Page 1 of 1

Voltage output levels of VS1003

Posted: Wed 2019-05-08 3:03
by Berry
This maybe a basic/silly question, so I apologize in advance if that’s the case. I’m working on an embedded project where I’ll be using the VS1003 MP3 decoder along with a class D amplifier to play back audio using a mono on PCB speaker in differential mode (SM DIFF). I want to select an amplifier with gain that will not result in clipping. My thought process is as follows:

1. The VS1003 datasheet shows the following for output voltage on the “L” and “R” channels: Full Scale Output Voltage (Peak-to-peak): min: 1.3 , max: 1.7 Vpp

a. Does this mean that regardless of the MP3 clip recording level and regardless of the volume selected, the output amplitude will always be 1.3-1.7 Vpp?
b. Does this include/superimposed on the DC offset from CBUFF of 1.24V? ie.
i. the output signal range (AC+DC) really is 0.39V to 2.09V for the loudest signal ( min=1.24 – 1.7/2=0.39, max=1.24+1.7/2=2.09v)
ii. the output signal range (AC+DC) really is 0.59V to 1.89 V for the least loudest signal ( min=1.24 – 1.3/2=0.59, max=1.24+1.3/2=1.89v)

2. Assuming my class D amplifier is powered by IOVDD and let’s say it’s 2.8V to match the table in the datasheet under which the Vpp is specified (in reality my IOVDD is 3.3V). Let’s assume the class D amplifier chosen output voltage can go as high as the power supply.

a. To avoid clipping at the amplifier output, the amplifier gain must be such that ampflier output never exceeds 2.8V. This means the ampflier gain must be unity gain? Even a gain of 2 will result in clipping (1.7 *2 = 3.4, which is larger than IOVDD at 2.8V)

Can someone help clarify and comment on my thought process for each of the above points?


Re: Voltage output levels of VS1003

Posted: Wed 2019-05-08 13:09
by pasi
1a. 1.3Vpp to 1.7Vpp is the test limits for the analog swing with maximum (generated) digital signal. The actual level depends on the actual RCAP voltage (varies slightly depending on the individual chip and ambient temperature), and the load at the output. The actual level of a decoded signal also depends on the volume setting.

1b. The swing is from 0V to 2x RCAP (and below AVDD).

2. RCAP is nominally 1.24V, with that the max voltage out is 2.48V . If amp output needs to be limited to 2.8V, you're close to unity gain.

Re: Voltage output levels of VS1003

Posted: Wed 2019-05-08 21:30
by Berry
Thanks Pasi, just to confirm my understanding:

1a. If I understand correctly then, assuming a nominal RCAP of 1.24, then the signal with full scale output voltage of 1.7 Vpp is as shown in this diagram. Is this correct? Maximum voltage is 2.09V and minimum is 0.4 V, for this particular test scenario.

2. So this means the absolute maximum voltage output for “L” or “R” is from 0V to 2RCAP, taken into account any volume settings, load, & temperature? Assuming RCAP is 1.24, this means amplifier gain must be chosen such that the output voltage at “L” or “R” never exceeds 2RCAP to avoid clipping ? (ie. gain can't be allowed to cause output voltage to go power supply of 2.8V as mentioned earlier)

3. What is tolerance on RCAP? 1.24V is nominal, but how low or high will get?

Re: Voltage output levels of VS1003

Posted: Thu 2019-05-09 10:17
by pasi
1a. The output swing is around the RCAP voltage, yes.

2. I don't see what vs1003 maximum has to do with your amplifier output maximum. I assumed 2.8V was the limit of the amp due to its power supply. Once the signal is out of vs1003, 2x RCAP is no longer a limit.

3. The actual RCAP voltage we tend to see is 1.2V. The test limits are about +-8%, but due to the expected Gaussian distribution, most are around 1.2V.

Re: Voltage output levels of VS1003

Posted: Fri 2019-05-10 2:16
by Berry
Thanks, I got everything I need now. Re-reading my question #2, I didn't properly word and I agree it's confusing. Fyi, I was confused when going through this reference design and hence my questions in this thread: ... icsV16.pdf

It's not clear what the power supply for the amplifier is in that case, but I assume 5V and in which case, it will definitely clip since the gain is 9.7, which means 1.7Vpp*9.7 = ~16Vpp, or 8Vpp per rail. It now dawned on me that for that case, there must have been attenuation by playing with the volume settings of the VS1003, such as to avoid clipping, in which case, things makes sense.