Single ended EL34 amp design

[Stephen1966]

Second Harmonic distortion

To find this we need Vpeak between QB and QC. There is a useful set of notes on harmonic distortion in https://www.analogethos.com/post/unders ... distortion and just from eyeballing the points BQ and QC on the chart, we can see that when the gain is asymmetrical it tends towards more even order harmonics. If the two sides, either side of Q were more or less even there would be a higher odd harmonic content. To find the percentage of second harmonic (2H) distortion we apply the formula:

2H = 50% x ((BQ - QC) / (BQ + QC)) x 100%

When BQ = 312V and QC = 250V

2H = 0.5 x ((312 - 250) / (312 + 250)) x 100 = 5.54%

Again, this is not exactly the same as the calculator where the THD for the 2nd, 3rd and 4th are given as parameters of the screen voltage and the headroom voltages, but are close enough. Actual THD may be way off when fire this amp up - and again, as working figures, they are enough to be getting on with.

The goal is not to eliminate THD altogether as it might be in a HiFi amp, guitar amps can benefit from the 2nd harmonic distortion and it all adds to the tone we seek. Underloading the tube with a Zpri of 3500 Ohm increases the THD significantly, 2nd harmonic is increased, but so too is the generally unpleasant 3rd and higher odd harmonic content. The goal then, is not to drive the output tube into all that much distortion. The distortion we want should be driven mainly by the preamp tubes and similar load line characteristics will need to be drawn up for the triodes when we get to that stage. As design is a recursive process, we will then see what new parameters we need to plug in/include in the load lines for the power tube.

[Stephen1966]

A very useful link here from the late, great Patrick Turner. The last posts here were worked out using his methods.

https://www.turneraudio.com.au/loadmatc ... todes.html

[Stephen1966]

Here are an additional set of charts I redrew from the Mullard and Brimar datasheets. Includes the plate characteristics chart with Vg2 = 250V, a to-scale interpolation of the Brimar mutual characteristics chart alongside, and a copy of the enlarged Mullard mutual characteristics chart on the second page.

EL34 anode and mutual characteristics.pdf

Enjoy!

[Stephen1966]

Finding the screen-grid voltage, Vg2
For me, the most challenging part of understanding the loadlines and the charts is how we arrive at the voltage sag we need to arrive at with the screens. Pages and pages of internet later I've arrived at the following. I'm still not sure I am doing it correctly but I keep returning to a set of datapoints which seem to confirm I'm on the right path at least.

Using the anode and mutual characteristics charts to plot the equivalent screen voltages I have this:

EL34 - Va 350 - Zpri 5000 (2)-01.jpg

I tried following Merlin's tutorial where he morphs the curves down - Vg1 OV becomes -8V and -8V becomes -16V and so on but the key piece of information as I understand it, is that the resulting findings are based on "peak" conditions. Whereas in his book on designing preamps he looks at what happens to the grid curves as they are morphed under load with Rg2 = 100k. The grid curves compress down and the knee becomes softer with the increased load. There, he states:

However, it must be appreciated that [the compression] only applies to dynamic conditions, that is when the valve is actually amplifying. For biasing or working out DC voltages and currents we would still use the static characteristics...

And by that, I think he means the mutual characteristics should be derived from the quiescent operating point of the circuit as designed.

The screen and anode currents are 'more or less constant' for a given anode voltage as expressed in the ratio: m. So it follows that when the anode and screen voltages are charted as equivalent, voltages on the screen are dynamically related to the voltages on the plate as expressed in the mutual characteristics chart. The relationship is non-linear but as the tube approaches saturation the current goes up as the voltage goes down and vice versa, in the opposite direction past the operating point; as the voltages go up, the current goes down.

I needed to find the voltage drop across Rg2 the screen-grid stopper. With voltage and current, I can calculate the resistance of Rg2

In the mutual characteristics chart I found the earlier Vg = Vg2 = 250V curve to be off (the original is greyed out) The new curve plotted from Va = 250V shows the corresponding Ia and Vg axis in green. Interpolating the Vg1 = 0V on my anode characteristics chart across to my 350V operating point. I end up with the points plotted in magenta - the 0V point on the mutual chart is slightly great than 250V but the other points correspond closer to 250V at lower Vg1 voltages.

[Stephen1966]

Calculating the values of the screen-grid stopper, Rg2

In a design twist on the Champ and similar single ended amps I'm thinking it will be good to use a 10H choke as the screen-grid dropper. The 10H/100mA/130 Ohm (DCR) here: https://primarywindings.com/product/10- ... -dc-choke/. Given the reputation I've been hearing about - for SE pentode power stages being some of the noisier topologies - I think the circuit would benefit from the filtering such a choke would offer. A low noise floor being one of the goals here. Following Merlin's descriptions of screen-grid droppers and stoppers, this is the equivalent of the screen grid dropper. And the voltage drop through 130 Ohms is negligible.

The quiescent screen current (9.4mA) and the projected current draw of the preamp sections (6mA) is just 15.4mA giving a voltage drop of:

130 x 0.0154 = 2.002V

For the purposes of calculating the values of Rg2 (the stopper) then, and unless we find there is a much bigger voltage drop in practice, we can ignore this small difference and assume the stopper has to drop its voltage 100V; from 350V to 250V.

I now take point (A) from charts, the peak anode current and utilizing Merlin's "fudge factor" say the screen "steals" all this current. When (A) = 133mA the minimum value of Rg2 becomes:

100 / 0.133 = 752 Ohms

* We talked earlier about compression and I am shifting my view of it. It may be true that compression tends to level off dynamic levels to some extent, but the more I learn about it, the more I realise it may be a good thing in small doses and may actually lead (counterintuitively to the way I was thinking) to better note articulation and touch sensitivity. The component, Rg2 is a hefty resistor in circuit but relatively easy to swap out and audition for other values.

Increasing the value to the next available and going with a 5W wirewound I arrive at 800 Ohms

Providing a voltage drop of:

800 x 0.133 = 106.4V

Increasing the resistance of Rg2 to 1k, the voltage drop becomes greater:

1000 x 0.133 = 133V

In 'Controlling Screen Current', Rob Robinette says:

Increasing the resistance value of the screen resistor will increase the overdrive screen voltage drop and therefore increase power tube distortion, compression and sustain.

https://robrobinette.com/Tube_Guitar_Am ... Distortion

The distortion will come heavy in the form of 2nd harmonics but as that increases, so does the 3rd. Rob recommends a 1k5 for the screen so there is plenty of scope for experimentation here.

There seems to be a lot of speculation about the role of the grid stopper out there - here, for instance https://el34world.com/Forum/index.php?t ... #msg230307 - and many, seem to suggest the smallest possible value yields better results but opinions are like... well, you probably know how that old saying goes!

[Stephen1966]

Find bias resistor
A cursory look at the operating point on the anode characteristics chart seems to suggest that the bias resistor needs to drop around 16V. Merlin's method however, using the mutual characteristics chart on the left stipulates the Vgk = -24V.

EL34 - Va 350 - Zpri 5000 (3)-01.jpg

I get this by summing the quiescent current (Ia) with the quiescent current of the screen (Ig2). The ratio of screen to anode current (m) is 6.7 so:

When Ia = 63mA

Ig2 = 63/6.7 = 9.4mA

and so,

Ik = 63 + 9.4 = 72.4mA

and

Rk = 24/0.072 = 333 Ohms

There's a bit of rounding going on here but it's very close and the nearest value is 332 Ohms

We can also use the same figures of Ik and Rk to find the minimum dissipation:

P = 332 x (0.072)^2 = 1.72W

The higher rated, 3W wirewound resistor is very similar to what Merlin found in his calculations. 3W or higher is the recommendation there.

[Stephen1966]

Find cathode bypass capacitor

The final steps are mostly uncontroversial and straightforward...

Calculating for a low, (10Hz) roll-off when Rk = 332 Ohms, we get:

Ck = 1/(2 x 3.141 x 10 x 332) = 47.9uF

And with a voltage rating capable of handling 3 times the expected voltage: 100V.

Merlin suggests most people don't really give it a thought and just go for 100uF

Likewise, I've gone for a nominal 4k7 (control) grid stopper and a 470k grid reference resistor in line with Merlin's choices. Some more working out of those values will probably take place during the design of the preamp stages and and again, once it's up and running, but for the moment, it's enough to be going on with.

Here then is a schematic of the work so far:

SE amp - 01.pdf

The power transformer is a Blues Junior type: Hammond 290XEX for Fender Power 049967 / Blues Junior

The output transformer is: https://primarywindings.com/product/500 ... sformer/

and the choke: https://primarywindings.com/product/10- ... c-choke/

I've documented the calculations here as both a learning strategy and an attempt to clear up what otherwise seems a somewhat opaque process. I invite comments/corrections and I'm sure I will change my mind all over again. Different Va, different Zpri on the OT, and so on. But, as long as there no holes in my understanding of this part of the project I'm sure I will arrive at an optimum set of values for this tube.

Other things on the agenda now are a study of the champ preamp circuit and a look into regulated switching DC heater supplies. The power factor loss on a standard regulated DC supply is about 50% and a switching supply promises greater efficiency. If I can get that, I might be able to avoid a hefty transformer for the filaments in the PT (and use a regular PT) or avoid the need for an altogether separate filament transformer. Space isn't going to be a problem but weight is a factor I would also like to keep to a minimum.

[bepone]

Rg1 as small as possible, so 220-330k

[bepone]

OT taken is too large, no need so big, and price is following 10-15W SE is more than enough..10w will also pass even if output power is 12, 13, 14W this is all the same, will saturate sooner from bottom part (distortion on bass side)

[bepone]

this for the output section is clear.. i have built one great SE amp 30W with 2xKT88 where preamp was 2xECC83 / trainwereck style.. it was killin' how good and simple it was
speaking that SE can also pass for the higher gain without a problem , with little bit of NFB

[bepone]

this is recorded with potato, full in cliping, but if you remove that in the brain with internal masking filter, you can hear some sound from it>
https://www.youtube.com/watch?v=5CDivirjQNY
all is diy and own design, PT, OT, etc .. OT was huge but very nice sounding

here is another one, on small speaker 10" , working with 1 KT88 tube - 15W
https://www.youtube.com/watch?v=4qqA4Gzn1yY

[Stephen1966]

this is recorded with potato, full in cliping, but if you remove that in the brain with internal masking filter, you can hear some sound from it>
https://www.youtube.com/watch?v=5CDivirjQNY
all is diy and own design, PT, OT, etc .. OT was huge but very nice sounding

here is another one, on small speaker 10" , working with 1 KT88 tube - 15W
https://www.youtube.com/watch?v=4qqA4Gzn1yY

That is one killer tuber - potatoes are not vegetables my friends

The ultralinear doesn't have the same heft but still, packs a real nice punch.

Great designs! Superb!

[Stephen1966]

Rg1 as small as possible, so 220-330k

Thank you bepone. As I understand it, the grid balance resistor also has something in relation to the NFB you mention below.

So that I may understand it better, could you explain how and why you arrive at 220-330k for Rg1?

[Stephen1966]

OT taken is too large, no need so big, and price is following 10-15W SE is more than enough..10w will also pass even if output power is 12, 13, 14W this is all the same, will saturate sooner from bottom part (distortion on bass side)

You are probably right, I like this transformer because it looks well built and has a good bandwidth compared to similar Hammond transformers for example. I'm not even sure I will stay with this though, there are possibilities for extra gain using different primary impedances.

What do you think of the power transformer though?

[bepone]

Rg1 as small as possible, so 220-330k

Thank you bepone. As I understand it, the grid balance resistor also has something in relation to the NFB you mention below.

So that I may understand it better, could you explain how and why you arrive at 220-330k for Rg1?

thx,
In case of too big grid leak, out. tube will be less controlled, you can have thermal runnaway, and current in the output tube will raise more and more and more and more until melting.. or some fuse activating.. it is rare the case but is happening, if you want to be sure, just put less than half recommended, (0.7 Mohms in EL34 datasheet)..

This will load previous stage, but this is also ok because load line 330k in parallel with anode resistor in previous stage *ECC83, will generete more 2. harmonic what is also good for us!