This page last updated: 25 July 98
14 JUNE 96 COMPONENTS. These constructional notes should come accompanied by a full component list, and so only particular points of interest are dealt with here. Please note that while all possible care has been taken in compiling this list, no responsibility can be accepted for any errors in part numbers, etc. Capacitors: Finding affordable close-tolerance capacitors is not easy; the best solution seems to be (as it was in 1983) the axial polystyrene type, available at 1% tolerance for 30 to 50p, depending on the value. (Part number BX46A etc, from Maplin) As described in the EW articles, the actual effective capacitance is to a closer tolerance than 1% as a result of paralleling several components. As also mentioned in EW, the subsonic filter capacitors C21,22,23 can generate distortion around the rolloff point, and this can be greatly decreased by using the highest voltage rating available, which will probably be 100V; the PCB is designed for this size component. Special 400V capacitors intended for mains filtering would probably be even better, but this has not yet been tried. They are physically large so will require a little ingenuity to mount them on the PCB. Please note that this distortion will typically only be measurable at signal levels a hundred or so times above the nominal operating level, so the problem is not too serious. Resistors: The resistors are all 1%, (Part number M1R etc, 4p from Maplin) which is no longer expensive or exotic, though anything more accurate certainly would be. Nothing is to be gained by using anything more exotic than carbon-film in most parts of the circuit, as there is no DC voltage across the component and so no possibility of excess noise being generated. The MC amplifier stage does have resistors carrying significant DC voltage, such as R4, R7, R14 etc, and it is just conceivable that the use of metal-oxide resistors here might give a barely-detectable noise improvement. However, I have never been able to find such an improvement myself. Opamps: I have never detected any significant performance differences (as regards the audio parameters of noise and distortion) between 5532s from different manufacturers. My experience has always been that if they work, they work well. The situation is quite otherwise for TL072s, which can vary quite widely in linearity. The best source is Texas. However, in this application IC2, the sole TL072, is working as a DC integrator so this hardly matters. The offset voltage is of no real importance as it is not necessary for IC2 to hold IC1a output exactly at 0V; a 100 mV error is of little significance for headroom. Having said that, if you do have more than +/-10 mV at IC1a output, it probably does indicate a fault somewhere. While many opamp types will function satisfactorily in this preamplifier, it will be difficult to improve on the 5534/5532, which is very well suited for the impedances prevailing in this design. A good example is the MM stage, which is primarily optimised for a cartridge impedance of 610 ohms plus 470 mH (eg Shure M75ED) For four different opamps, the calculated noise output of the RIAA- equalised MM stage, using the mathematical model MAGNOISE is: 5534 -92.5 dBu OP27 -94.8 dBu TL072 -87.1 dBu LT1028 -91.8 dBu This demonstrates that the TL072 is noisier than the 5534 because it gives optimal noise performance with a higher load impedance than the cartridge presents, while the expensive LT1028 is noisier because it is optimised for a lower load impedance. The figure for the OP27 is misleading, because the measured noise is actually 2 dB higher than for the 5534; this is believed to be due to the extra common-mode noise generated in an input-bias-current cancellation system, which is not taken into account in MAGNOISE. This extra noise is believed to only be a problem in applications such as this, where the impedances at the two op-amp inputs are very different. Relays: For some reason I cannot begin to guess at, it is still possible to buy relays that are hopelessly unreliable. Bitter experience is often the only guide. The type recommended in the parts list has proved reliable in my experience. CIRCUIT OPERATION. The MC stage has proved entirely stable at high frequencies. However, there is always the possibility with hybrid discrete/opamp stages that instability may show itself due to some unfortunate combination of circumstances. If this appears to be the case (and you will need an oscilloscope to be sure) then C6 may be cautiously increased in value to lower the "dominant pole" around IC1a. Do not increase the value by more than necessary as this may degrade the HF distortion, though this will typically only be measurable at signal levels a thousand times above the nominal operating level (ie 5 V rms at IC1a output would be required to reveal the effect) This stage, in conjunction with the MM stage following, has a high gain, and RF breakthrough is a possibility. A small ceramic capacitor of up to 1000 pF directly across the MC input phono socket is likely to be an effective cure. (This capacitance could probably be greatly increased, but this depends on the cartridge characteristics, and must be left to you to decide) Attention to the record deck grounding may also help. The MM disc stage opamps are 5534s without internal compensation for unity gain; they are specified for stability with a gain of three. In this design they should need no external compensation capacitor, but the safety margin is not enormous, so provision is made for adding external capacitance between pins 5 and 8. If it is needed then 5 pF should be ample; if 12pF (which would compensate the opamp for unity gain) is required for HF stability, then something is probably wrong with the feedback network. The Active Gain Stage will definitely show signs of HF instability if C43 is not present. CONSTRUCTION. General: This is a double-sided PTH (Plated Through Hole) board, and desoldering components can be difficult. A good solder-sucker is essential. You will find that the solder pads adhere much better than for a normal PCB, but conversely it is much harder to get all the solder out of a plated-through hole, so the component lead can be released. Opamps: It is strongly recommended that these are socketed. Unsoldering an 8-pin IC from a PTH board is not amusing, and can easily lead to PCB damage. In the absence of specialised desoldering equipment, the best strategy is to cut off the body of the defunct device with a strong pair of cutters, so that the legs can be removed one by one. Power supply: Build this section first, and check that the output voltages are correct. Note that insulating thermal washers are essential between regulators and heatsink. Input/output sockets: The PCB is designed so that phono sockets may be directly mounted into it, facing upwards. This assumes that an external case is used that has much less height over this part of the board than the rest, and which will therefore need to be specially made. If you do follow this option, remember that the insertion force of "audiophile" phono plugs can be high, and if the PCB distorts under the pressure the solder joints may suffer. To alleviate this several fixing holes are provided around the phono socket area, so that the board may be firmly supported from a rigid base. Alternatively, the phono sockets can be mounted to the inside of a standard case, and wired down to the PCB with single screened cable. Screened cable is essential if intersource crosstalk is not to be impaired. CUSTOMISATION. Many variations are possible on the basic preamplifier, whilst still using the unmodified PCB: MC stage gain: This can be customised by changing the tapping points on the feedback arm; (R7,10,13) gains of between x5 and x100 are easily obtained. If less gain than this is required then the MM input might as well be used, and the volume control advanced further, bearing in mind that there is at least 6 dB of gain in hand. Make sure that the total resistance of R7 + R10 + R13 does not fall below 600 Ohms, or linearity may be degraded by the extra current demands placed on IC1a. MM stage gain: Altering this is not simple as changing R0 requires small but significant changes in Ra and Rb if the RIAA accuracy is to remain within +/- 0.05 dB. I regret I am unable to enter into correspondence concerning different gain structures- the work involved in recalculation is significant. Line levels: These have an operating level of 150 mV, but this a purely nominal figure. The maximum signal capability of the following stages is about 10 Vrms, so the safety margin is large. Even if maximum boost is used in the tone control stage, this is only +10 dB maximum, and so the preamp inputs will not overload before an input of 3V rms is supplied; domestic equipment can rarely if ever produce such levels. It therefore follows that simply adjusting the volume control will cope with most variations in input level. The CD input may however be configured for different gains by altering the values in the input attenuator R57,58. The resistor values should be kept as low as possible (without excessive loading on the source) to ensure that the Johnson noise generated by these components remains negligible. Tone control: If you don't want tone controls, then the entire sidechain (Ic6, IC7b, IC8 and associated components) may be omitted. The two controls VR2a and VR4a are omitted, and pin 5 of IC7a connected directly to ground by a wire link that replaces R38. Active Gain Stage: The maximum gain of this can be altered (for example, if you have an exceptionally insensitive power amp requiring more than 2 Vrms to drive it) by changing R53. Increasing R53 increases the maximum gain in proportion, though the gain at VR6 mid-travel is only slightly altered. If R53 is increased to more than 22K, C43 should be reduced to 47 pF to prevent premature HF rolloff. Relay muting: This can be omitted entirely, including all the control circuitry, if you are sure the preamplifier will only be used with power amplifiers that have their own efficient muting system. The normally-open relay contacts must be linked across. Douglas Self.