[Mannevond]

Hi

I got my hands on a few LM49720HA opamps (TO 99 metal capsule), and got this crazy idea that I should test it in my CMoy. First ghetto style attempt was to just cut the legs to length and pop it in the IC socket. Worked out well, sounded ok but I broke one of the legs when swapping between that and the others I normally use. So I decided to mount a new one on a small piece of proto board and solder the legs to make it more sturdy and easy to fit.

Problem now is that it gets VERY warm when playing loud, and still quite warm on moderate volumes. This did not happen with the one I broke. Legs were a bit shorter than on the protoboard one. After a while there is some noise as well, as if it was picking up RF or something, but I think it is something else. Only happens when I have played for a while and the heat has built up. Turning it off for a few seconds gets rid of it until the opamp gets warm again.

Is this oscillation? I have decoupling caps connected directly to the DIL8 socket, but they are a bit further away from the IC than they are when using standard DIL8 ICs, since I have mounted it on my ghetto adapter. Would it help to move the decoupling caps closer to the IC? There is space on the protoboard...

My CMoy has input caps, 2*470uF/25V power caps, 18V power supply, rest is more or less standard. Gain is set to 11, but considering lowering since it is way too loud long before it starts distorting around one o'clock on the volume control.

I really want this one to work. The sound is very nice until the opamp gets warm and starts the noise show...

 

[fzman]

it's probably oscillating - it should not run that warm/hot. you can stick them right into machined dip-8 sockets, and they will stay put - not sure about portable use, though.

they probably have too much dc offset for a cmoy design- i would check before i use it with my good cans.

hope that helps.

 

[Mannevond]

DC offset is around 10mV pr. channel, goes up to 11 mV when I connect a source with no signal. A bit high, but should be acceptable according to what I have read (under 20 mV being safe).

Would bandwidth limiting at around 30 kHz help?
How about moving the bypass capacitors onto the breadboard "adapter"?

None of my other opamps behave like this; JRC4560D, OPA2134PA, AD823, AD8620. I believe the 8620 is the most difficult of the others, but as I said, no problem with any of them.

 

[fzman]

also, the metal case is connected to V-, iirc, so make sure it's not touching things it should not touch!!

i use those in several sound cards, and have not had problems, or had them get hot. if you are not careful, though, you can get accidental contact from a leg to the case, especially when you trim the leads short to get them into the sockets

 

[Mannevond]

Nope, measured it with the multimeter. 7.32 MOhms from case to V- leg, otherwise nothing connected to the case. Measured between neighbouring soldering points as well, to make sure there are no shorts. Doesn't seem to be...

 

[FallenAngel]

LM47920 isn't meant for driving headphones, it's rated for 600R load, most headphones aren't.

 

[Mannevond]

FallenAngel: Ah, saw that in the datasheet now. That explains the heat when driving 32 Ohm cans. I guess the opamp sees a short circuit...

Would this opamp be a good candidate for a preamp output stage?

 

[FallenAngel]

For line driving duty the LM4562 (or LM47920 - same opamp, just a different name) is pretty damn good.

 

[FallenAngel]

LME49723 is not the same as the LME49720 (or LME49710 for single), it's a different opamp with different specs - and higher noise floor.

As for the LME49710HA vs LME49710MA or single vs dual packaging, I have yet to hear a difference, specs are identical, build process is identical... might sound identical?

Haven't heard TLE2062, might try some day - specs look pretty decent for driving headphones - 80mA output, 7-36V operating range, but it swings 3V away from rails.

Off 8.4V (standard rechargeable 9V battery), you'll have +/- 1.2V swing... that's pretty low.

 

[Mannevond]

I use 2 9V batteries for 18V power (16.8 with rechargeables). That'd still give me plenty more swing than needed until the batteries die...

How about putting a buffer in front of the 49720? Would it have to be inside the feedback loop or not? What are the pros and cons? What's a good buffer for such a thing? I can easily steal power for the buffer, but getting it inside the feedback loops would need some more work in my cmoy. Probably better to build one from scratch if it's necessary...

 

[fzman]

there's a design on the national web site, which uses the 49720 along with the 49600. i think it's referenced via the ap note for the 49600 buffer.

 

[FallenAngel]

You don't want a buffer "before" LME49720, you want one "after".

 

[Mannevond]

fzman: Thanks. I just looked at the data sheet and it seems mighty impressive on paper.

Implementing it seems like fairly easy stuff. I'm gonna order a few and start building something. It'll be interesting to see if I can make this work for portable use. Current draw can be quite big, but if the resulting sound is pleasing, I have to learn to live with it

 

[Mannevond]

FallenAngel: I did mean "after". Bad wording on my part