 Hi, on MPI. Hi, on MPI, brought to you by Digi-Key and everything. Digi-Key, RichTech, is it? That's right. You're a power partner. You're a power partner. That's right. We've covered them before. This is their second time on the show. I think we did an LED driver from RichTech before. They do a lot of automotive, a lot of industrial. This week, we're going to be looking at an I2S audio amplifier. I do love I2S. This is a similar chip, but they didn't have an image of the package. But this is the RT9120S, particularly S. And this is an image of it on the eval board. So you can see it's like a 24- to 28-pin QFN. And it's a very powerful I2S amplifier. Despite this being a tiny little chip, this is a... Actually, this is sorry. This is the wrong screenshot. It's a 30-watt, not 20-watt, because it's the 90120S. It's a 30-watt stereo inductorless I2S input, class D amplifier with stereo or mono output. Like halfway through, I realized that there's the 90120 and the 90120S. And we're actually talking about the 90120S. That's the new version of this chip, which is more powerful. It's pin-compatible. So if you have been using the 90120, just FYI, it's the same, but better, because it's got better specs. So the 90120, in general, is an I2S amplifier. So it's digital input and analog output. It drives a speaker directly. It can do stereo or it can do mono. The I2S signals can be 1.8 or 3.3-volt input. But for the power to the amplifier part, you'll need 5 to 26 volts. And I like that, because it's a nice wide range. If you don't need the full 30-wats, you can do less than we'll show the power rating. You can do your 12 volts or less. But if you want to use an automotive or industrial system with 24-volt power, you can power it with 24 volts. No problem. And because it's class D, you actually have very little power dissipation issues. Like most people who are used to amplifiers that are dealing with 30 watts, you're going to have a gigantic heat sink on your class AB. Not necessary here, because it's a class D amplifier. So it does. I think they have this spec here. It's like 1.5 megahertz PWM. I think it's either 75 or 1.5 megahertz. And this chip can also do a variety of different inputs. I2S, right justified, left justified, and as well as TDM. OK, so this is actually made a little mistake. So what we're talking about today is actually the RT9120S. You can see that they are almost identical, same 32 QFN. The 90 120 is the older version, 20 watt max. The new version is 30 watt max. But otherwise, they still have stereo output, 4.5 to 26 volts. I believe they have almost identical I2S command set. It's also the 21S is a little bit more efficient. It's got 94% rather than 92%. And it supports TDM, whereas the previous version did not support TDM. But there's also other power amplifiers, by the way. So it's like, oh, you want higher wattage, lower wattage. You can do a wide range from RichTech. We're just going to be only talking about the 9120S here. But they do have tons. So how do you control it? Well, you can give it I2S input. And one thing, and then I2C to control it. So you will need a driver. It doesn't have a free one mode. You can't just pipe I2S in and have it just automatically output audio signal. But the I2C is at least well documented. There's a full register map inside. And then there is a fault output. And there's some power supply stuff. But it's actually a fairly simple, what I like is there's not a lot of extra components needed. You don't need an extra crystal. You don't need M clock input. You don't need a PLL, extra capacitors and inductors and whatever. It's very much like data in audio output. So one thing that I do like about this is that if you wanted to use an audio amplifier, a digital audio amplifier with the Raspberry Pi, the Raspberry Pi doesn't have an M clock output. And so you want something that can take the bit clock and PLL it up. And then it can calculate how much the main clock frequency it needs based on the bit clock and does it automatically. So there are existing I2S chips that do that. But many chips require an M clock. What I really like about this one, if you go back one, it doesn't need it. It just needs left, right clock, bit clock. And then you see there's SDI and SDO. That's because it can route the audio output. It doesn't actually have a microphone input, even though there's a data output. You only need three digital inputs to control the I2S. OK, typical use case. So like I said, it's actually, I like how simple it is. Because it's a class D, and especially if you're using longer wires, you're going to want that LC filter. In the documentation, they do mention you can use ferrite beads. You don't need to use big inductors. But the eval board does use fairly large 10 micro Henry inductors and capacitors to make a filter. Probably for EMI reasons, depending on your product, if you're going to be doing EMI testing, you might be able to get away with shorter, if you're using very short wires, non-juice power inductors, you can maybe use ferrite beads. But that's on your design requirements. So probably you can use either. But because this is designed maybe for stereo systems that go in automotive or whatever, they're expecting to have a lot of noise going to be coupled in or that it could couple into something. So they want to be extra careful on the output to make sure that they don't get that 1.5 megahertz PWM square wave emitting all over the place. You can also, it turns out, connect it into dual bridge tied loads. So it is a bridge tied output for stereo. But if you want 50 watts, you can get that by tying both the positive and negative outputs of the left and right channel together. And then you can configure it over I squared C to say, hey, I want you to do mono output. You can drive, I think, four ohms with that dual bridge tied load. Otherwise, if you're using stereo, eight ohms where it's at. Although it's a little unclear. Maybe you can use six or more four ohms as well. It just won't be as efficient. You're going to get the most efficiency with eight ohms at 20 watts. You'll get 95%. Nothing to watch out for is the total harmonic distortion depending on your power supply, how much power you're trying to push through, and your load resistance. On the left, it's eight ohms. On the right, it's four ohms. You want to keep it under 10% total harmonic distortion where you can actually start to hear the effects of the crackling effects. But up to 18 volts, you can still get 30 watts into four ohms. And if you want 30 watts into eight ohms, you'll need 24 volts. OK, some nice things. I squared C interface means you can set things like the gain over I squared C instead of digitally by changing your max integer value on I2S. Of course, there's lots of other settings available, like what format you want, whether you want that mono mode or stereo or mute. But you will need a driver for the I squared C interface. The one cool thing that this chip does have is dynamic compression. I'm trying one with dynamic range compression. So one issue that people have when they have these very powerful amplifiers is that it's very easy to have clipping when you get to the higher amplifications. If you have an all-in-one system, you can probably tune it so that you'll never generate an audio signal so loud that it would clip on the output. But if people are using the long power supply or they're using different spec speakers and you expect you could have the bing and you want to avoid that because it sounds terrible. And so what you can set up is this dynamic compression that as the signal gets louder and louder and louder, it doesn't amplify as much. So you see there's this cutoff where after DRCT, whatever gain or volume, it starts compressing the amount of gain to be lower and lower so you never hit that clipping rate where you might get square waves into your audio signal. It's also got a compensation filter, which is kind of nice. So the output LC filter does have a slight effect on the audio, especially at high frequencies because you don't have perfect inductor and capacitor. And so if you can measure the effective transfer function that is caused by the LC filter on your audio, you can then precompensate it and so you get a little bit of a gain at the high frequencies to basically have an even equivalent DC gain across all frequencies. Let's see. Oh, there's an error mode, which is kind of nice. So of course, because there's an I2C, you can read from it, not just set settings. And so there's a couple of different error modes. It can flip up that GPIO, you can read it. It'll tell you you have open circuit shorts, you're missing 2S data, you have a DC voltage where it's not supposed to be. So that's kind of nice, good if you're trying to have a better user interface for people, especially again, if people be doing their own installation with this amplifier. And I did notice that there is a kernel module written by the RichTech team that was submitted into mainline in 2021. And so if you want to use this with a single board computer and running Linux, you don't have to do any work because you'll already have a sound driver that's built into the kernel ready to go. Finally, there's the Valboard. You want to get started quickly. Big chunky connections for power and speaker output, and then you can just control it over I2C and give it I2S data, and you're good to go, nice looking Valboard. Also in stock. Bill Belandischke. And the price is nice. It's like $2 to $3 in quantity. $3 for small, $2 for low quantity. That's a pretty good deal for stereo I2S. No M-clock needed. 30 watts or 50 watts mono stereo I2S amplifier. So a nice chip class D. Check it out if you need a big honking stereo system. This is a good I1FK. Good work. Hi, on IMPIA.