Quote:
Originally Posted by tubes /img/forum/go_quote.gif
Dan, maybe you could comment on passive I/V conversion vs opamps as it is thought that no opamp is the best opamp. Also, what are your thoughts on the use of Pulse Transformers at the input of a dac?
|
Pulse transformer: They are there for a reason. In fact, pro gear (AES/EBU) electric connections REQUIRE a transformer in the signal path. I am referring "electric" as opposed to say optical connections where the signal is isolated between driver and receiver by optical means (light pipe).
So why a transformer?
First, it is a way to isolate gear (source and destination) grounds. When you have 2 pieces of gear (separate chassis), the ground potentials may be different. If you connect the grounds (such as for signal return path), you end up with potential ground loop currents. That un intended activity "adds to" the signal itself. Often it is line frequency or twice the line frequency, but it can be due to other sources as well. The transformer "breaks" that direct connection between grounds.
Second, when using transformers, one can (and really should with a good circuit) take advantage of the enhanced common mode advantage they offer. Transformers is one of the ways to achieve "balanced connection". One makes both sides (there are 2 wires in your cable) "look the same". give The wire pair in the cable already "looks the same" (they are in the same space, nearby and parallel to each other). If you happen to pick up an electric disturbance, and both sides pick the same amount, the transformer does not pass that energy from the input winding to the output winding.
Why? If both sides of a winding "jump up" by 100V simultaneously, the voltage DIFFERANCE between both sides is un effected, so the current stays the same, and there is no change in the magnetic filed, so the secondary coil is not impacted. We call it common mode rejection - common mode for both sides impacted together "in common". There are also some down side to using transformers. A lot of it is about good circuit design. As a rule, pulse transformers is a good thing, at the hand of a good designer.
The I to V question will take a very long answer. many of the better DA's internal circuits are based on steering currents, not voltages. Well, the end result may be a DA circuit that has an output where the signal is given as a current source, and not a voltage source. So one needs at some point to convert the current (I) to voltage (V).
When you use a passive circuit, you are up against a number of problems. You need to come up with a current source that is capable of operating well (no distortions) while the voltage across the circuit changes all the time (with the signal). When you use an OPamp, the current source is working into a virtual ground, but now you have to deal with whatever limitations the OPamp has. To answer that, one has to look at specific implementations and circuits.
At this point in time, I would say that trying to save an OPamp is like looking at one tree in a large forest. The audio signal first appears at the output of the mic, and by the time you hear it, chances are that it passed through dozens and possible hundreds of circuits such as an OPamp. In fact, inside the micpre, the AD, the DA, the power amp... there are a lot of such circuits. You do not see them, unless you get into the inner details of circuits such as AD and DA, but they are there.
Folks are confused. An Opamp is mostly about PAKAGING of an electronic circuit, mostly for lower current voltage and power applications. The discrete parts circuits we used to make out of transistors and FETS, we can get prepackaged, with much better components matching and so on. Yes. opamps have limitations, but when a designer makes sure NOT to go up against the limitations, you end up well.
Also, there is that false notion floating around that less parts is better. Often such is the case, but there are huge number of cases where the opposite is true. Say you want to amplify by 1000 (60dB). Say you single device can only do a good job of it up to a gain of 10, and then it really falls apart. The intelligent solution is to use 3 devices in series (10 X 10 X 10 =1000). But someone in this crazy audio industry will sell a one device solution with notion that with one device you are doing less to the signal. It is not the "device count" that matters to what you do to the audio. It is the end result that counts and often it requires more devices.
Simplistic thinking yields to very wrong conclusions. Also, the subject of "good opamp" vs. "bad opamp" just wrong. There are opamps that are bad under all conditions, but as a rule, an opamp has to be fitted to a circuit and the circuit fitted to the opamp. Say opamp A is good at circuit 1, and opamp B is good at circuit 2. You switch opamps and both circuits may be real bad. So which opamp is good or bad?
When folks hear an opamp in one or very few locations of limited numbers of specific designs (often one location in one design), they should not generalize that it is “good” or “bad”. Such favorite pass time ends up with much spread of questionable "findings", and many get miss informed. My advice, do not be "active circuit shy". This day and age, a passive I/V still needs to be buffered by a number of active circuits (opamp or not) before it gets to your ear. You can not connect directly to a passive I/V without causing much damage to the signal. The question is how good of an I to V can you make. It is not about saving a single opamp, which is usually the argument for passive.
I do have to “hold back” a lot of sensitive information from potential competitors. I hope what I said is helpful to you.
Regards
Dan Lavry
