[Audiovista]

Part I: Concept

Part II: Wish List

Part III - Conceptual Design

The way I see it, logical, deliberate and careful design should get me to 90% of the work done. Another 5% will be lab work, tweaking the circuit parameters to validate design and meet the specs. 

Upon validation, 95% of the design work is done, unit performs safe, reliably and within specifications.

Then....last 5% ....you guessed it - listening tests... This is where the product gets its personality (or multiple personalities - we may like more than one option). Friends and I listen to different components, different gain distributions, grounding schemes.... 

It's all about right compromises.... and I rely on the Concept and Wish List to guide me through....

First thing to decide - solid state or tubes?

Considering requirements for high accuracy, small size and reasonable price, solid state wins. Furthermore, for a while now I have been tempted to try some of the popular operational amplifiers ("op-amps") and some of the new ones, designed specifically for high end audio Only a few years ago the category of "high-end audio op-amps" did not really exist, we had to make do with instrumentation or video amplifiers and end result was pretty much hit-or-miss proposition.

Next - active or passive RIAA equalization?

One quick look back, at the concept and requirement for high accuracy and flexibility makes passive equalization a clear winner. Separating gain stages from filtering required to meet RIAA specification allows to treat accuracy separately from gain related issues (gain, THD, dynamic overhead).

Well, let's then see what will the block schematics look like:



Going from left to right we see the first gain stage, which also acts as a low impedance source for RIAA filter and second gain stage which also provides high impedance load for the filter. With low impedance input and high impedance at the output, filter's job of modifying signals as predicted (calculated) is pretty easy.

Then, there is the Offset Compensation circuit. This circuit's job is to cancel the DC offset at the output, so we do not have to use output capacitors. Some capacitors sound bad, some may even sound good (euphonic second harmonic distortion of Aluminum Electrolytic capacitors is not uncommon), but all of them add distortion. Less capacitors, less distortion. Keepin' it simple.

[Audiovista]

Non-inverting circuit will be used for gain stages. Schematic and gain equation are shown below.



The other option might have been an inverting amplifier. Advantage of inverting amp is somewhat lower distortion, due to non linearities of the input capacitance vs. common mode voltage (common mode voltage for inverting amplifier is always zero). In our case this is only theoretical advantage, as our common mode voltages are only few mV (for the Gain Stage 1), or few hundred mV (for the Gain Stage 2). Big advantage of the chosen, non-inverting configuration is very high input impedance. That will make it easy to set input impedance of the preamplifier, by simply adding a resistor from input to ground. This circuit will be used for Gain 1.

Circuit for Gain 2 needs to have provision for offset cancellation. We have two options here:



Both circuits are very similar to the previous one and will provide non-inverting amplification. Circuit on the left will have its gain modified by the extra resistor (R3), while the one on the right side will have its gain unchanged. For this application, the circuit on the left may be a better choice. Voltage source Voffset is not really output offset, but rather voltage needed to bring the output offset to zero. It's real name would be an "error voltage" (more accurate, but never used term may be "error correction voltage"). The error voltage is output of an error amplifier (as we'll see below) and our representation as an ideal voltage source is a bit misleading. Voffset will have a certain series impedance associated with it. It will be in the order of 5-50 ohms (depending on the error amplifier). To put things in perspective - to keep the noise low, R2 will need to have relatively low resistance (1-5kohm). R3 will be in the 20-50kohm range. Series resistance of 50ohm will have much less effect on the overall gain when added to R3, then when added to R2.

It is very likely that this is mostly academic, but we'll stick with the circuit on the left for now. (NOTE: I am very eager to use the other circuit as well, and the next mini project may offer ideal application for this circuit in somewhat different role - it will not be used for offset cancellation but for.... well, we'll see....).

Actual gains of the two Gain Stages will be determined once we have our block diagram translated into complete draft schematics. We are getting close....

[Audiovista]

Before we move on, let's make a small digression. It will help us understand what follows. Operational amplifiers, when used as amplifiers (they have some other uses too), always work hard to maintain zero difference between positive (+ or non-inverting) and negative (- or inverting) inputs. That's all they know. If the voltage on the (+) input becomes higher than (-) input, an op-amp will increase the output voltage hoping to cancel the difference. If the voltage on (-) input becomes higher than that on (+) input, an op-amp will reduce the output voltage. In a properly designed circuit, the (+) and (-) inputs will equalize well before output reaches it's maximum voltage (which, depending on the component, is 0-3V below power supply voltage and is called saturation voltage).

Let's see now what will the Offset Compensation circuit look like. As we said, proper technical term would be the Error Amplifier and simple schematic is below:



Here we have another operational amplifier in action. Output voltage is filtered by the R-C filter. This filter basically averages the output signal. Average of the useful, AC signal, is zero (so it's not being considered for cancellation) and what is left is a DC component, i.e. output offset voltage. This DC voltage is brought to the positive (+) input where it's compared to the voltage on the negative (-) input. Negative input is grounded, so its potential is at zero Volts. Now remember, the op-amp does not want the two inputs at different potentials. Difference of the DC offset (+ input) and zero voltage (- input) is amplified by the op-amp and resulting error voltage Voffset is brought to the negative input of the Gain 2 amplifier. This amplifier then changes it's output DC voltage (without affecting useful audio signal) until the offset error amplifier has voltage on the positive input equal voltage on the negative input (which is fixed and is zero). Other set of R and C adds extra filtration of the AC signal. 

Ideally, the voltage on the (+) and (-) input is equal and zero, while in practice they will differ by the input offset voltage of this op-amp (input offset voltage is a voltage which is below threshold of the op-amp's sensitivity - the op-amp considers this voltage to be zero; the offset voltage can be positive or negative and changes with temperature). This offset can be from 0.1-10mV, depending on the op-amp used. Even 10mV is acceptable and not that bad, compared to what would offset be without the compensation (about 100 times higher). We will pick an op-amp with reasonably low input offset voltage of 1-2mV max. Spending a lot of money here would be wasteful (there are no sonic contributions of this stage).

[Audiovista]

Part III now on the website.

http://audio.engineeringvista.com/AudioKits/Phono-1/Design/Part3.htm