1-800-PSAUDIO

# Chips vs. discrete

May 29, 2018
by Paul McGowan

In yesterday’s post, I explained the difference in terminology between a discrete component op-amp and an integrated circuit op-amp was the way they were manufactured and packaged. The circuit itself could be identical. But they are typically not.

An op-amp is a functional amplifier that always has two defining blocks: an input differential pair and an output gain stage. There’s often a third block added, an output buffer. Let’s look at yesterday’s drawing.

The Diff pair is the input and the feedback point. Any signal put into one or both of these two inputs will be amplified if there is a difference between the two. The gain stage takes the output of the diff pair and does two things: it amplifies the AC signal and gets the DC where we want it. (The DC at R1 is very high, nearly at +V. We want this taken back down to halfway between +V and -V—which is where started at the input). Once the signal has passed through the diff pair and gain stage—larger now for AC signals and the DC is equal on input and output—we can then wrap the output signal back to the – in through a few resistors to set the gain. That is how we apply negative feedback.

Of course, there is much more to learn about op-amps but these are the basics of its operation.

Understanding these basics, Stan and I realized a few things about the limitations of the integrated version of this circuit. The first is rather obvious. Because the IC op-amp is encased in plastic we could not choose what types of devices were used, nor what the values of the resistors and other parts were. But worse, and this was a big deal for us, we could not change the + and – voltage. IC op-amps have limits which are typically +15 and -15 volts—parameters that could not be changed in an integrated package.

Tomorrow I’ll explain why we needed more voltage and what we did.

Subscribe to Paul's Posts

### 7 comments on “Chips vs. discrete”

1. mjbrenneman says:

Could you please explain what the diodes are used for? I understand the rest of tbe circuit but have never seen diodes being used.

1. Without them the output is class B, there is a region of about 1.2v where neither transistor conducts, and you get crossover distortion. By putting two forward biased diodes between you introduce a 1,2v offset between the bases, the output becomes class AB, and the distortion is much reduced (although not eliminated entirely).

In this case the diode’s rectifying capability is not being used. They are just a cheap and easy way of getting a voltage offset which corresponds to the turn on point for the transistor. They are particularly effective in ICs, since they will be at the same temperature as the output transistors. In a discrete circuit they should really be mounted on the output heatsink, since offset bias varies with temperature.

A slightly better, but more expensive, approach is to use a ‘rubber diode’, a transistor and two resistors, to provide the offset. The resistor values allow you to determine the exact offset you want. This also gives better thermal matching possibilities.

1. The last method Chris refers to is called a vbe multiplier and is what we have used for years on power amplifiers to set their bias. Making one of the resistors a pot allows easy bias set and if the transistor is mounted on the same heatsink as the output transistors then, as Chris mentions, you get thermal tracking as well.

2. Audiomano says:

Apparently the topic today, has not been of interest to the assiduous subscribers to this forum, which by the number of comments it is concluded that those who know a lot, are not interested in the subject and even less to those who do not want to know of the components that are inside their devices, for which many times they have to pay juicy sums of money.

There have been talented designers who have used ICs with excellent results, they have achieved an exquisite sound (no noise) in their amplifiers, together with the obvious simplification of the circuits, which is very useful for users, since it allows them to repair by they themselves failures that can be presented, that like any electronic device is never free of them.

I put the case of the IC: LM-391-N-100, which is a preamplifier that delivers the signals to the drivers: D44C11 and B633 which in turn deliver the amplified signal to the power transistors 2N5686 and 2N5684 (50 amps and 300 w ea). respectively. This is a circuit so simple that any hobbyist can dare to repair it.

On the other hand is the complete preamplifier (MM & MC) designed by Allen Bradley: the C-21 consisting of several tens of discrete transistors.

Both products sound extraordinarily well and here comes the question: Why some audiophiles “purists” detest the devices with ICs, if being in the digital age almost all CD players, DSPs and DACs use them and nobody disgusts them as before ? The TL-702 is still used today after several decades.

I think that’s the topic of today: “Chips vs. discrete”

3. jeffstarr says:

Paul, just because many of us have not commented, does not mean that I don’t find this interesting.

1. Longplayer says:

I agree. It is more difficult understanding a schematic that does not look anything like the actual parts than say looking at a cut-away or exploded diagram of an engine where the diagram actually looks like the parts of the engine, but it can be interesting, nonetheless…at least at a basic level.

1. jeffstarr says:

I like flow charts too, although I can follow a simple schematic.

Founder & CEO