China is remarkably diverse in a way that is difficult to describe to people who have been denizens of first world nations, knitted together by rail and telegraph for a century and a half, radio and car for a century, and TV and airplane for 70 years for their whole lives. These technologies only began affecting China, to the extent that pronunciation began to standardize and similar changes to take place, in the past 30 years.
While it's true that 90%+ of the nation are of the Han ethnicity, that classification means about as much as "white" did in the US in 1900; that is to say, it is defined negatively, as the state of not being of any particular, narrowly-defined minority group, and thus, like "white" in the US a century ago, covers truly vast cultural and regional divides. Heck, "Han" Chinese from Guangzhou and "Han" Chinese from Heilongjiang look just as different or more so from one another as they do from most of the minority groups, on average of course. So to say that China would be more diverse if it were less dominated by the Han majority is questionable; that majority didn't arise through genocide or ethnic cleansing but through assimilation on a grand scale, and that process was very much a two-way one.
That said, if you want a more diverse China, one with a much smaller Han majority, without fracturing the country as a whole unrecognizably, IMO the best POD would be to preempt the Han Southern expeditions between 140 and 100 BCE. The Yue peoples of the region were more primitive than China and would inevitably still be drawn into the cultural orbit of a Han Dynasty whose territory proper was confined to the traditional Han heartlands and peripheral territories to the north and west. If the region's peoples follow a tributary relationship, which is likely, that territory (modern South China) will still be brought under direct control whenever an outside people invades and conquers China from the north (the Mongols and Manchus both directly administered many regions which the Song and Ming had considered tributaries, not territories), which means that it will likely be part of China by the modern era, but it will be a more diverse region occupied by larger proportions of non-Sinified and semi-Sinified people, as are Tibet, Xinjiang, and to some extent Inner Mongolia today. However, because South China can and will support much higher population densities, it will much more dramatically affect the demographics of the country as a whole compared to the northern periphery IOTL.