My first valve amp - help appreciated :)

[Krstfrs]

Hi,

I'm new to valves and hope to get a bit of help from you guys
The goal is to design a SE amp with an EL 34, because:

- I read that the perfect Raa for pentode operation is Ua/Ia, being 2,5k at maximum power (though I'd like to lower the power a bit to lengthen the lifespan of the valve)
- I also read that the perfect Raa for triode operation is about 2 to 3 times Ri, being 1k8 to 2k7 (the Mullard datasheets are very informative)

So a single output transformer with 2k5 input and 4/8 output should be enough, although my speakers are 6 Ohm. There are no output transformers for 6 Ohm

So far I designed a preliminary pre amplification stage for line level (mainly my computer) and would greatly appreciate any comments for this:

                     +----- +250V
                     |
                    +-+
                15k | |
                    | |
                    +-+
                     |
                     +----||----->
                     |    15n
                   --+--
                  /  |  \
                 |  ---  |
    15n          |       | 1/2
>---||---+-------+-- - - |  ECC81
         |       |       |
        +-+      |   *   |
   500k | |       \     /
        | |        --+--
        +-+          |
         |           +-----
         |           |    |
         |          +-+   |
         |      270 | |  --- 220μ
         |          | |  --- 
         |          +-+   |
         |           |    |
>--------+-----------+----+---GND

- I don't know what influence the grid resistor has, so i chose 500k, which I have seen a few times in other projects and which is Rg,max/2
- The working point was chosen to be -1,5V. I don't have any Ig1 graph, so this was a rather conservative choice
- What is the influence of my choice of Ra (leading to Ia,max=16,5mA). I if understand correctly, all I need at this stage is voltage amplifiying, which seems to be about μ=30 regardless of Ia,max. What disadvantages/advantages do other choices of Ia,max have?
- Why the hell do I use a ECC81? I could also design a stage with μ=30 with the ECC83, drawing less current.
- By the way the μ=30 wasn't chosen, it was calculated after choosing the working point.
- So far no feedback is designed in, the 220μ was chosen, so that it is much less than 1/3*270 Ohm for 20Hz. 220μ/350V is damn expensive. I'd like feedback (the linearity isn't that good so far...) but found no equations on the internet.

The goal, as said, is to drive a single EL34. I would also use 250V for that, although i have 270V AC, resulting in about 380V DC from a cheap standard power transformer. The EL 34 datasheet just says 250V is typical...

The transformer has 270V 400mA, which should be enough. A cheaper model has 230V 200mA. Is it possible to limit the 2 EL34s (stereo) to about 70mA, so that this pre amp (16,5mA max) and the output stage draw just about 170mA? I have the bad feeling that i forgot the Z current and other stuff, so i think I have to use the 270V 400mA transformer and waste 130V DC?

Phew, that was much. Many thanks if you read the whole bunch and much more thanks if you answer some of my questions, I'm one of these toobz noobz

[aardvarkash10]

you are designing the wrong way around. Start at the speakers and work backward: speaker > output transformer > output stage > driver stage > input stage.

Don't worry too much about the (very) nominal impedance rating of your speakers - they vary wildly according to frequency anyway. But with any SE amp, its important how efficient your speakers - you won't be getting a huge amount of power out of this puppy! Low efficiency speakers = disappointment....

[kavermei]

Hi, just some unsorted thoughts:

- The 500K grid resistor is fine
- With 15nF and 500K, you have a -3dB point of just 21Hz, I think. You should be aware that your amplifier will have multiple such RC highpass filters in it, and if you place them all at 21Hz you may lose too much of the low-frequency audio just above 20Hz. So maybe make the 15nF capacitors just a little larger?
- A 2k5:6 transformer ratio can be obtained from a (approx) 1k9:8 or a 3k8:4 transformer
- You don't need 220u/350V for the cathode bypass capacitor. The voltage across it will not exceed Rk*Ia,max. I'd say use a 35V rating or so, which will be plenty.
- You should first determine how much voltage swing you need to drive the EL34, and what your line input level is, and determine the required mu for the preamp stage from that.
- The max. DC current you can obtain from a transformer depends on the type of smoothing filter that is used, e.g. capacitor input vs. choke input, full-wave vs. bridge rectification, ...

- Roughly design it and just build it, you can always tweak later... Sound will change with different bias currents of the EL34... If I were you, I'd go with the slightly more expensive transformer since it allows you to do more experimenting, or maybe try other tubes, or push-pull configuration later, too.

Best
Kenneth

[Merlinb]

Originally Posted by Krstfrs

I don't know what influence the grid resistor has, so i chose 500k, which I have seen a few times in other projects and which is Rg,max/2

You can safely go up to 1Meg with an ECC81, although 500k is fine.

- The working point was chosen to be -1,5V. I don't have any Ig1 graph, so this was a rather conservative choice

I would suggest you increase the bias. The ECC81 is, well, not exactly a very good audio valve unless the bias is at least -2V, because it draws a lot more grid current that most other preamp valves.

- What is the influence of my choice of Ra (leading to Ia,max=16,5mA). I if understand correctly, all I need at this stage is voltage amplifiying, which seems to be about μ=30 regardless of Ia,max. What disadvantages/advantages do other choices of Ia,max have?

Making Ra larger will reduce distortion and increase output voltage swing, and also increase gain. 15k is rather low, unless you like a lot of 2nD harmonic distortion. I would increase Ra to perhaps 33k or 47k. However, you can always adjust the gain by adjusting the amount of cathode bypassing.
Increasing Ra will also reduce your current consumption too.

- Why the hell do I use a ECC81? I could also design a stage with μ=30 with the ECC83, drawing less current.

Yes you could! SOme people here will mock the ECC83, but the fact is it is a very linear valve, and a good deal more 'hifi' than the ECC81.

I'd like feedback (the linearity isn't that good so far...) but found no equations on the internet.

Adding a small resistor in series with the cathode bypass capacitor will introduce feedback. You can adjust this to whatever you want.

[m6tt]

Install a switch or at least try without the bypass cap, for some types of music the difference will be night and day.

Grid resistor value on the first valve doesn't affect the DC conditions of the valve until around 1.5meg for most small signal valves like ecc81. However, it does affect the impedance seen in conjunction with the cable from the source. Anything with an impedance higher than about 100k may pick up more than you intended when driven from a low impedance source such as a computer output, probably 1-10k output impedance. 10:1 is plenty! This can be made a pot until you find the optimum loading for your source. Guitars and some tube equipment likes at least 250k.

[SY]

I would suggest you increase the bias. The ECC81 is, well, not exactly a very good audio valve unless the bias is at least -2V, because it draws a lot more grid current that most other preamp valves.

I generally bias them at 1.7V, 2mA idle current, with absolutely zero problem with grid current or linearity.

SOme people here will mock the ECC83, but the fact is it is a very linear valve

It is indeed IF the plate load is very, very high (e.g., CCS) and the cathode is not degenerated. Otherwise, it's pretty mediocre.

[Krstfrs]

I finally found a nice book on tubes (Designing Valve Preamps for Guitar and Bass, first chapter "The common cathode, triode gain stage" is online), which explains how and why to choose values, most other books I found just explain how to choose them.

Theres just a little bit that disturbs me: The author says that increasing Ra decreases distortion, but increases odd harmonics in this little distortion. I don't understand that, because increasing Ra means decreasing Ia, which pushes the operating point to a much less linear region in the transfer characteristics figure, which explains why there is more odd distortion (relative) but not why there is less overall distortion (absolute). Any explanations?

Many thanks for the answers.