Noisy PGA2310

Hi Ferenc,

Welcome to the E2E forums!

I'm sorry to hear about the noise in your system, however without more information it will be hard to diagnose the problem. Layout issues can be a common problem for noise, what is the layout of your system? Have you tried using the EVM?

Regards,

~John

Dear John,

I wouldn't think that layout could go so wrong. I tried many way to remove noise.

I now follow star ground topology, which works perfectly in my other audio stuffs, such as LM3875 amplifier and my headphone amplifier, as well.

For PGA2310 the star ground point is the common pin of the two 10uF capacitors. From this point leads the pin10 and pin15 through seperate wires. Also this point is connected by a short wire to the pin5.

It is just too simple to screwed up. 

As for the input, intresting thing that there is no difference in noise if input is shorted to the ground or not.

I haven't tried using EVM because I thought it is very simple to make PCB for it (only 6 capacitors). But now I would be interested if the EVM could solve noise issue, or the chip is not good enough. But EVM board is too expensive for me.

BR,
Feri

Hi Feri,

Well, the EVM would be able to demonstrate the performance charts on the d/s. Do you have a schematic I could take a look at?

It's interesting to note that there is noise if the input is grounded, do you have any filtering from the output of the PGA2310 into your amplifier?

Regards,

~John

Hello Feri,

We usually recommend following the application schematic provided in the product data sheet, or the schematic from the EVM which can be found on page 6 of the PGA2310 User's Guide. 

I am happy to send you a PGA2310EVM so that you can evaluate the device's performance on an ideal test platform. Please provide your shipping address to my e-mail at ian@ti.com.

Best regards,

Ian Williams
Linear Applications Engineer
Precision Analog - Op Amps

Hello,

here are the pictures of my circuit. Ii is unnecessary to attach the schematic because it is the same as in the datasheet.

Hi Feri,

Unfortunately, a noise problem in a hand-soldered board is tough to solve; remotely even harder. I recommened taking Ian up on getting the EVM and modifying that for use in your system.

Regards,

~John

Hi John,

EVM is on the way, thanks for Ian. We will see the result.

(But it should work on a protoboard as well)

BR
Feri

Hello Again,

I have received EVM board and it has the same noise problem as mine. 

If I use PGA output running through a power amplifier and I stand 3m distant from the speaker, I hear hiss.

With HeadPhone, the noise kills the music if that is quiet.

At the moment I use the simpliest method:

-Short the PGA’s input to the ground

-Connect +/-14V and AGND

-Connect VD+ and DGND

-Connect a headphone/Amplifier to PGA’s output.

Result is harmful noise.

Hi Feri,

I sent you a couple of TPS7A30-49EVMs to ensure that you're getting clean power to the PGA2310. 

At this point I'm unsure where your noise is coming from, although it seems to be from an external source since you're using the EVM and the inputs are grounded. As we discussed via e-mail, you're using a headphone amplifier between the PGA2310 output and headphones, right? The PGA2310 can't drive headphones directly.

Is it possible to take a picture of your test setup and post it here or send it to me via e-mail? 

Best regards,

Ian Williams

Thank You Ian again!

I am looking forward to get it!

Until that, I attached the picture you asked.

BR,

Feri

Hi Feri,

Thanks for the photo. 

I see that you're powering VD+ from your digital board by soldering a wire to jumper J1. However, it doesn't look like you're providing a GND connection from your digital board to the PGA2310EVM. Make sure to connect a common GND between them. Also, since you're in three-supply operation you should remove jumper J2.

If that doesn't eliminate the noise, disconnect the digital board and use the PGA2310EVM software to communicate with the EVM using the serial cable. Make sure to install the correct jumpers for three-supply or two-supply operation as shown in page 2 of the PGA2310EVM User's Guide.

Best regards,

Ian Williams

Hi Ian,

I provide GND from the digital board through the blue wire connected to pin 'DGND', above U2.

J2 is not shorted by jumper, it just hangs on it.

Unfortunately I don't have LPT port on my PC's back, only have header on the motherboard. I am going to get a cable this week.

But shouldn't it work without digital control? I mean if you powering only the analog part of PGA, it should work with no noise.

Datasheet says: 

On power up, all internal flip-flops are reset. The gain byte
value for both the left and right channels are set to 00HEX,
or mute condition. The gain will remain at this setting until
the host controller programs new settings for each channel
via the serial control port.

So, it should work without serial communication, just at mute condition. Am I right?

BR,
Feri

Hi Feri,

I see that now. It's a little bit difficult to see everything on the photo.

You're right that the PGA2310 will work without digital control, but you must still apply digital power, VD+, for the part to start up properly. As you have stated, without digital control it will start up in a mute condition, but to check noise and distortion levels in normal operation you'd have to configure for some non-mute amount of gain.

I'd like for you to check without your external digital signals applied since, other than the headphone amp, that is the only difference between our test setups. Of course, I assume that your headphone amp works normally.

Best regards,

Ian Williams

Dear Ian,

Fortunately, I could get a PC with LPT port on it.
Today I installed EVM software and connected LPT cable to EVM. It works just perfectly.
However, noise is still alive.

My headphone amp works normally if you mean that it has no noise when its input is shorted to the GND.
Do you have any other suggestion?

BR and thanks,

Feri

Hi Feri,

Have you received the LDO boards yet and tried powering the system with clean power?

Can you try disconnecting the headphone amp and probing the PGA2310 output with an oscilloscope?

Best regards,

Ian Williams

Hi Ian,

I haven't received it yet. But it will be so welcomed!

This situation is a nightmare. I every day try and try to solve it finally, but with no success.

I couldn't suspect my headphone amp is the source of the problem, because my power amplifier is noisy as well with the PGA, and a headphone directly connected to PGA's output reveals the noise, (let's ignore now that PGA cannot drive a HP properly) too.

I don't have oscilloscope.

Anyway, what is your experience with setting PGA2310 gain over 0dB? As for mine, the noise is increasing significally, and over +20dB the noise is so strong as if it was raining heavily. (if you need, I have recorded it)

Feri

Feri,

I am also an audiophile, though I don't have experience on PGA2310. However, take a look at the PGA2310 datasheet, you can see the noise @ 0dB is 9.5uVrms. When you set gain to 20dB, the noise will likely increase to 95uVrms, a high enough value for headphone. If you amplify it further using amplifier, the noise will be obvious.

From my experience, to get the best performance, you should set the input signal to high amplitude and the gain of the following amplifier to low value, since the signal quality (SNR, THD+N...)normally is better at high volume (just before clipping occurs), while the amplifier with feedback becomes better at low gain.

Now return to your case, if you use 2Vrms line out signal as souce, and set the PGA2310 gain to 0dB, the SNR @ PGA2310 output can be calculated to be 20*log(10,2Vrms/9.6uV) = 106dB. A very good figure. Even if you use the around 500mVrms MP3 or mobile phone output, the SNR is still above 90dB, still a goo figure.

If you indeed need a x10 gain stage, and can't tolerate the noise of PGA2310, I suggest to build one x10 amplifier using low noise OP and put before PGA2310. 

Best Regards

Frank Pan

Hi Frank,

I am glad to see you helping me.

In my case, the headphone amplifier's gain is set to 2V/V, so there is not much noise amplification.

The PGA2310's gain is limited at 0dB in software because I don't need to amplify the signal, only attenuate it.

Still, I have noise at output.

My only idea is that I must use a voltage divider between PGA and headphone amplifier to decrease not only the music level but also noise level, and, as you suggested, "build one x10 amplifier using low noise OP and put before PGA2310."

I hope this will help me to totally eliminate noise at headphone.

Or, can you take a look at my DAC output circuit? I could increase its gain setting to have enough level to drive PGA2310 properly, and I wouldn't need to build another amplifier before PGA?

http://www.wolfsonmicro.com/documents/uploads/data_sheets/en/WM8740.pdf
Figure 27

I use in this circuit 2x OPA228.

Thank You very much!

Feri

Hi Feri,

I take a look at your DAC, it can output 2Vrms voltage. So a x10 amplifier between DAC and PGA2310 is not needed.

Since your HP amplifier has gain of x2, when the gain of PGA2310 is set to 0dB, the maximum output of your HP amplifier is 4Vrms. It is rather high voltage for headphone, even high impedance ones. 

So you can decrease the gain of the headphone amplifier. If your headphone amplifier is quiet enough, the output noise is proportional to headphone amplifier gain. However, some headphone is not stable when set to low gain, and some can't be set gain to below to 1. You can check your headphone amplifier. 

You are right, you can also put divider resistors between PGA2310 and headphone amplifier. In this case, the divider resistors should be placed near the headphone amplifier side.  You can also increase the DAC output voltage (might to 8.5Vrms if you use +/-15V supply for OP and PGA2310) , then divide it to low value by divider resistors, this will also help decrease noise.

Best Regards

Frank Pan

Hi Frank,

I would like to choose to put divider resistors to their place according to our discussion.

The question is that, how can I change DAC output voltage to about 8Vrms? This circuit is too complicated for me because it is some kind of parallel.

Could you please help me to calculate the right value of resistors?

Thank you!
Best Regards,
Feri

Hello,

I think Frank was right. I ordered a PGA2311 as a sample and replaced my PGA2310 with it, and noise almost has gone!

I think it is because of the 9.5uV output noise of PGA2310.
PGA2311 has 2.5uV noise and it seems low enough to hear almost no noise at all.

After that, at the beginning I told that PGA2310 is too noise. Time has proved it true.

Anyway, Thank You very much for your help!

Best Regards,
Feri