Those exquisite electronic designs that make people applaud

I have always disliked those who blindly worship foreigners, but sometimes I have to admire foreigners in the industry. They also make mistakes and write some messy articles that hurt people, but there are indeed many designs that are so exquisite that people applaud.

I was watching the CAN bus a few days ago. So many devices were connected to a single information bus, all wanting to talk, but without a leader. Wouldn't that be a mess? I found out that they assigned an ID number to each connected device - an ID card with different sizes. Unlimited competition was carried out based on the binary 01 level, which enabled multiple devices to compete for the single bus without a leader. After watching it, I felt wonderful. These foreign devils seem to be smart, at least not dumber than me.

Let's look at the amplifier. To detect the current of a load, one method is to connect a detection resistor in series in the loop. As long as the voltage difference across the resistor is obtained, the current flowing can be calculated. Everyone knows this. But where is the series resistor connected? Is it the high side, that is, the top of the load, or the low side, that is, the bottom of the load? So, I know that there are two detection methods, namely High side and Low side. Both methods have their own advantages and disadvantages: The biggest advantage of low-side detection is that there is almost no common-mode voltage across the series resistor. For example, one end is 0V and the other end is 0.1V, and the voltage difference is 0.1V. This can be directly detected using an instrument amplifier, which is very convenient. But it also has a disadvantage, that is, the bottom of the load is no longer 0V, but 0.1V. If the current is fluctuating, this 0.1 will be unstable, just like standing on the first floor, but the floor is shaking, and the result is that the load is very uncomfortable. You are a detection instrument, you want to detect the current in the load, but you make the load very uncomfortable, just like a doctor makes a patient very uncomfortable, which is a bit bad.

Therefore, a large number of designs use high-side detection. However, high-side detection also has its problems. For example, the load working voltage is 100V. When working normally, the bottom of the load is 0V and the top is 100V. Now you have connected a small resistor in series on the top of the load, with a 0.1V voltage difference, which makes the high end of the resistor 100V and the lower end of the resistor 99.9V (that is, the potential on the top of the load). From the effect, the load is actually very comfortable. It is very stable at the bottom, 0V. Yes, it is a little floating on the top, about 99.9V. We know that general loads are not very sensitive to voltage fluctuations on the top, so it is very comfortable.

But if the load is comfortable, the measuring instrument will not be comfortable. The measuring amplifier circuit must detect the voltage difference between the two wires, which are 100V and 99.9V respectively, and the common mode is 99.95V. Such a large common mode voltage will burn out any instrument amplifier immediately if it is loaded on it.

What should I do?

Foreigners designed a differential amplifier, such as ADI's AD628, the circuit is shown below. It uses two sets of voltage-dividing resistors to divide 100V to less than 10V. The voltage actually loaded to the internal op amp pin is only about 10V, which is safe. However, we found that the differential voltage of 0.1V to be detected was also attenuated by 10 times to 0.01V, so they added a 10-fold amplification at the output end of the subtractor, which not only protected the internal op amp from being burned, but also ensured that the voltage difference of 0.1V was not attenuated, and the output was the 0.1V we needed to detect.

Isn't it wonderful? Actually, it's not wonderful at all, the wonderful part is yet to come.

We all know that shrinking something first and then enlarging it always makes people feel uneasy. Is there a circuit that can achieve: first, resist high common-mode input, and second, not attenuate differential mode?

At this time, I began to admire the foreigners. They designed an AD629, which is the younger brother of AD628, to meet this requirement. The circuit structure is shown in the figure. It is said to be able to withstand a common-mode voltage of up to about 270V and achieve a one-to-one differential voltage detection. How did they come up with this? It seems that their beef is not eaten in vain.

Texas Instruments' INA117 has the same structure as the AD629, and the resistors inside are also similar.