Opamp temperature during use

Hello everybody,

I made a small board which contains a headphone amp, using a dual opamp (OPA2134 in DIL8) as a gain stage, and then four unity gain opamps into 1.6ohm mixing resistors in parallel, per channel.
I used a quad opamp per channel, it's a OPA4134 in an SMT package. The OPA2134 is lukewarm, but the smaller quad opamps are pretty hot, I'd estimate about 55 degrees or so.

From a reliability standpoint, I'd rather have them a bit cooler, or is this higher temperature no problem? Would lower rails (currently +/-15volts) help? And a higher value for the mixing resistors? This amp is meant to drive 32ohms headphones, so I could go upto 8 or so ohms mixing resistors?

Hi.
Did you measure the offset of each opamp before the mixing resistor.

I see no good reason for the degeneration resistors to be that small if the load is only 32 ohms..

If they are getting hot sitting at idle, it could be DC offsets,,,  cooler is always better for reliability.

JR

Thanks for the replies, I did check for offset, but only after the resistors, total offset was 10-15mV. Will check individual offsets tonight. I'll also up the series resistors.
Maybe I'd just have to glue a heatsink to the buggers

Why would you do it that way, when there are a huge number of i.c's designed as power amps?
Any of the National power opamps would be better. And they come in all powers.
Start again with a proper power opamp.

radardoug said:

Why would you do it that way, when there are a huge number of i.c's designed as power amps?
Any of the National power opamps would be better. And they come in all powers.
Start again with a proper power opamp.

Click to expand...

Exactly!

-a

maybe he had too many opamps laying around...

If he was talking about a mass production design I'd share the concern, but for one off DIY, it looks like a learning opportunity...

I have actually seen product designs influenced to use up old parts that were in excess,,,, but that often ends badly with more of the undesirable parts reordered and ending up left over again... 

JR

Some people ( myself included) like to "run whatcha brung"... Nothing wrong with that.

Thanks for the replies guys, it was indeed a matter of using up some parts in the stash. Someone at DIYAudio.com did a headphone amp design with two parallel NJM4556 opamp stages, so initially I was going to try that as well. But since I don't have those opamps (and those are the more muscular variety, think they can do 70-100mA?), I figured I'd use up the SMT quad opamps I had in my stash and wasn't going to use for anything else in the near future.
They were bought together with some other stuff I actually needed at the time, I guess we all know how this goes.

Changed the resistors to 10 ohms, so Zout is 2.5 ohms. More than 12 times smaller than the intended load so that should be ok. 6-8mV of offset, and they get warm but not blistering hot, so that's ok.
Sounds nice, but I am actually more fond of the other opamp I built using LME49600 line drivers, that one sounds really great, and is a really low parts count build.

Glad to hear that cleared it up.. That heat is kind of like the class A bias in a class AB power amp, a little bias current can make a lot of heat when dropped across the entire power supply. 

I have paralleled opamps before in a commercial design, when I had only one circuit in an entire design that needed slightly more drive capability and I didn't want to add a discrete buffer or bring in a different higher current part.

Pushing GP opamps to their current extremes could easily expose the weak sisters in the group.

JR

Let's hope 160mA is enough  ;D

A common design technique for headphone outputs is to add a series resistor, so even 3,2ohm cans will not draw amps of current. Some might prefer driving high impedance cans with less than that 10 or 20 ohms of series build out R.

JR

Somewhere I have calcs which suggest that 27 ohms in series with a 7V source will put near-max power into a range of typical headphones of different impedances.

Some higher-Z cans have large Z variation but most "32" phones have very flat Z curves.

So I'd go 50 or 100 ohms in series with each of the quad-team amps. And for worst-case 10mV offset, that puts offset-induced cross-current down below 1mA.

PRR said:

Somewhere I have calcs which suggest that 27 ohms in series with a 7V source will put near-max power into a range of typical headphones of different impedances.

Click to expand...

I think it's this one.

The old server moved, picture links lost, and I can't edit Cmoy's pages. Here are the images again:

"....parsing the data in Rane's paper, there are trends."
http://headfonz.rutgers.edu/head-power2.gif

"Here is my chart with "typical" performance of op-amp based designs sketched in."
http://headfonz.rutgers.edu/head-power3.gif

In this instance, small binder clips epoxied to the offenders might make them, and you happy.  I do this with the bridge rectifiers for the heater circuit in Lydcraft equipment.  It eliminates burned PCB's, blown components, and downtime in the future.  They are designed right on the edge, and usually fall off after 8-10 years.

Into the future your chip choice might not last as long as a higher power rated part, but they got the job done.
Mike