An interesting question about a trivial problem appeared on Audiogon recently. The poster asked "why do preamps pop when they get turned on?" As most audiophiles know, one should always turn on the preamp about 30 seconds before turning on an amplifier. When you turn the amp before the preamp we are often treated to rather loud and scary sounding thump instead.
But why is that?
That turns out to be a rather interesting question. There are two reasons:
- Turn-on power supply instability
- DC offset
Protection Circuits
This
discussion is meant to be a generally a fun post about basic preamplifier
behavior, but of course more advanced preamps have a number of ways
to mitigate turn-on problems including output relays, soft start
power supply circuits, etc.
If you are fortunate enough to have a system with these features you will now have some idea of the problems the turn on circuits are trying to avoid. Hopefully even if you have never heard this problem you'll gain some perspective on what happens in the first 100 milliseconds after powering on your gear.
Preamp Design
In the drawing at the top we have a very basic outline of a schematic for a discrete solid state preamplifier. We exclude most parts but focus on the chain between the wall socket on the left, and the output on the right. The amplifier input load, which is outside of the preamp, is represented here by the only resistor on the far right going to ground.
It's worth noting that there are a lot of similarities between a discrete preamplifier and a so-called power amplifier. The power supply and complementary output devices are in fact arranged similarly so if you think you've seen this before, you probably have.
High-end preamps with discrete circuits often have a pair of complementary transistors on the output, represented by Q1 and Q2 here. The actual type of transistor doesn't matter. What matters is that they are each tied to their respective voltage rails. Often the + and - are around 12V.
Turn-on Instability
The first, maybe the biggest problem with a preamp is what happens the instant when the main power switch is closed. We start with C1 and C2 completely discharged. The top and bottom rails start at 0 Volts and eventually reach + and - 12V respectively as caps C1 and C2 reach full charge.
The problem is that C1 and C2 are not charged evenly and simultaneously. When the power switch is closed the current flowing through the bridge rectifier may only flow through the + or the - terminals (or neither) but never both at the same time. This in turn means that the moment power is turned on it either starts to charge C1 or C2. It may even start to charge C1 and then flip, depending on exactly when in the AC waveform you flip the switch.
The point is, that for a couple of AC cycles C1 and C2 will not be charged symmetrically. This means that Q1 or Q2 can't maintain 0 Volts on the output. Whichever transistor has the most voltage will "win" pulling the output towards that rail until it's complement reaches the same voltage charge. This dynamic tug-of-war continues until the voltage rails stabilize, a process which takes in the neighborhood of 50 milliseconds and is seen as large up and down swings on the output voltage. Essentially multiple fractions of the wall socket's AC waveform is making it all the way to the output, though at greatly reduced amplitude. Frightening!
Consider the alternative, let's say C1 and C2 could be charged exactly evenly, something not possible due to the bridge rectifier. So, each rail would grow from 0V to 12V but with opposite polarities in lock step. That would allow Q1 and Q2 and preceding components to remain balanced during the entire turn on process and it would be easy to maintain 0V at C3.
Other potentially good ideas are to add a little resistance between the bridge rectifier and final filter caps, which if large enough would still ensure high current, but not fast turn on. The tube power supplies I've seen do just that, and IMHO this is an excellent solution to power line noise as well as mitigating turn-on charging inrush current.
It's correct to point out that the voltage rail imbalance is more complicated than this analysis. There's a lot of active (powered) parts between the audio input and the output transistors, any of which can add their own complicated behavior to what happens in the first 50 milliseconds after power is applied.
What About Voltage Regulators?
An excellent question!! We did not include this in the diagram, above but preamps often have either discrete or monolithic voltage regulators to keep the rails stable and noise free. The problem is that below their setting they function as purely pass through devices. That is, a +12V regulator won't do anything until the input voltage is at least +12V. From 0 to 12V the regulator does nothing and is wide open. Only once the input voltage exceeds +12V will the regulator start to clamp down on the voltage. Because of this voltage regulators alone don't solve the uneven charging problem.
DC Offset
Another issue for preamps (or really any amp) is how accurately they settle to 0V on the output. When idle, a preamp or amp should have 0V at the outputs. We call this value the DC offset. In a perfect world with no music playing this is exactly 0V relative to the signal ground, ignoring noise which is not DC but AC. It is not uncommon to find that a preamp does not maintain an absolutely perfect idle voltage of 0. Idling at plus or minus 5 millivolts would not be a terrible preamp but excellent for an amplifier.
If we ignore the turn-on issues the next issue which may cause turn-on thumps is DC offset. If the preamp has a capacitor on the output, as is often the case in tube preamps, then the voltage will spike and drop to zero over time, as shown below. This
graph is borrowed from Learn About Electronics. :
The magnitude of the spike is equal to the DC offset, which in a tube preamp can be quite high if not controlled somehow, but it eventually settles to zero. Tube preamps very often do not have complementary outputs, and the signal floats between one high voltage rail and ground, so they don't have exaclty the same problem as a solid state pre, plus they often have a resistor in series with the power supply caps, thus mitigating how big a spike they produce. Sometimes called a pi filter because of the schematic shape. Because of the filter in between the primary and secondary filter caps they can have a slow ramp-on power supply, which is a very good thing.
