Fullscreen HTML5Med Res (??MB)HTML4

A dummy load box for large audio amplifiers Checking the power output of a large audio amplifier is not a trivial exercise. A stereo amplifier rated at around 500-1000 watts per channel will need to be tested into 8, 4 and 2-ohm loads, as well as requiring a 1-hour soak at 33% of its rated power. The load box described here meets this need. By LEO SIMPSON Since we do quite a lot of testing of audio amplifiers from time to time, we often have need of a load box which is able to dissipate a lot of power. We set a target rating of 1000 watts per channel with load impedances of 8, 4 and 2Q. And while amplifiers rated to deliver 1000 watts per channel into 80 are not commonplace, many amplifiers will deliver quite 62 SILICON CHIP surprising amounts of power into 20 loads for short periods. You might ask why it should be necessary to test power output into 20 since the vast majority of loudspeakers have a nominal impedance of either 80 or 4Q. The answer is threefold. First, if two 40 loudspeaker systems are connected in parallel, the resulting load impedance will be nominally 20 and that is what the amplifier has to drive. Second, the impedance of some 40 loudspeakers may dip down to as low as 20 at some frequencies. Third, many amplifier manufacturers make a big point about how much current their amplifiers can deliver. The specification for amplifier resistive loads can be found in IHF-A202, published by The Institute of High Fidelity, Inc, New York, in 1978. To quote the relevant section: "The resistor shall not have more than 10% Above: a high power load box requires a big case and plenty of forced air ventilation. This case measures 540mm wide, 210mm high and 350mm deep. It was salvaged from an obsolete computer. reactive component at any frequency up to five times the highest test frequency and shall be capable of continuously dissipating the full output of the amplifier while maintaining its resistance within 1 % of its rated value". Temperature problems While it may not be immediately apparent, that is a very stringent specification . While it may be reasonably easy to obtain a resistor with a tolerance of 1 % , getting it to maintain that tolerance while dissipating a lot of power is quite another matter. Most large wirewound resistors will operate with a surface temperature of up to 300°C if they are run at full power without fan cooling. Clearly, you can't afford to have the resistors run up to those temperatures, otherwise their temperature coefficient will ensure that the resistance is well above (or maybe below) its nominal value. Interestingly, the temperature coefficient of wirewound resistors can be positive or negative or maybe both; ie, positive for lower temperatures and negative for higher temperatures. So the first problem is to ensure that the load resistors maintain their value within that ±1 % range up to the full power rating. That means extensive cooling and derating; ie, not running the resistors at their full power rating in order to keep their surface temperatures down. When you think about it, a total rating of 2000 watts is This view shows the internal wiring to the four relays and the 12V regulated supply which energises their coils. in the same league as domestic electric radiators and they get red hot! Even fan forced radiators pump out hot air, so their internal elements run at quite an elevated temperature. Reactive component The other problem is the requirement that the reactive component of the load resistance does not exceed 10% of the nominal value up to five times the highest test frequency. What this part of the specification is saying is that the inductance must not be too high. But because power resistors are "wirewound", they naturally have inductance and sometimes quite a lot of it, relative to their nominal resistance. Typically, the highest audio frequency used for power testing (as opposed to frequency response testing Below: this view shows how the banks of jug elements were mounted on brass rods. The three fans can just be seen behind the elements. "J}ii '··•· liJ. - : :,t9 -•· . AUGUST 1992 63 I A 4n ~ RL1a C 1kW RIGHT CHANNEL RL2a ~ 4Q RL 1b 1kW LE.FT CHANNEL RL2b A~ c-.bo s2a MONITOR 8 ~ o---1o S2b Fig.1: four relays are used to perform the load switching. One resistor bank is used in each channel for 4Q loads, while two resistor banks are connected in parallel in each channel for 2Q loads & in series in each channel for an loads. which is usually done at low power) is 20kHz. Therefore, five times the highest test frequency is lO0kHz and at this frequency the load resistor must have a reactive component not more than 10% of the nominal value; ie, no more than o.zn for the zn range, 0.4Q for the 4Q range and 0.8Q for the 8Q range. So the maximum inductance for the zn range should be no more than 0.3 microhenries; for the 4Q range, no more than 6.4 microhenries; and for the sn range, no more than 1.3 microhenries. These are extremely low values of inductance because even if the resistors had no inductance at all, the inductance of connecting wires would still be considerable. For example, the inductance of a single 1-metre length of 2mm-diameter wire in free space is around 1.2 microhenries. If the wire is curved or near magnetic material such as steel, that inductance can be quite a lot higher. It _ is possible to obtain resistors rated up to 250 watts under forced air cooling conditions, or up to 500 watts or more with water cooling. However these are usually only available with a tolerance of 5%. You can obtain, to special order, non-inductively wound 64 SILICON CHIP resistors with similar ratings and value of tolerance although they are very pricey. For custom wound non-inductive resistors with 1 % tolerance , the price goes through the roof. In fact, for a stereo load box with a rating of 1000 watts per channel in the three load impedances listed above, we were looking at a cost of several thousand dollars and that was just for the resistors. There had to be another way. Our approach was to use a variation of an old idea - the humble jug element. In the past, we have used tapped combinations of jug elements to provide dummy loads for amplifiers. Naturally, they have to be immersed in a bucket of water but they work well. The only problem is that with a high power amplifier, the water soon boils. That presents a real hazard, especially if the bucket is kicked over, as happened on one occasion in our workshop! How to cool it Having decided on using paralleled jug elements, we next had to addres~ the question of cooling. We ruled out water cooling right at the outset because that would mean a substantial tank together with a radiator core and fan. So that left oil cooling or forced air cooling. We ruled out oil cooling because a substantial tank and a finned radiator would again be required. So forced air cooling was chosen by default. We then had to decide how much power a single jug element could dissipate. Our method was to feed current through a single jug element in still air and measure its temperature rise and resistance shift. With SOW being dissipated, the element became moderately hot but stayed below red heat, although the wire began to discolour (ie, turn blue) after 5 minutes or so. For this temperature rise, which we estimated at less than 200°C, the resistance shift was less than 1 % which is right on the button as far as the aforementioned IHF specification is concerned. Based on that, we decided that each jug element should be able to dissipate 100W if forced air cooling was used. Typical jug elements have a resistance of around 36-39Q. That is quite convenient because with 10 jug elements in parallel, we could then obtain a resistance of close to 4Q which would be able to dissipate 1000 watts. Four such resistor banks would be required and the resultant 4Q resistors would be switched in parallel or series to give zn or 8Q. The resistor banks would need to be switched simultaneously for each channel and give a selection of no load, sn, 4Q or zn. Relay switching And this brings us to the next problem. How to do the switching? The currents and voltages involved are quite high. For example, an amplifier delivering 1000 watts into an 8Q load will put out close to 90V RMS. The same amplifier could be expected to deliver at least 40A continuously into zn loads (before fuses blow) and possibly a great deal more on a pulse power test. There isn't any multi-pole rotary switch (that we know of) which can handle the voltages and currents involved. That left us with relays or multi-pole circuit breakers. Ultimately, we decided to use relays, each with two sets of changeover contacts rated at 240VAC and lOA. These have more than adequate ratings as far as the likely applied volt- Starring KT OAT KT 386SX-20 KT 386-25 KT 386-33 KT 486-33 KT 386 NOTEBOOK * * * * • New Costumes • All Australian Support Cast Our Promoter is seeking Dealers wishing to present this talented new range of PC Performers to the Australian Public. Don't miss your opportunity to book your place in this ever expanding dealer network. ,. KT TECHNOLOGY ~~~~m~ iu~1~::~ ~ ~~~~evard, Port Melbourne 3207 Tel: (03) 646 5755 Fax: (03) 646 7997 \ \ The resistor banks were mounted in a rectangular tunnel fabricated from sheet aluminium. Three high-capacity computer fans mounted at the back switch on & provide forced air ventilation when ever a load is selected. Even when running at high power, the exhaust air is cool. ages are concerned but are underrated as far as amplifier current capability tests are concerned. We think that they will do the job but when testing big amplifiers we will have to reduce the signal level before switching ranges, so that they don't have to switch those high currents. The relays are energised from a regulated 12V DC supply derived from the 240VAC mains. The relay switching is arranged so that the loads cannot be connected unless the 240VAC mains is switched on. This ensures that the three fans always run when the loads are connected - we don't want a meltdown while testing a big amplifier. 66 SILICO N CHIP The three fans are 120mm computer fans, each rated at 105 cubic feet per minute. Together, they pull quite a draft. The mains voltage to the fans is reduced to 220VAC which gives a slight reduction in noise while not appreciably reducing the draft. Method of assembly You can see the result of our work in the accompanying photographs. The load box is housed in a large plastic case salvaged from an old computer (from the days when 8-inch floppy drives were standard). This is fitted with perforated steel at the front and the three fans at the back. The small control panel at the front accommodates the substantial binding post terminals for both channels and a rotary switch which controls the internal relays. Forty jug elements were connected in four banks of 10. Their brass connecting wires were removed and they were mounted on 3 70mm lengths of 1/8-inch threaded brass rod. The start and finish of each jug element was soldered to the brass rod. The resulting resistor banks were then suspended in a rectangular tunnel fabricated from sheet aluminium. The brass rod connections to each element were isolated from the sheet metal sides using a sheet of rigid fibreglass suitably drilled. Four relays, wired as shown in the circuit ofFig. l, do the load switching. One resistor bank is used in each channel for the 4Q load condition. For the ZQ load condition, two resistor banks are connected in parallel in each channel, while for the 8Q load condition, the same resistor banks are connected in series. In effect then, the load box could handle 2000 watts per channel while in the ZQ and 8Q load condition, and 1000 watts per channel while in the 4Q condition. However, we think it will rarely be used at powers of 1000W let alone 2000W. Note that the relays provide for disconnection of the loads at each extreme setting of the rotary switch. This is handy when doing measurements such as damping factor and in testing for stability. Switch 2 enables external monitoring equipment, such as a noise and distortion meter, to be connected to the left or right channels of the amplifier under test. And how close did we come in meeting those resistance and inductance conditions described at the start of this article? All three load settings give cold resistance values within 5% of the nominal values and they come closer as the temperature rises. The inductance results were as follows: 0.8 microhenries on the ZQ range; 1.19 microhenries on the 4Q range; and 1. 76 microhenries on the 8Q range. These figures are a little higher than indicated in the IHF specification if Z0kHz is used as the highest power test frequency, although in practice this should not have a significant effect. SC