# #03Simplifying digital designs using AKM's build-in A/D converter function

Episode 3. We will introduce the outline of AKM's ADC and sensor products group.

Senior K

Oh, the Lissajous waveform. Are you making an encoder?

Young A

I have two Hall elements from last time. I'm trying perform an A/D conversion... Well, the signal amplitudes of phase A and phase B are slightly off.

Senior K

Yes, the phase of A seems to have slightly larger amplitude.

Young A

I guess the location of the sensor is still not good.

Senior K

It is difficult to match perfectly via analog only. Afterwards it's better making fine adjustments of the gain digitally.

Young A

…I guess you are right.

Senior K

Well..., I saw that you are not good at digital!

Young A

Wow, can you believe it that I'm the only one assigned to this? It must be M Manager's conspiracy!

Senior K

I understand. It's difficult doing everything.

Young A

Doing both mechanical and analog circuits by myself... The addition of digital is quite a stretch for me.

Senior K

Plus, digital circuits are not started easy.

Young A

On previous projects, D handled the digital portion. Honestly, I do not understand it. It is easy to calculate with Excel using the output of the ADC, but it is difficult to drop it into a digital circuit. I am not good at it...

Senior K

Then, AKM's ADC is good news for you!! It has a register function to digitally fine-tune the signal gain of phase A and B!

Young A

Oh, this is really good news. Which products?

Senior K

The AK9232/33/34 for single ADCs and AK9242/43 for ADCs with input amplifiers. Offset errors and gain errors can be corrected independently for L and R channels by the built-in digital function. (Figure 1)

Young A

The AK9242/43 has PGAs in the analog block, but what is gain error correction?

Senior K

It's a digital gain error correction for the ADC output code. The analog gain correction from the PGA is in 3 dB steps, but with this digital correction you can set the gain in 0.0018% steps.

Young A

Very precise!

Senior K

Since the correction is unnecessary with the microcomputer and the FPGA, it is possible to utilize the computing capacity effectively. In addition, angle calculation functions and phase adjustment are available assuming that phase A and phase B signals are input to 2 channels, respectively.

Young A

Angle calculation ...... Does that mean arctangent?

Senior K

Yes, the angle calculation is done by so-called CORDIC.

Young A

Well then, is it possible to output θ without calculation at the output stage!? Why didn't you tell me earlier?

Senior K

Since it takes computation time and capacity when executing angle calculation with a microcomputer, it is effective that this calculation is done in the ADC. Results as little as 2us from the start of A/D conversion to the output of angle calculation results. This is not easy to achieve without hardware implementation.

Young A

Very useful, but anything in particular that needs to be caring?

Senior K

When the power is turned off, register values are initialized. So be careful, you need to store the values in an external nonvolatile memory and incorporate them into the power-up sequence.

Young A

Well, OK, if you say so, registers are quite difficult, aren't they? Incorrect values can be output without proper application. Reliability is also very important for industrial applications!

Expert M Manager

That is also considered!

Young A

My God! M manager!

Expert M Manager

From the viewpoint of improving reliability, this product group adopts data multiplexing and majority logic in registers. (Figure 2)

Young A

Multiplexing and majority logic?

Expert M Manager

When the user writes to the registers, the ADC internally keeps the written information in triplicate. In addition, it adopts a configuration that takes the consensus of the multiplexed data and determines the function of the ADC.

Young A

Well, how is it effective?

Expert M Manager

Even if part of retained data got corrupted during operation, it will not fail since the data is multiplexed and majority logic is applied. (Figure 3)

Young A

I see. It is reliable.

Senior K

In addition, there is an alarm for the situation when all three of the multiplexed data does not match

Young A

How do you use this?

Senior K

When an alarm is detected, registers must then be rewritten, which will resolve the error, stop the alarm, and return it to normal operation state.

Young A

Well, so the ADC will tell you, "There's an error in the internal multiplexed data! The operation will not fail since majority logic is taken, but please rewrite the desired registers as soon as possible because it is dangerous".

Expert M Manager

That's right. By incorporating an alarm function, a fault detection function is added to the ADC itself.