TOA VM-1120B Amplifier

 A customer brought me a faulty public address amplifier for repair. Not easy as I couldn't find a circuit diagram.

Here's the manufacturer's information
VP-1120B Power Amplifier (120W)
120W RMS output power.
Wide 40 ~ 16,000Hz frequency range.
2 program and 2 priority inputs with program input muting during priority operation.
Operates on both AC mains and 24V DC.
Rack-mountable with optional TOA mounting accessories (VP-1060B / VP-1120B: 1 / 2 3-unit size / VP-1240B: 3-unit size.

Specification
Output Power 120W RMS
Frequency Response 40 ~ 16,000Hz ± 2dB at rated 1 / 3 output
Distortion Less than 1% at rated output (f=1kHz sine wave)
Inputs 2 program inputs (parallel); 0dB 100k O (balanced) / 2 priority inputs (parallel); 0dB 100k O (balanced)
S / N Ratio (20 ~ 20,000Hz) 80dB
Indicators Green LED for power indicator / Red LED for priority indicator
Output and Load Impedance 100V / 83 O, 50V / 21 O, (70V / 42 O, 4 O)
Power Consumption At no signal, 29VA / At rated output 350VA
Power Requirements 220 / 240V AC, 50 / 60Hz, 24V DC
Dimensions (W x H x D) 216 x 144 x 340mm
Weight 10.3kg
Colour Black

Below is the example received for repair, clearly mechanically rather different. This is puzzling because there's nothing on the case to indicate anything other than a model VM-1120B, yet it's completely different. For example the amplifier shown below measures (W x H x D) 430 x 120 x 320mm (excluding the 19" mounting brackets). How on earth does the manufacturer handle factory change processes? Presumably through serial numbers or an in-house code numbering system?

 
 The amplifier arrived here with the usual "not working" fault description leaving me to dismantle it and carry out some tests before proceeding. Stripping out the main circuit board involved detaching the power supply rectifiers, front panel control board and the large heatsink to which is mounted the main board. Below is a picture of this assembly.
 

 What you see above is the board reassembled after repairs. Circuit checking has to be carried out with the board detached from the heatsink. Three screws hold the board in place and seven to secure the transistors to the heatsink. I found that one of the power transistors was short-circuited. The original types were 2SC2987 (click to see their spec) which has ratings of 120W, 140V, 12Amp. I replaced all four with type 2STC5242 (click to see their spec) which has ratings of 150W, 230V, 15Amp. Both have very similar gain figures, which is very important. Why change all four you might ask? Well the transistors are connected in parallel push pull. A pair connected in parallel as an emitter follower to one side of a transformer primary and the second pair connected to the other side of the primary winding whose centre tap goes to ground. The transformer is a universal type having 100V and 50V outputs connected to the rear termination panel plus other unused outputs.

Along the circuit board you can see a row of ceramic resistors. The outer pairs are used to balance the currents to the paralleled transistors. Without these ballast resistors one transistor might hog most of the output current and the resulting unrestricted unbalance might destroy it. The balancing is necessary because transistor gains are not readily guaranteed. Gain will vary at different collector currents, and for parallel connected transistors the one with the highest gain will draw more current than its partner resulting in it getting hotter. The increase in temperature of its semiconductor junction will increase its gain and very quickly thermal runaway will result in meltdown.

I tested the new transistors and matched them from the batch I'd bought as well as I could. Using an external power supply set to 24 volts and current limited to an amp or so I checked the voltages across the ballast resistors. These are recorded on the chart below. The ballast resistors are 0.2 ohm.

 Total Current

 Q5

 Q6

 Q9

 Q10

 223mA

 3.8mV

 4.4mV

 4.3mV

 3.9mV

 302mA

 7.5mV

 9.3mV

 8.6mV

 7.6mV

 354mA

 9.8mV

 12.6mV

 11.3mV

 9.9mV

 354mA

 10.3mV

 13.6mV

 12.1mV

 9.9mV

 675mA

 10.8mV

 15.5mV

 12.3mV

 9.8mV

 675mA

 11.2mV

 16.8mV

 12.7mV

 9.6mV

 675mA

 10.8mV

22.7mV

13mV

8.4mV
 You can see exactly what's happening. The total power supply current represents the amount of audio drive which is kept at a very low level because I'm operating without the heatsink. Where there is more than one reading at a particular current the increase in voltages represents heating. Only a minute or so was left between readings and at the end of the short test Q6 was too hot to touch even though its dissipation was only of the order of 3.5 Watts. Clear to see are the effects of both thermal runaway and current hogging as Q6 is extracting current from its partner Q5 as is Q9 from Q10. At full output these readings would be around 350mV. To help keep the amplifier protected I connected the thermal monitors to the transistors having the highest gains. These being the inner pair.
 

 Preliminary testing was carried out using a laboratory power supply providing 24 volts limited to 1 Amp. The picture above shows the amplifier's power supply. On the left is the mains transformer providing about 24 volts RMS, rectified by the lower of the two bridge rectifiers on the right and the capacitor just visible on the left to 36 volts DC. The output voltage drops under full load to 24 volts. The upper bridge rectifier on the right is connected as a single diode and is used for connection to a 24 volt standby battery, used in the event of mains failure. On the right of the mains transformer is the output transformer.

At its rated power output of 120 Watts I reckon, at 75% efficiency, the DC current requirement is about 7 amps. A 10 Amp fuse is fitted in the emitter circuit.

 

 Above is a view of the chassis. In this version, used for announcements at a ferry terminal, there is no provision for inputs other than that for a pre-amp for a microphone. The output from the pre-amp meets the rear socket input spec of 0dBV which represents an output of 120 Watts from one volt RMS of audio. This input socket is labelled "Priority Input".

The pictures below shows the repaired board refitted into its confined space. Surprisingly there's no cooling fan fitted. On the heatsink is an overtemperature sensor.

 
 

 Here are some details of the repair. Transistor Q6 was short-circuit. I fitted a pair of 2SC3263 transistors and returned the amp. The replacements are of roughly the same vintage as the originals, however the thing was returned with the message that it still wasn't working. I found everything was OK the second time it arrived on the bench, but for some unknown reason Q6 failed again during testing. The initial problem I had was that I had no information about the input equipment. Although I'd checked the amp and thought it was working, I hadn't measured the power output. Normally I get faulty items with only one reason for failure and it was reasonable to have assumed that in this case a transistor had failed.

Fitting a set of four new transistors as I mentioned at the start of this story was now not the end of the job. why didn't it work the first time round?

At the rear is a priority input socket which is supposed to provide an audio input that over-rides other channels. OK, but there are no other channels fitted so the priority input is the only input. The DIN connector is marked to show a pair of inputs plus a voltage of 24 volts and a feed to which this voltage can be routed. Connecting these terminals operates a relay whose terminations are carried to the rear panel. I had in mind that these connections where needed to prioritise the input but testing seems to indicate that this is not so.

I applied a dozen millivolts of audio to the input and traced its progress. It passed through an op-amp then a preset volume control (under the socket) then went to an 82kohm resistor from which it didn't emerge. Something was intercepting the audio and preventing it getting through. Around this area of circuit board is a digital transistor so was this involved? I turned it off and its collector went from zero to 24 volts but it didn't remove the muting. I then noticed a zener diode whose leg was cut. Is this a case of sabotage I wondered or a factory mod? I joined up the cut ends and... no luck, the muting was still present.

What next? As the circuit board was screwed to its heatsink, all I could do was make measurements at component legs, which I proceeded to do. Now, I actually had a clue to the problem. I'd noticed that turning off the amp resulted in a very brief burst of output. This also happened for a briefer moment when turning it on. This burst seemed to be progressive rather than an abrupt or instantaneous effect, like a capacitor charging or discharging.

An old favourite of mine is connecting a resistor on a lead to ground to various points on a circuit. A marginal condition can sometimes be altered one way or another by this means and an 8.2 kohm resistor is likely to be quite safe as regards potentially damaging results. I found an odd voltage which measured about 8.5 volts and, touching the diode cathode carrying this with my 8.2kohm resistor, brought the amplifier to life. The diode connects to a 47uF capacitor which measured as OK but could the diode be leaky resulting in too high a voltage on the capacitor Unlikely because it's the cathode so the problem must be too high a voltage at the diode anode.

I suspect, all things considered, that we're looking at a fault indication which is muting the amp. If 8.5 volts represents an over-temperaure indication from one of the three monitors (two transistor plus the heatsink) then the audio would be cut off until things cooled down. Alas, with no circuit diagram I'm going to take a gamble. I soldered the 8.2kohm resistor in place. The voltage now measures 6.5 volts and there's no muting. Maybe the amp will now work when re-installed at the ferry terminal?

 Return to Reception