Hello, I just have a simple question about tubes vs transistors.
According to wikipedia tube 'soft clipping' is actually wave compression. Wave compression causes mostly even order harmonic (musically related) distortion but retains information.
Whereas with transistors 'hard clipping' actually cuts the top of the waveform off and loses all information that was in the chopped off part, while adding mostly odd order information (distortion) not musically related.
_________________________________________________________________________________________
My question relates to soft/hard clipping in relation to RMS vs Dynamic power:
My confusion lies in what exactly 'Dynamic Power' is. Many companies advertise 'Dynamic Power' but do not explain exactly what it is.
For example:
•I buy an amplifier and hook it up to my 'X' source device and O-Scope for testing
•I send a continuous 0dB 1khz sine wave to the amplifier and turn up the gain/volume until distortion is visible on the O-scope.
•I now never go beyond that level to remain distortion-free.
•I now hook up my speakers, and play music at max predetermined 'distortion free' power.
•Will music somehow make the amplifier send more power than a 0dB sine wave? A.k.a 'dynamic power'
I just dont understand how 'dynamic power' can exceed RMS power if RMS is measured with a 0dB signal- which is the maximum signal music can contain. 😕
If this is possible, then valve amplifiers are at a huge advantage as no information is lost during soft clipping to cover the dynamic range. While a solid state amplifier would require tremendous power to cover the dynamic range of music without losing information (hard clipping).
According to wikipedia tube 'soft clipping' is actually wave compression. Wave compression causes mostly even order harmonic (musically related) distortion but retains information.
Whereas with transistors 'hard clipping' actually cuts the top of the waveform off and loses all information that was in the chopped off part, while adding mostly odd order information (distortion) not musically related.
_________________________________________________________________________________________
My question relates to soft/hard clipping in relation to RMS vs Dynamic power:
My confusion lies in what exactly 'Dynamic Power' is. Many companies advertise 'Dynamic Power' but do not explain exactly what it is.
For example:
•I buy an amplifier and hook it up to my 'X' source device and O-Scope for testing
•I send a continuous 0dB 1khz sine wave to the amplifier and turn up the gain/volume until distortion is visible on the O-scope.
•I now never go beyond that level to remain distortion-free.
•I now hook up my speakers, and play music at max predetermined 'distortion free' power.
•Will music somehow make the amplifier send more power than a 0dB sine wave? A.k.a 'dynamic power'
I just dont understand how 'dynamic power' can exceed RMS power if RMS is measured with a 0dB signal- which is the maximum signal music can contain. 😕
If this is possible, then valve amplifiers are at a huge advantage as no information is lost during soft clipping to cover the dynamic range. While a solid state amplifier would require tremendous power to cover the dynamic range of music without losing information (hard clipping).
Last edited:
RMS power is usually based on sine waves.
But with square waves you can get considerably more power out than the sine wave rms value.
Valves are preferred by rock guitarists as they output soft distortion rather than square waves which a bipolar amp will output if clipping is reached.
Square waves sound terrible.
As a guitarist I use a soft limiter into a solid state amplifier and get good results.
But with square waves you can get considerably more power out than the sine wave rms value.
Valves are preferred by rock guitarists as they output soft distortion rather than square waves which a bipolar amp will output if clipping is reached.
Square waves sound terrible.
As a guitarist I use a soft limiter into a solid state amplifier and get good results.
Symmetrical clipping (soft or hard) produces odd harmonics. Assymetrical clipping produces primarily even harmonics. (Probably something wrong with the P-P amp if it does that, although a SE amp probably does that.)
Low feedback generally gives soft clipping, which sounds better, ie similar to compression (getting low odd harmonics). Since tubes generally use lower feedback and suffer more gradual saturation, they clip softer. One could make a P-P Mosfet follower output stage with soft clipping, if the NFdbk were dropped around it. Of course the distortion spec would go kaput too.
It's low harmonics that sound best, not even versus odd. Its just that the lowest one is even (2nd) so they get the rep. for sounding better.
Low feedback generally gives soft clipping, which sounds better, ie similar to compression (getting low odd harmonics). Since tubes generally use lower feedback and suffer more gradual saturation, they clip softer. One could make a P-P Mosfet follower output stage with soft clipping, if the NFdbk were dropped around it. Of course the distortion spec would go kaput too.
It's low harmonics that sound best, not even versus odd. Its just that the lowest one is even (2nd) so they get the rep. for sounding better.
Last edited:
Interesting! Thanks for clarifying all of that. Its nice to see that Feedbackless SET amps actually do have quantifiable advantages.
This may or may not be the case. Tubes will not clip as fast as transistors when run open loop. Including NFB will make any amp clip hard and fast. It's especially bad with solid state, since solid state needs much more NFB to linearize. Other things can also affect over drive performance. This is especially true of finals that are capacitor coupled to a driver. A positive control gris excursion will turn on the GK parasitic diode, and the resulting current flow will charge the coupling capacitors negative, pushing the finals into a less linear part of the plate characteristic until it leaks off. This can degrade sonic performance.
You get a similar phenomenon with cathode bias if the cathode bypass capacitors charge to too high a voltage. This is the "farting out" guitar players sometimes complain about.
"Dynamic Power" is a marketing gimmick. As for what it actually is, only the marketing dept knows for sure. The only reliable metric is RMS power.
Music signals typically spend a good deal of the time at some low level. The peaks can rise 20db or more above the average level. A VT amp can sound louder than an equivalent SS amp since the SS amp needs to stay well out of clipping, whereas the occasional clip isn't heard with a hollow state amp. As for that "dynamic power" nonsense, that's all it is: nonsense.
I suspect what the manufacturers are calling dynamic power may be instantaneous peak power.
A tube amp is generally limited to how much current it can supply by saturation effects in the tube. While a SS amp may be able to supply huge instantaneous peak current and hence power if the load impedance is low enough. The power supply is generally not designed to maintain this, only its capacitance discharge supports it. The transistors are able to provide the high instantaneous current without saturation effects, just short of instantaneous melt-down. Hopefully, some sort of safe operating area limiter in the amp protects it from disaster. This might also be a concern for melted speaker driver voice coils in speakers that have large impedance dips. So the SS amp may be able to drive difficult speakers better, if it doesn't finish them off.
A tube amp is generally limited to how much current it can supply by saturation effects in the tube. While a SS amp may be able to supply huge instantaneous peak current and hence power if the load impedance is low enough. The power supply is generally not designed to maintain this, only its capacitance discharge supports it. The transistors are able to provide the high instantaneous current without saturation effects, just short of instantaneous melt-down. Hopefully, some sort of safe operating area limiter in the amp protects it from disaster. This might also be a concern for melted speaker driver voice coils in speakers that have large impedance dips. So the SS amp may be able to drive difficult speakers better, if it doesn't finish them off.
When you start pulling power from an amp, the power supply Voltage starts to go down, same as a car slows pulling a trailer. Less Voltage means less power. Power is measured with continuous sine waves that cause the most sag of the power supply. For short bursts before the power supply sags, it can temporarily deliver more power. If the power supplies were regulated (more expensive) the RMS and dynamic power would be the same.
G²
G²
I suspect it often refers to the fact that an amplifier will deliver more power for a short period than it will continuosly, because the rails start to sag. You might get 1-second worth of higher available output power, which is actually useful for music with its brief peaks, but you won't see it on continuous sine wave testing.
As Miles said dynamic power is pure marketing but there is a simple musical reason why even order harmonics sound more pleasant than odd order ones.
If we start with a 1kHz sine as the fundamental (aka 1st harmonic) the 2nd is 2k, the 3rd 3k, the 4th 4k, the 5th 5k and so on.
But if we put that into a musical notation using octaves the 2nd harmonic is one octave up from the fundamental but the 3rd is an octave and a half up, the 4th is 2 octaves up but the 5th is 2 octaves and a quarter up from the 1st harmonic/fundamental.
In other words technically they are all harmonically related to the fundamental but musically speaking (ie what we hear as pleasant) only the even order harmonics are.
PS: Going up one octave is a doubling of frequency. The A above middle C is usually 440Hz, the next A one octave up has a frequency of 880Hz and the A two octaves up is at 1760Hz and so on.
If we start with a 1kHz sine as the fundamental (aka 1st harmonic) the 2nd is 2k, the 3rd 3k, the 4th 4k, the 5th 5k and so on.
But if we put that into a musical notation using octaves the 2nd harmonic is one octave up from the fundamental but the 3rd is an octave and a half up, the 4th is 2 octaves up but the 5th is 2 octaves and a quarter up from the 1st harmonic/fundamental.
In other words technically they are all harmonically related to the fundamental but musically speaking (ie what we hear as pleasant) only the even order harmonics are.
PS: Going up one octave is a doubling of frequency. The A above middle C is usually 440Hz, the next A one octave up has a frequency of 880Hz and the A two octaves up is at 1760Hz and so on.
Dynamic power is a SS thing eg short term power (peak), it's the reason for smaller heat sinks and power supplies. Continuous RMS at high levels using sine wave test signals are unnecessarily difficult for commercial amps designed for music. Most SS amps rated for 4 ohm loads would infact burn up if run continuously at max power at low impedances in "Lab." conditions.
The complication is that not only do the voltage rails sag, but they are severely modulated by the demands of current from the output stage in this situation - the supposed DC voltage effectively becomes a very distorted variant of the input audio signal, putting stress on the circuitry's PS rejection, injecting interference in the worst place possible.
Normal PS's are helpless in the face of this, they can't draw energy from the mains fast enough to properly compensate. Only very heavy duty supplies, or interesting variants of a normal supply, or well done SMPS's can overcome this - which is why most amplifiers start to sound sad when high SPL's are asked for ...
This is why many amplifiers have an RC decoupling on the LTP/VAS side to stop ripple/audio getting through.
Wolfram alpha can do some cool stuff with this if you ever get bored.
https://www.wolframalpha.com/input/?i=play+440Hz+tone+++880Hz+tone
There are already such errors in these statements that I believe they can only originate from the Wikipedia "tube sound" article.
It really should be deleted from the face of the Internet.
e.g.
- Any clipping, whether hard or soft, is compression.
- Any clipping, whether hard or soft, removes signal information
- even and odd order harmonics are actually related to symmetry or asymmetry of the clipping, not its hardness or softness
- hard vs. soft clipping is related to amount of harmonics being produced
- tube amplifiers quite often actually clip hard
- many solid-state amplifiers can clip quite soft if someone just designs them to do so
etc.
Last edited:
The problem is that this concept applies only to instances where distortion actually isn't preffered.
When it is, and is designed-in to the circuit, you can throw those rules out of the window. Real life examples of circuit design aiming for "nice" distortion simply fights against all those rules outrageously.
It would be probably good to get rid of that Russell O. Hamm "Tubes Vs. Transistors" -trife as well.
Yet a typical tube power amp output, when clipping distorted, looks very much like this:
Symmetric, squarewavish clipping with plenty of crossover distortion. I can only visualize the astounding amount of odd, high order harmonic distortion of such waveform.
Surprisingly it's the kind of revered tube distortion that guitar amp designers are vigorously trying to capture.
Last edited:
- Status
- Not open for further replies.