube, then the resistor R can be computed as follows: R = 1.25 / 0.030 = 41.67 Ohms. So rounding up to the nearest value of R (in the E92 range) we either use 39 or 47 Ohms, or we use two resistors parallel of 82 Ohms yielding 41 Ohms nett which will result in a bias current of Ib = 1.25 / 41 = 30.49 mA. Typical operation in triode mode is 34 mA, but I'd like to stay below that value, so this must be OK for the moment.
The power supply for EL84 amps is often very simple, and we rely on PSRR of the Push-Pull circuit to eliminate for very crude power supply. One way to get a better power supply for a power amp is to get it a capacitor multiplier (sometimes called an electronic choke) or gyrator.
The power transformer should be able to deliver 230-250 Vac @ 150mA minumum (4*30 mA + 4*8mA), and for the heaters 6.3Vac @ 4* 760mA + 2* 600 mA = 4.24 A
The most simple version has a reference to a specific voltage, mostly by using a zener diode with a sufficient large value (say 270 Volts). Picture would look something like this ...
It's also possible to build a version for which the FET is floating wrt to ground and is given a reference to the B+ rail coming from the power supply. I found the idea for this setup in an on-line article of audioexpress (which was the simplest) and several on diyaudio.com.
Note: Be careful when building these Gyrator circuits above until I've tested them myself....
I do have lots of criticism on the Suppo amp wrt safety and construction quality. But the schematic and values used for the components seems to be usable. That's whay I started with something I knew and could refer to and tried to improve from there.
The schematic is still based around a Concertina splitter and that works quite well.
| << Page 1: Introduction |
^^ Back to DIY Page ^^
|