AFE (analog front end) is analog front end in Chinese. It refers to the battery sampling chip in BMS, which is used to collect battery cell voltage and temperature, etc.
Before writing this, I thought about it for a long time. There are too many things to write about AFE, and one small thing can be discussed for a long time. In the end, I decided to start from one problem and introduce some AFE related things in the middle, so that the content will not appear too rigid and dogmatic.
The question is: How to choose a suitable AFE?
The location of AFE in BMS (picture from TI official website)
According to normal thinking logic, of course we should analyze our input needs and then make a choice; in reality, the above statement is only half right, because there are not many AFEs for us to choose from, and all needs must compromise with reality; it's like a chef who can make all kinds of flavors to meet customer needs, but has only one ingredient, tofu, and is always a little stretched.
The above figure is the internal functional block diagram of ADI 's LTC 6813. The internal structures of AFEs currently available on the market are similar, and the differences are more concentrated in hardware resources; another is the architectural form designed for functional safety requirements. Simply put, the main differences are the number of sampling channels, the number, type and architecture of internal ADCs (about ADC, we can discuss it separately later).
The external input we obtain needs to be mainly in two parts: domestic and international standards and customer needs. There will definitely be overlaps between the two parts, which requires us to distinguish them ourselves. The good news is that generally, customer needs are stricter than standards.
The main standard that BMS can refer to is QC/T 897-2011. Since the updated version has not yet been released, let's talk about this one first. The most important requirement for AFE is acquisition accuracy (as shown below), which is our bottom line. The indicator requirements in the standard are not very strict, and the test conditions are written very vaguely. The latest discussion draft is much better. The measurement and verification of AFE voltage acquisition accuracy seems simple, but how to achieve an accurate voltage reference source, especially in EMC testing, is a question worth discussing.
There are many demands from customers, and the main factor affecting the selection of AFE is the configuration of the battery module . For example, how many series and parallels does the smallest module have? How many modules are there in total? Then there are some detailed requirements, such as sampling accuracy, number of temperature points, functional safety requirements, etc.
In real applications, for the cells connected in parallel, we sample them as the same cell; further, the cells are basically connected in parallel first and then in series to reduce the demand for sampling channels; but what makes us headache is that the number of cells connected in series in a module is not fixed, and the total number of cells connected in series is not uniform. In this case, we need to match the number of cells in each module. Even more unfortunately, the maximum number of voltage channels of AFE is discontinuously distributed (mainly divided into 3 levels: about 6s, about 12s and about 18s), so we need to carefully select the appropriate number of channels for matching, so as to avoid waste and force. Therefore, especially for third-party independent BMS manufacturers, because of their limited voice, sandwiched between the OEM and the cell factory, it is difficult to make a platform version of the product.
It also involves a special scenario, which is sampling by connecting the AFE across the module (as shown below). Simply put, the voltage drop on the copper busbar is collected together; we may use an independent channel to collect it, or put it together with the battery cell to collect it. In either case, it involves a negative pressure sampling problem, which requires the AFE sampling channel to withstand negative pressure. At present, there are not many manufacturers who can do this, and many manufacturers' products are evolving in this direction.
In addition to sampling accuracy, what is easily overlooked is the number of temperature sampling channels of AFE. The recommended ratio of temperature channels to voltage channels is 1:2. Do not make it smaller, because it may cause insufficient temperature channels. In addition to collecting external NTC, there may be some analog signals in the board, so temperature sampling channels are a relatively tight resource.
There are some automotive AFE resources available for us to choose from (see the table below). Some of the information may not be accurate, and some are still in the sample stage. They are for reference only. It categorizes US and non-US devices. In fact, NXP and Renesas have a close relationship with the US, which is a bit unclear. China's AFE started a little late, and we can see that there are almost no AFEs in cars.
In summary
, I have briefly introduced several main criteria for selecting AFE. In fact, if the company's platform has been established, it is difficult to select a new AFE, and it can only evolve repeatedly on the old platform. For engineers, taking mature circuits and using them directly is both a good thing and a bad thing. In the long run, it is best to go through the process from 0 to 1.
|