Electron Engine ™
Printed Circuit Boards by Emissionlabs ®
Tuning the EE20 Board.
First, tuning is not really necessary, because in the data sheet are given recommended values for a fixed network, and this network works definitely good. The same applies for data sheets which do not even mention the use of a network. So in case, without oscilloscope, just use the transformers by the data sheet, and you are have the specified performance already by that. However you can outdo this result by circuit tuning, this is what this text is about.
Tuning with an oscilloscope will give a better linearity at frequencies, outside the audible range. This is required to reproduce a square wave with good accuracy. It stays a matter of taste if this is strictly needed, because if a square wave shows distortion like sagging or overshoot, which is caused by added or removing frequency components outside the range of the human ear, it is said this kind of distortion can not be heard anyway. Yet when I can see how nice a 5kHz square wave response can be improved by circuit tuning, I see not reason not to do this, just because it is supposed to be not a not audible.
How this works.
A tone transformer has always a self resonance point, which means in a practical situation the output signal will increase by itself when coming closer to the resonance frequency. When this happens, a large phase error gets introduced, at relatively small the amplitude error already. Even so, the resonance is not just a little bit, it can be 1/3 or more of the signal. As long as we apply no sine wave signal outside the intended frequency range, the self resonance will not take place, and we might say the self resonance plays no role. However when we apply a square wave within the audible range, like 5kHz, this signal contains higher harmonics, which are in the range of the self resonance frequency, and this causes so called ringing. This means at every edge of the square wave, for a very short moment, the transformer resonates. So instead of the expected edge, we see an overshoot which by itself consists of 2...3 periods of the self resonance frequency (with some 5..10% of the amplitude of the square wave itself). I find it difficult to say, if this is a problem or not. But given we can eliminate this by a simple RC network on the output, it may be a good idea to do it.
Here is how it is recommended in the LL1544A data sheet.
However this is just an average number, and the value needed depends on the windings configuration we choose, and also on the capacitance and resistance of the load.
One the EE board, we use an RC filter, with three capacitors to choose from, and an adjustable resistor in series with a fixed resistor. The fixed resistor is just a minimum value we need anyway, and it prevents a shorted output, in case the variable resistor would be set at zero ohm.
This little part is all we need.
Please note, for the ringing compensation, what counts is the RC-time, and a variable resistor can be used to adjust the RC time. Three different capacitors give three ranges, and connecting all three capacitors together gives the fourth range.
For adjustment there are a few ways
- Use the data sheet. These are average values and in any case good. For LL1544A for instance, the data sheet specifies 6k7 and 470pF for the RC network. Connect Capacitor C1 by setting switch C1 to 'ON', insert a 6k7 Resistor at the place of R1, and add a wire bridge instead of the potentiometer. Like this LL1544A is wired by the data sheet. The next part is about more precise adjustment, using an oscilloscope.
- Adjustment with an oscilloscope. I will replace the hand sketch by good oscilloscope pictures later. Please note the value of 70kHz was measured here with an LL1636 at 1:20 step up. Another transformer, and another step up ratio, will give another result.
Here is the procedure: Set C1, C2, C3 to 'OFF'. Apply a square wave of 5kHz to the board input, with a low impedance source. The ringing will now become visible at the output. Choose C1...C3, and adjust the potmeter such that the first half period of the ringing is still there, but the second half period is already gone. This is called critical damping, and it will exactly flatten the frequency curve if tested with a sine wave. To verify this, switch C1...C3 off, but do not touch the potentiometer. Now sweep the oscillator from 5kHz to 200kHz. At some frequency, the output signal will rise by itself, and have a bad phase error too. This is the self resonance point. Now repeat, while the originally chosen capacitor is switched 'ON' again. You will see, the resonance and phase error are gone now. At another gain setting, or another load impedance, the adjustment must be repeated.
Moving Coil input
Damping
A damping network is not mentioned in the Lundahl datasheet, because this depends on too many factors.
What no Cartridge Manufacturer can say in advance.
Not a part of the transformer specifications however, is the reaction of the MC cartridge to amount of damping. The RIAA amplifier, to which the EE21 board is connected, is assumed to have 47k impedance, which they usually have indeed. This is not the issue. However this impedance is transformed to the MC side by the square root of the transformer gain. Which transformer gain is NEVER specified by the MC cartridge manufacturer. So perhaps one chooses a gain of 10x or 20x. A gain of 10x will have only 25% of the damping compared to a gain of 20x. So depending on what MC transformer is taken, there will be quite a change of damping of the cartridge, which is only by coincidence the right one.
In short we can say, at too much damping the sound becomes dull, and at to little damping the needles looses good contact with the groove. How much damping is ideal, depends also on the weight of the head shell and stiffness of the tone arm. All of this can never be specified by the manufacturer, and they will avoid the subject, by just say what is the "minimum" load and that is all. So this is why we recommend variable damping, and switch it on and off, to hear for yourself if there is the difference.