Fullscreen HTML5Med Res (??MB)HTML4

By LEO SIMPSON Fixing transformer BUZZ in the Class-A Amplifier Since the 20W Stereo Class-A amplifier was published in 2007, it has been widely acclaimed for its sound quality. But there has been a consistent niggle experienced by quite a few constructors – low level buzz from the power transformer. We recently took a look at this problem and have come up with a cure. W HEN WE PUBLISHED this amplifier in 2007 we were well aware of the low level buzz from the power transformer and we regarded it as inevitable. After all, in a Class-A amplifier, the load current is constant and always high, regardless of how much audio power is actually being delivered. That and the amount of heat produced are the two major drawbacks of class-A amplifiers. But that constant load on the power transformer means that it is always working hard. In this case, it is delivering over 2A from the balanced ±22V (nominal) DC supply rails. That means the peak rectifier currents can be expected to be at least 10A and it is these heavy pulse currents at 100Hz which cause the audible buzz from the transformer windings. But we judged at the time that the buzz should not be a problem with the lid on the case, even when playing quiet passages of music in a quiet room. 40  Silicon Chip And while we judged our prototype to be quite satisfactory, it is apparent that normal production variations mean that some transformers are noisier than some constructors would like. Just recently we have had a number of readers complaining and we were cogitating about the problem: maybe pot the transformer, use pliant mounting or some-such . . . Then there was a moment of serendipity as one of our staff who had recently been reading one of the articles on Vintage Radio had a sudden thought: what if we tried a chokecapacitor filter for the power supply? Choke input filters were widely used for the HT (high tension) rails in valve sets because high-voltage capacitors were expensive. And before permanent magnet loudspeakers became common place, the coil for the electromagnet in the loudspeaker did double duty as the choke for the power supply filter. But all of this clever circuitry fell into disuse as highvoltage capacitors became cheaper and more readily available and permanent magnet speakers became the standard. Such is the progress of technology. So the design of choke input power supply filters has become something of a lost art. In typical valve radios, the iron-cored choke would have had an inductance of around five Henries and be rated for a current of no more than about 50 milliamps. The class-A amplifier has lower supply rails but much higher currents and in any case, we would not want such large values of inductance. Why not? The answer is that a choke input power supply works quite differently from today’s capacitor input power supply filters. In the latter supplies, the rectifier diodes conduct for only a short time during the peaks of the AC waveform, producing the high current pulses at 100Hz, as mentioned above. siliconchip.com.au S1 250VAC A F1 T4A T1 160VA TOROIDAL 16V 0.02 * 230VAC INPUT BR1 35A/400V _ L1 470 H + A 0.02 * 16V N E Fig.1: the modified power supply has two 470µH chokes (L1 & L2) connected in series with the positive and negative outputs from the bridge rectifier.  10,000 F 35V 10,000 F 35V 10,000 F 35V 100nF * FOR TESTING L2 470 H SC K 2.2k 1W GND CHASSIS EARTH 2011 +19.5V NOM 10,000 F 35V 10,000 F 35V 10,000 F 35V 2.2k 1W 100nF A  K –19.5V NOM 20W CLASS-A AMPLIFIER POWER SUPPLY By contrast, in a typical choke input power supply, the rectifier diodes, whether they be thermionic (ie, valve) or semiconductor, typically conduct over most of the 50Hz AC waveform so the high 100Hz pulse currents don’t occur. The result is that the 100Hz ripple on the DC supply is more sinusoidal rather than sawtooth, as it is for capacitor input power supply filters. Hence, we could expect the addition of chokes in series with rectifier outputs to the power supply board in the class-A amplifier might be able to reduce the transformer buzz. But not so fast. There is much more to choke input power supplies than meets the eye. In capacitor input power supplies, the DC rails are usually only slightly less than the peak value of the AC input waveform. For example, when 16VAC is rectified, you can expect a DC rail of about 22V. But in a typical choke input filter as used in vintage radios, the DC voltage will be a great deal less; about 90% of the RMS value of the AC voltage. That would be unusable in the Class-A amplifier because the reduction in the DC supply rails would greatly reduce the available audio power output. So before we even started we knew that we would have use small-value chokes – just enough to give a useful reduction in the peak rectifier currents without an undue reduction in the DC supply rails. We won’t detail the attempts that didn’t work; instead, we will cut to the chase and give the solution which siliconchip.com.au involves a pair of cheap and readilyavailable iron-dust toroidal chokes with an inductance of 470µH and a current rating of 5A. These chokes are much, much smaller than the chokes typically used in vintage radios. Circuit details Fig.1 shows how the chokes, L1 and L2, are connected in series with the positive and negative bridge rectifier outputs to the capacitor bank of the power supply. Electrically, the effect of these chokes is quite modest but it is enough to give a major reduction in buzz from the power transformer. To demon- strate the effect, we have produced a number of scope grabs showing conditions in the power supply with and without the chokes. Fig.2 shows the conditions from the standard power supply, without the chokes in circuit. The green and yellow traces show the 100Hz sawtooth ripple voltages superimposed on the positive (green) and negative (yellow) supply rails. The magenta and cyan traces depict the rectifier currents flowing in the secondary windings of the power transformer. These short pulse currents charge the 30,000µF capacitor bank in the power supply. Notice that the individual winding currents are The two 470µH chokes are secured to a piece of blank PCB material using Nylon cable ties. This assembly is then mounted in the chassis on 12mm standoffs and secured using M3 x 6mm machine screws and washers. April 2011  41 This close-up view shows how the chokes are mounted and wired into circuit. The outputs from the bridge rectifier go to the leads on one side of the chokes, while the leads on the opposite side go to the corresponding positive and negative inputs on the Power Supply Board. Be sure to use heavy-duty hook-up wire for the choke connections. at 50Hz but since they are 180° out of phase, they result in 100Hz ripple on the supply rails. We measured those pulse currents with a 0.02Ω (20 milli-ohms) shunt in each secondary winding, as shown on Fig.1 which is modified from the original power supply circuit published in June 2007 (note: these resistors are shown in red and are not to be fitted for the purpose of this modification). Given that the scope sensitivity for the current measurement is 100mV/div and the resistance is 0.02Ω, the pulse currents have an amplitude of 11A peak. Furthermore, the period of rectifier conduction in each half-cycle is about 3ms. No wonder the transformer buzzes! Fig.3 shows what happened when the chokes were wired in place. Not only are the rectifier pulse currents slightly reduced in amplitude, down to about 10A peak, the period of conduction is now extended to about 5ms in each half-cycle. By the way, for this measurement, we inverted channel 4 of the scope (cyan) so that the pulse voltages are neatly superimposed. OK, so the pulse currents are only a little less savage but that is not the whole story. You can also see that the 100Hz sawtooth ripple voltage 42  Silicon Chip waveforms are now smoother but there is also a drawback to the choke modification: the supply voltage has been reduced, from around ±19.5V to around ±18.15V. That is a significant reduction and will lead to a reduction in the maximum power output of the amplifier of a few watts. Fortunately, the audible effect of that order of reduction will be unnoticeable. Installing the chokes The recommended chokes are 470µH toroidal units rated at 5A and wound on sintered iron cores. They have a DC resistance of just 0.05Ω (50 milliohms). This low resistance is important because even with this low figure they will get hot. They are available from Altronics (Cat. L-6630) and Jaycar Electronics (Cat. LF-1278). Two chokes are required. We installed the chokes on a piece of blank PCB material. This should be cut and drilled as shown in Fig.4, after which you can secure the chokes in place using Nylon cable ties. Bend the leads of the chokes at right angles, as shown in one of the photos The completed choke assembly is mounted in the chassis between the Power Supply Board and the leftchannel power amplifier board. You will need to carefully mark out the chassis mounting hole positions, then drill the holes to 3mm. It’s best to drill the holes from the underside of the chassis but you have to be careful to capture the metal swarf. Several layers of sticky tape on the inside of the amplifier and some strategically placed plastic wrapping can keep the swarf from contaminating other parts of the amplifier. In addition, you should cover the ventilation slots for the power transformer on the underside of the amplifier before you start drilling. Be sure to position the holes accurately and use an oversize drill to remove any metal swarf from around the holes. The choke assembly can then be mounted in position on 12mm tapped stand-offs and secured using M3 x 6mm machine screws and washers. Connecting them up The chokes must be connected into circuit using heavy-duty hook-up wire. The first step is to unplug the positive and negative leads from the bridge rectifier at the power supply board. The spade connectors are then clipped off each lead and the wire ends stripped and soldered to the leads on one side of the chokes. These connecsiliconchip.com.au Fig.2 – standard power supply: the green & yellow traces in this scope grab show the 100Hz sawtooth ripple voltages superimposed on the positive & negative supply rails, while the magenta & blue traces show the rectifier currents flowing in the secondary windings of the power transformer. tions should be insulated using short lengths of heatshrink sleeving. After that, it’s just a matter of running leads from the other side of the chokes to the positive and negative terminals on the Power Supply Board. These leads can be terminated with fully-insulated 6.3mm spade lugs to plug into the quick connect terminals. Mains reduced to 230VAC Keen-eyed readers who compare these voltage figures with those originally quoted in the 2007 articles will have notice a big discrepancy: the supply rails were originally ±22V. Why the big reduction? There are two reasons for this. The first is that Australia’s mains voltage is now officially 230VAC and we frequently see mains voltages below that in the SILICON CHIP offices. Partly that is because the mains voltage is now generally lower but it also happens because of heavy machinery being used elsewhere in our building. Indeed, if were presenting the Class-A amplifier in 2011 we would now specify a transformer with 18V secondaries rather than 16V, to cover this reduction in mains voltage. Unfortunately, the picture is a little more complicated though because there are areas of Australia when the mains voltage still exceeds 250VAC. This reduction in mains voltage was highlighted by another problem which became apparent after we had siliconchip.com.au Fig.3 – modified power supply: here’s what happens when the 470µH chokes are wired in place. The 100Hz ripple waveforms are now more sinusoidal with rounded peaks, while the rectifier pulse currents are slightly reduced in amplitude. In addition, the period of conduction has been extended from about 3ms to about 5ms in each half cycle. 90 A A 32 10 A 8 20 A 10 A 20 8 CL A 34 ALL HOLES A ARE 3.0mm DIAMETER ALL DIMENSIONS IN MILLIMETRES Fig.4: this diagram shows the dimensions and hole positions for the choke assembly PCB. installed the chokes: the relay on the Speaker Protection & Muting board was reluctant to operate. While its internal LED was lighting up, the relay contacts were not closing. The quick and easy cure for this is to replace diode D1 on the PCB with a wire link. If that proves ineffective, the relay will need to be changed to a 12V type (Altronics Cat. S-4311) and a 100Ω 1W resistor installed in place of the link originally shown for R2 on the PCB. These changes to the Speaker Protection Board are only necessary if the relay operation proves unreliable. Performance testing To verify that the addition of the supply chokes had not any deleterious effect on the performance of the 20W Class-A Stereo Amplifier, we ran all the significant measurements with the mains input voltage adjusted to 240VAC. There was no real difference apart from the absence of transformer buzz, although there is still a very subdued hum from the transformer. The only difference in performance is a very slight increase in distortion from the left-channel amplifier, due to the proximity of the chokes. This could be avoided by mounting the choke assembly on the rear panel, between the power transformer and the Loudspeaker Protection Board. Conclusion If you are bothered by the buzz from your transformer, then you should consider installing the chokes as we have described. They will make a considerable difference. However, if your mains voltage is low (ie, below 230V), you might think twice. If you do go ahead, you may need to also replace diode D1 on the Speaker Protection Board with a wire SC link, as described above. April 2011  43