Protection devices for Audio Amplifiers
There is a lot of theory put into amplifier design and when one looks at these complicated circuits often the protection circuits are more sophisticated than the gain stages. Purists believe the only true amplifier is a piece of wire with gain. Engineers often criticise the protect circuits of creating their own problems, along with causing inferior quality reproduction.
These are the major problems come across in the design of any amplifier:
- Heat – is there sufficient cooling for heavy duty work?
- Can the output transistors go out of their Safe Operating Area? (known as SOA).
- Can the amplifier drive a 2 Ohm load when minimum load spec is 4 Ohm?
- The dangers of mains voltage boosting and sagging?
- Oscillations and instability.
- Over driving and distortion.
- Loudspeaker protection – expensive lessons to be learnt.
These are just a few of the problem areas in the design of amplifiers.
Heat – is there sufficient cooling for heavy duty work?
Fans that suck! Yes, fans that do not blow on to the heat sink but instead suck air. Ever seen a CPU cooler with a reverse running fan and you’ll know the immediate dangers. Sucking air to keep a cabinet cool or causing a specially designed case to have a designed volume of fresh air moving through it is one thing but not when it comes to heat sinking and the forced air cooling.
A lot more emphasis today is given on audio amplifiers being minitiarised to look cool doesn’t mean it is cool. Heatsinks, moreover the output transistors should be kept as cool as possible. Large aluminium heatsinking is absolutely vital to longevity. Professional duty amplifiers have plenty of this plus fans or blowers. Class D audio amplification has done away with a lot of expensive stuff, large capacity transformers, reservoir capacitors and aluminium or copper cooling. Not so with Class A or anything higher than 25W.
The rule of thumb has always been: As much aluminium as possible.
Can the output transistors go out of their Safe Operating Area? (known as SOA).
In the authors view there is a lot of mystique left in the air when it comes to seeing some of the designs on the internet. The mere fact that a designer has thrown in hundreds of output transistors in parallel to spread the current flow doesn’t mean it’s safe. Just one transistor current hogging because of cooling problems can lead to disastrous results. Emitter resistors which are prone to heating can cause bad joints. This impacts the integrity of the entire circuit.
The biggest complaint in this area comes down to designer claimed powers. Because an amplifier has 10 output transistors per channel does not make it more powerful than one with 4. It means it can drive a lower value load possibly and the potential power is possibly higher into very low impedance loads but this brings up our SOA margin for error again.
Often one will see circuits where the output transistor Vce rating is lower than the supply rail voltages. It may be safe into a resistive load but inductive/capacitive loads often take a transistor out of spec. Transistors need to be kept within their SOA. An example below:
Preferred Device, Complementary Silicon Power Transistors
Complementary silicon power transistors are designed for
general−purpose switching and amplifier applications.
• DC Current Gain − hFE = 20−70 @ IC = 4 Adc
• Collector−Emitter Saturation Voltage − VCE(sat) = 1.1 Vdc (Max) @ IC
Derating is very important in electronic work. In other words, use something which will not fail under most circumstances – this comes at a cost.
Can the amplifier drive a 2 Ohm load when minimum load spec is 4 Ohm?
This is definitely one which is more critical than what is presumed. Tied to cooling and SOA many professional series amplifiers (and automobile) are designed to have 2 Ohm loads. When seeing the mounted output transistors, doing a check of the SOA of these devices and doing the maths they are simply NOT designed for 2 Ohm loads. Especially with the inadequate cooling. To make matters worse they get designed for bridging which at the slide of a switch we get more power but designed for 4 or 8 Ohm and not 2 Ohm. Who follows these rules? I am sure that in 99% of cases where we have premature amplifier failure this is caused by not adhering to manufacturer spec. To bridge or not to bridge: Don’t Bridge.
(Q5 and Q6 protect the output pair from driving into a short circuit – each transistor is arranged to reduce drive when current sensed through the emitter resistor goes above a threshold voltage.)
Bridging can allow for 4 times the power output. Specs indicate double because the manufacturer assumes the user will use 4 or 8 Ohm loads and not 2 Ohms or sometimes 4 Ohms.
And yes, manufacturers do nowadays spec lowest load as being 2 Ohms but not for Bridge Mode. In most instances at the very least the cooling is inadequate unless the manufacturer specialises in the design of these amplifiers. Read the instructions.
The dangers of mains voltage boosting and sagging?
This one is an interesting one only because who checks their household wiring and supply voltage? In the author’s own experience he has come across two instances where an audio amplifier kept burning output and driver transistors because the optimum setting on the mains transformer was 220V but should have been 240V. In both cases the output transistors were 2N3055/2N2955 (and not the Hometaxial). Rails were lifted to over 80V into a four Ohm load when the mains voltage was sitting at 250V. A simple case but where the output stage was taken out of SOA.
In another, maybe not unusual case, a well known professional audio company had one of their class D amplfiers thrown back at them because it would cut out randomly. The company in question argued that it was the mains voltage which was sagging to the point twhere the power supply could not power up or regulate. We won’t go into details of this case, it’s public knowledge that the company went out of their way to assist but the user would not back off.
In the design of any audio amplifier switch mode supplies have taken a route of their own. I don’t like them necessarily for audio use but they are becoming more popular. Your standard EI or toroidal may have it’s own set of rules but are resilient against most mains-borne attacks, SMPS not. Make sure you have anti-surge, anti-spike anti-whatever protection before your expensive Class D equipment.
Oscillations and instability.
The fearsome one. The silent killer.
A well designed audio amplifier should be designed to play into any load type. The VAS and power stages should be well guarded against trying to play audio into the stratosphere, meaning in the RF spectrum. Unfortunately it does happen and when it does happen the results can be catastrophic.
Early MOSFET amplifiers if not carefully designed had a habit of going into oscillation. In many cases DIYers used their own board designed not knowing the full implications of working with high gain stages with very high impedances. Even BJTs succumb to oscillation, especially where not enough attention has been given to feedback and high gain stages where parasitics could be a concern. Modern audio amplifiers are often designed with large power bandwidths.
Although the Phase Linear 400 and 700 series are rumoured to have had instability problems we read more about this now than then. I do believe the company would have had adequate engineering change notices to protect the user and their own good name. High powered equipment can burn down houses, even small ones – when the R&D is dodgy.
Over driving and distortion.
A very popular problem in the home consumer and professional market. We have all at some stage or other pushed an amplifier to it’s limits but very few of us know what dangers lurk below the surface.
One of the major drivers in the over-distortion business is alcohol. Nobody likes the sound of distorted music but when driving an amplifier outside it’s designed for parameters which is essentially what distortion is unless it was designed to distort we are heading for a problem. The typical scenario is clipping and then grand clipping, distortion all the way, nothing resembling the input waveform. Of course the loudspeakers don’t like it and of course besides a hot voice coil we have very hot output transistors. A fuse is the last safety circuit in the link – for it to blow we would be looking at burnt CE junctions or worse, expensive blown MOS devices. In the DJ circuit often the amplifier survives but the speakers don’t.
In a well designed amplifier often the engineers understand the effects of alcohol and they do have current limiting circuits, even circuits which prevent distortion. Often an amplifier such as this seems to get softer the more you turn the volume up. The good news is that just perhaps you have such a circuit which automatically reduces drive when the current is too high or the thermal ceiling of the transistors has been reached.
When manufacturers state that you should get an amplifier rated at twice that of the loudspeaker you should know why. The reality is if you are such a party animal you would need a system which makes your ears bleed.
Loudspeaker protection – expensive lessons to be learnt.
Monitoring DC at the loudspeaker output is not a luxury, it’s an absolute necessity. Some of the older juggernauts did not have speaker protection built in. (Phase Linear 400 is just such an example).
Anti-thump, DC detection, distortion and over-drive detection are all good reasons to have speaker protectors. In some cases the speaker protection circuit protects the output stage of the amplifier as well when your amplifier is ill. Nothing like a dry-joint to cause DC offset.
The circuit above shows one of the most commonly used circuits for anti-thump and DC sense. There are additional diodes and NPN to act as a switch to disconnect load when other abnormalities are detected, e.g. fan speed sense, high temp, current sense etc. The blocks in blue are RC timers: C1 and R6 form the timing network for triggering Q1 and Q2 in the event of DC sense over a period, usually a few milliseconds. R3, R4 and C2 prevent the relay from pulling in until the amplifier has stabilised preventing loudspeaker thump.
Q1 and Q2 through their respective diodes sense either a negative or positive voltage over a set time period before triggering Q1 or Q2 which in turn switches off Darlington pair Q4 and Q5.