Do I need an amp for my project using a PGA2311

My project's goal is to have an audio input jack (from a phone, for example), and then an output jack for headphones.

I am hoping to use a PGA2311 to adjust the volume of the audio.

My question is: do I need an amp after the PGA2311, or can I just directly plug in the headphone left and right channels? The headphones normally work with music playing from the phone (which is input for PGA2311).

If I would need an amp, can anyone suggest the easiest solution?

Thanks!

Yes, you need an amp. The [u]datasheet[/u] says:

The PGA2311 also provides the capability to drive 660-Ω loads directly without buffering.

In case you don't know this, that's the minimum resistance/impedance, and since headphones are typically 32 - 64 Ohms, they won't work. (There are some 600 Ohm headphones, but they typically require higher drive voltage.)

can anyone suggest the easiest solution?

The easiest solution is to buy a
[u]headphone amplifier[/u]
.

If you want to build something, the [u]LM386[/u] is pretty easy to use (although you may need to knock-down the signal or reduce the gain because I think the default design has too much gain, and you may not want any gain for what you're doing). And, you'll need two for stereo.

Or, TI has LOTs of appropriate power amp/headphone amp chips. They might even have a headphone amp with a built-in digital volume control. I know they make digitally-controlled amplifiers (DCAs) but I don't know if they are just preamps or if they can drive headphones.

Thanks for the detailed reply.

This makes me wonder.. Am I correct in attaching the phone's aux left and right channels directly to the PGA2311 inputs? I assume the audio is already amplified in some way, because headphones can be plugged in, but can that be then plugged into a preamp?

Am I correct in attaching the phone's aux left and right channels directly to the PGA2311 inputs?

I would say yes.
Their is a wide gain you can program into that chip, if you set it too high then all that will happen is the signal will clip and it will sound distorted on the peaks. This sounds like clicking.

The lm386 has 3 orders of magnitude more distortion than the PGA2311, you might want to find
a simple high current buffer amp to boost the output current handling with low distortion
as the PGA2311 already has significant voltage gain for a line-level input. Unless of course
0.2% THD is fine for you.

I've used the PGA2311 for a couple of projects, and use an OPA2134 as the buffer amp.

Works very well.

Ian.

Looked at the specs for the OPA2134 - 35mA drive isn't very much.

I was thinking more like the AD8656 - 220mA of drive and truly rail-to-rail.

Incidentally talking of opamps I had the misfortune to play with an LM358 I had knocking around
at the weekend and found just how poor an op-amp it truly is at 5V - starts to flake out above
5kHz with large signal drive, completely utterly unusable at 50kHz... I had assumed it could at least handle
an audio signal (!)

RedyTedy:
Thanks for the detailed reply.

This makes me wonder.. Am I correct in attaching the phone's aux left and right channels directly to the PGA2311 inputs?

The spec sheet says that distorion rises if the sources output impedance is too high. I would probably put an opamp buffer before it if you can. I have a pga2310 on breadboard at the moment. It has two inputs via relays, one to an Arcam rDac and one to a valve buffered device, it does not play well when connected direct to the later. Also you should provide RF filtering as close to the input connector as you can.

MarkT:
found just how poor an op-amp it truly is at 5V - starts to flake out above
5kHz with large signal drive, completely utterly unusable at 50kHz... I had assumed it could at least handle
an audio signal (!)

5v is very low voltage for most opamps. It may not be optimised for that case.

You may already know but if you are building on bread board you need get the feedback and input shunt resistors within a few mm of the package. Also a 0.1u decoupling cap is needed applied right on top of the package.

davetcc:
The spec sheet says that distorion rises if the sources output impedance is too high. I would probably put an opamp buffer before it if you can. I have a pga2310 on breadboard at the moment. It has two inputs via relays, one to an Arcam rDac and one to a valve buffered device, it does not play well when connected direct to the later. Also you should provide RF filtering as close to the input connector as you can.

Audio line outputs are generally lowish impedance. They are traditionally speced to only drive 47k or higher loads,
but in practice most will drive lower impedance loads fine (not as low as headphones though!). Put another
way these days you can assume an audio signal is produced by something like an opamp and shouldn't need buffering.

Valve circuitry is another matter! I'd add protection circuitry for a signal from a valve amp as large
voltage spikes are a risk (at switch-on and switch-off ?)

It's not so much that the impedance is low on the input of the 2310. In fact the input is relatively high as it's essentially an opamp with feedback.

It's more that if the impedance of the source is not low enough in terms of AC circuit theory - and that includes all cables and RF filtering, as I understand it, it creates an uneven frequency response. Most circuits I've seen with a pga2310 usually have a unity gain opamp in front of it. You could probably even use a 5532 dual as it's being used at unity gain voltage follower.

davetcc:
It's not so much that the impedance is low on the input of the 2310. In fact the input is relatively high as it's essentially an opamp with feedback.

It's more that if the impedance of the source is not low enough in terms of AC circuit theory - and that includes all cables and RF filtering, as I understand it, it creates an uneven frequency response. Most circuits I've seen with a pga2310 usually have a unity gain opamp in front of it. You could probably even use a 5532 dual as it's being used at unity gain voltage follower.

True, neglecting those effects might well degrade the excellent THD performance somewhat too.

BTW 5532's are not low voltage rail-to-rail opamps, the one I mentioned is and has top performance too.

And NE5532's were superseded by NE5534A's decades ago I thought?

The 5534 is the single version of the 5532 dual. It has very slightly better characteristics but is not suitable directly as a voltage follower because it is not stable at unity without compensation. Yep I forgot we are talking about 5v rails. I'm used to always having a second 12-0-12 supply.

While we're mentioning split rail supplies. Can anyone suggest the best, and by that I mean lowest noise, method to get 12-0-12 in an automotive environment?

I'm currently using the PGA2311 with a Murata DC-DC converter to generate the +ve & -ve rails. This converter runs from a 5v linear regulator. I'd like to use the PGA2310 for the extra headroom it offers, but getting a relatively low noise, split rail, DC-DC converter that's happy with the automotive voltage range is proving tricky. I'd rather not use any sort of switching supply on an audio circuit, but I don't know any other way to get the -ve rail from a DC source.

Ian.

Personally ive not attempted this.

Have you looked though linear technologies site? I get their monthly newsletter and it's normally full of automotive voltage controllers. There may well be something there.

They are nearly all smd chips so depends if you are happy with that.

davetcc:
Have you looked though linear technologies site?

I haven't, but I shall. Thanks for the pointer.

They are nearly all smd chips so depends if you are happy with that.

No problem with SMD at all.

Thanks,

Ian.