Oops, we were discussing input or output level control, weren't we? Back to the reality. Well, in the end a low value potentiometer (say 10 kohms) at the output of an active line stage is, I believe, a very good solution. It poses no problem for normal amplifier inputs. Power amp input impedance should be at least 10 times higher that pre-amp output impedance, which is around 3 kOhms.
Note that in this way the line stage output signal level is rather high, and is partitioned only later. This certainly means a little more distortion, but in practice it is an acceptable amount. As there is really no audible noise.... A note: a few exceptionally good pre-amplifiers (the new Nagra one, for example...) have indeed two level regulations, one at the input and one at the output, just to cope at best with all possible situations...
Anyway, we still have two problems.
We want to get the same line stage output level from a normal MM pickup connected to the phono stage and a normal CD player connected to the line input. The output from the RIAA second stage is about 250mV @ 1KHz, 4mV phono input signal. But a CD output level is normally 2V. This means that if we take no precautions, the CD level would be 18dB higher than the phono level, requiring the listener to set the volume ever time he or she switches from phono to CD or vice-versa, which is exactly what we want to avoid. So we must do something to solve the problem, but what?
Well, there are only two possible solutions. We could increase phono stage gain by 18 dB or decrease line-input sensitivity by 18dB.
The first solution could be accomplished either by changing the tubes used (both of them could be ECC83's, for example). But this is not easy to accomplish as the phono stage tubes were especially selected. Or, we could add one more stage. However this must be rejected as it would invert the phono stage. So, the line input would be inverting while phono input would be non-inverting, making it impossible to preserve absolute phase of the musical signal from one of the two inputs (absolute phase seems to be audible or at least to have audible effects).
By the way, the preamplifier is always inverting, both as a line and as a phono preamplifier. You must take this into account and reverse the connections between the power amp and loudspeakers in order to preserve absolute phase.
Hence, the only viable solution seems to be the second one, which could be simply accomplished by a resistive voltage divider set immediately after the line-input pin.
A resistive trimmer could also be used in the place of the divider, making it possible to precisely set input sensitivity to the desired value, as in the new Nagra pre. But as far as it is possible for sound and noise reasons I would prefer to use a fixed divider. The second problem is that we want to use paper and oil coupling capacitors, and reasonably sized and priced capacitors are normally available only up to 0.47uF; with a 10kohm potentiometer set up after the coupling capacitor, obviously a big low frequency problem arises. So in the end I decided to use a 47kohm volume potentiometer.
This causes the output resistance to raise to 18Kohm at most, which could be acceptable for a 220kohm input impedance power amp.
Possible problems? Plenty. The major one would probably be that system becomes very sensitive to interconnects and that you could get a lot of RF noise because of the high impedance connection. The first problem can also be seen as an "opportunity" to use interconnects for sound fine-tuning. The second never arose in my environment, which is rather rich in high frequency interfering signals...
If you want to achieve a better low response and/or a lower output impedance, you should use a far higher value film (or paper and oil... if you can find where to put it...) capacitor (4.7uF, for example) and 10kohm potentiometers for both balance and volume.
Precautions against EMI and ground loops
Tube circuits often have a very high impedance. By the way, this is strictly true in the old time design practice, or if you follow tube design hints on tube data sheets. Often you find 1 Megaohm grid resistors or things like that in order to reduce coupling capacitors values, and hence overall cost. In facts you can reduce these values to far more acceptable levels (typically very few hundreds kOhms) without any problem.
Anyway, tubes internal circuitry has intrinsically rather high impedance and they can work up to very high frequencies, so it is always possible for an interfering high frequency signal to be picked up.
There are precautions to be taken to avoid this kind of problems, and they concern circuit shielding, circuit design and layout design.
All audio units must be completely shielded against high frequency (radio) signals. This also requires a good separation between power supply unit, which is both a direct source of high frequency noise due to switching diodes and a very good receiver of external EMI signals, either directly through the air, if not properly shielded, or through the mains lead.
In order to avoid this last kind of problem, special care has been taken, as you know, in designing mains connection. Just to be sure that no interfering signal is able to reach the power supply unit.
The power supply structure should be completely shielded by the cabinet from the external environment. The cabinet also shields the audio circuits from power supply unit. The HT power supply layout has been designed in order to avoid easy propagation of switching noise. In fact, the layout follows the logical circuit structure and you can find all components physically set in the same order as in the schema.
The rectifier tube is at the extreme right side of the box, as far away as possible from the line circuits, and the last filter capacitor at the other end, towards the centre of the box (and the audio circuit).
Moreover the transformer and the two coils have also been set so that mutual inductance should be minimised, each with differently oriented packs. The filament power supply unit has been finally placed near the power transformer, on the right side of the box. Even the audio units have been placed accordingly to their sensitivity.
The line stage at the centre, by the PSU but separated from this by a shielding section and the phono stage is at the left end of the box. All this keeps any chance of direct propagation of switching noise to the power supply output as low as possible. Just to be sure, as you know, one more RC filter for each triode has been placed in the phono circuit, so that they cannot be influenced by PSU conducted noise. Even earthing layout is very important.
In both the line and the phono stages, the component layout should follow, as far as possible, the functional distribution of the component. I normally use what is called bus ground lines, that is a solid core copper wire about 1.2 mm in diameter soldered along the circuit and with all other component connected either directly or indirectly to it.
For line stage the arrangement I normally use begins with the input pin. The input RCA sockets must be completely insulated from the cabinet. All the ground connections must be connected together with one bar of silver plated copper, which, one end goes to the cabinet shield connection, and on the other forks and goes to the line circuit of each channel. Each ground bar connects directly the cold side of the cathode resistors of the corresponding channel line tube. This provides the correct reference to the grid, which is connected (through the only input selector) to the RCA socket signal pin.
The bar goes through the whole stage, being the one ground reference for all purposes. At the output of the stage, the bars forks. One branch connects both channels grounds together and to the PSU ground, the others goes along the signal connection reaching the volume and balance controls, with the required connections to these potentiometers, and then goes to the potentiometer output socket.
The same pattern is applied to the physical layout of each channel's phono stage. Here the input socket signal pin is connected to the input load and tube grid by a thin teflon-insulated coaxial wire.
The ground is not connected to the other input socket through the ground bar, but only to the phono stage input via the shield of the coaxial wire, The shield is connected at the same node as the pickup load RC network and the cold end of first stage cathode resistor. From this point another ground bar goes along the stage as the ground reference for all the other purposes; the opposite end of the bars, at the phono stage output, are connected together and to the power supply ground.
This configuration seems to work perfectly in my prototype, but I would expect possible problems if different components were used or overall layout was different.
What I can suggest you is to try following the instructions above, but feel free of changing anything you think might make things better if you find any problems. As far as I can say, in my experience, this kind of layout seems absolutely effective in preventing hi frequency noise or low frequency buzzes disturbing my listening to music.
But in my experience, if problems arise, are very hard to solve...
When in doubt, remember – ensure that all the ground paths exactly follow the logical schema layout. Last but not least, a few precautions should be used even in circuit design. The typical circuit design uses low value resistors in series with the tube pins, named stoppers.
These resistors form, in conjunction with the tube pin inter-electrode capacitance (a few pF) a low pass RC filter, which gives good results at very high frequencies. In nearly any good design you'll find a grid stopper, but even anode and cathode stoppers can be used.
In the case of the line stage, where two triodes are connected in parallel, there are more leads connected together and hence a higher risk of interference. Moreover the two triodes could also interact in uncommon and unexpected ways. Hence, the grid stoppers are mandatory. Also cathode and anode stoppers, all placed close to the tube sockets pins have been used.
The phono stage can be subject to very high frequency interference (FM radio or similar), which can became audible because of different detection phenomena. While this is dealt with by good layout design and good shielding, a major problem can arise further up the chain.
Interference can be picked up by either the turntable interconnect wire or even the pickup or intermediate wiring. To reduce this kind of problem the most common solution is the grid stopper resistor. In my case I used a few hundred ohm resistor, but you could increase or decrease the value according to your needs and problems.
In very bad cases, you could even use a little inductance, but I would not suggest it unless it is really necessary and it is the one way to get rid of the problem. Also, take into account that your pickup input resistance could be of the order of a few hundreds ohms.
Dramatically increasing the stopper could add a significant amount of noise... Anyway before trying this solution you should try:
© Copyright 1998 Giorgio Pozzoli for TNT-Audio, http://www.tnt-audio.com
Go to Part 5
How to print this article