Hi All.
I am trying my hardest to find a good enough reason to bias an opamp like the LT1115 into class-A (according to an RIAA preamp on the LT1115 datasheet).
The distortion figures of most modern opamps are already vanishingly low. What is the advantage then of biasing these into class-A? I have tried to build a small circuit which does this using a BF245 Jfet. On my scope, I see no difference between the circuit in (2mA draw) and not in.
Can someone please help me understand..better still, are there any before after scope measurements? ALSO... lastly, does it improve the audio output significantly enough to justify adding 2 components to boas the device?
regards Robert
I am trying my hardest to find a good enough reason to bias an opamp like the LT1115 into class-A (according to an RIAA preamp on the LT1115 datasheet).
The distortion figures of most modern opamps are already vanishingly low. What is the advantage then of biasing these into class-A? I have tried to build a small circuit which does this using a BF245 Jfet. On my scope, I see no difference between the circuit in (2mA draw) and not in.
Can someone please help me understand..better still, are there any before after scope measurements? ALSO... lastly, does it improve the audio output significantly enough to justify adding 2 components to boas the device?
regards Robert
It will probably reduce distortion even further.
Normal "vanishingly low" figures come from MASSIVE NFB, biasing into Class A goes to the source of the problem.
Watching a scope screen is an incredibly crude way to detect distortion, will only show gross deviations, say 10% or plain old clipping, triangle wave slew limiting, etc.
I bet that preamp designers relied on measuring distortion with some sophisticated device.
Normal "vanishingly low" figures come from MASSIVE NFB, biasing into Class A goes to the source of the problem.
Watching a scope screen is an incredibly crude way to detect distortion, will only show gross deviations, say 10% or plain old clipping, triangle wave slew limiting, etc.
I bet that preamp designers relied on measuring distortion with some sophisticated device.
I agree, you cannot see such a small amount of distortion on an oscilloscope waveform trace.
You can measure such low distortion without too much expense, get a 24-bit audio-to-USB interface for your computer (in the $100 range), recording software (such as Audacity, free software) and some FFT analysis software (I recall Audacity has a basic version built in). You need a sine wave source that has less distortion than you're trying to measure. Audacity can generate an audio track with a sine, and play the track while recording the interface input, but I don't know if the sine would have low enough distortion. You could check it in a loopback test.
Three of the example circuits on that data sheet show a FET on the output sinking 2mA (for forced Class A, but there's no mention of that in the datasheet text), so I wonder if the distortion spec is with or without it.
You can measure such low distortion without too much expense, get a 24-bit audio-to-USB interface for your computer (in the $100 range), recording software (such as Audacity, free software) and some FFT analysis software (I recall Audacity has a basic version built in). You need a sine wave source that has less distortion than you're trying to measure. Audacity can generate an audio track with a sine, and play the track while recording the interface input, but I don't know if the sine would have low enough distortion. You could check it in a loopback test.
Three of the example circuits on that data sheet show a FET on the output sinking 2mA (for forced Class A, but there's no mention of that in the datasheet text), so I wonder if the distortion spec is with or without it.
By the way, an LT1115 is a very poor choice for a moving-magnet phono preamplifier because of its 3.4 pA/√Hz input noise current (when measured under realistic conditions). It is suitable for moving coil.
The datasheet claims that the noise current is 1.2 pA/√Hz, which is already rather high for moving magnet, but that is only with equal impedances driving both inputs. You can calculate the noise current under normal conditions from the Total Noise vs Unmatched Source Resistance graph.
The datasheet claims that the noise current is 1.2 pA/√Hz, which is already rather high for moving magnet, but that is only with equal impedances driving both inputs. You can calculate the noise current under normal conditions from the Total Noise vs Unmatched Source Resistance graph.
Last edited:
Lt1115 datasheet shows that 2mA current source loading the output for more circuits , not just phono circuits.I always felt it is just a trend to add current sources to load operational amplifiers, not sure who started this, probably Nelson Pass, but they should get warmer thus noisier and thermal noise is not smth you want for phono preamplifiers so I'd say it is a bad ideea to add 2mA current sources to op amps in phono or microphone preamplifiers, but I can't deal with N.P. aura protecting the ideea .Apparently it is a very good circuit for mc carts and the mm cart use comes as a bonus even though not ideal.A friend of mine uses lt1115 for mc carts and his system is based on b300 tubes and the most expensive Tango transformers so he might know something about it.
Op ams can and do have crossover distortion. It’s quite low most of the time, but load one heavily enough and those high order harmonics start creeping up. Those EF buffers will make any op amp drive low Z lines like 5532’s can - without crossover distortion. Also can drive headphones directly.
You are right about the noise increase, but the effect is small. Suppose the op-amp runs on +/-18 V, draws 2 mA extra from its positive supply and the normal current from its negative supply, and has a 200 K/W thermal resistance from junction to ambient (it can't be more than that, or the chip would overheat due to its quiescent dissipation). That will increase its junction temperature by 2 mA times 18 V times 200 K/W = 7.2 K. Under the same assumptions, its temperature would be roughly 350 K without the current source, so the absolute temperature increases by just over 2 %. The noise of a resistor then increases less than 0.1 dB, the voltage noise of a bipolar transistor biased at a temperature-independent current less than 0.2 dB.
I read that forced class A was first introduced by Walt Jung in his audio op-amp books, but I think those were all published years after 1972, but his books likely popularized the idea.
The hifisonix article brings up a good point, forced class A eliminates the half-wave (distorted) current pulses in the power supply rails. This also points out the importance of good power rail bypassing.
The hifisonix article brings up a good point, forced class A eliminates the half-wave (distorted) current pulses in the power supply rails. This also points out the importance of good power rail bypassing.
Thank you! I wouldn't know that.Without clear knowledge about such calculations I tend to be religious about cold electronics.