Are the input and output electrolytic capacitors connected properly, with their positive terminals to the amplifier,
and not to the source or load?
Mark,
All I can suggest is that you think about what you said you're driving with, and what the capability of LM380N chip is. According to LM380 | Mid/High-Power Class D Amplifiers | Audio | Description & parametrics it is specified to be able to handle ± 10 to 22 volt supplies, and to source 1.5 amps to the load.
You've got the dual chipped board hooked to ±15 V at 0.5 amps. This - to me - means “clipping is going to happen” at elevated levels.
One of the reasons your Sennheiser headphones headphones gave minimum output is because (if I remember right) they're also pretty high impedance. 32 Ω I remember - for the older ones. Many other headphones are either 16 Ω or 8 Ω.
In any case, the clipping current will be defined both by what your power supply can deliver (enhanced by the possible on-printed-circuit-board power supply reservoir capacitors), but for continuous SINE waves, it would be
E = IR
I = E/R
I = 15 ÷ 32 = 0.5 amps
I = 15 ÷ 16 = 0.9 amps
I = 15 ÷ 8 = 1.8 amps
See what's going on there? Current limits of the LM380 _will clip_ above their internal protection limits. You'll hear that. Likewise, the higher impedance Sennheisers will simply be drawing less power. You'll hear that too. Quieter operation.
Lastly, for music program sources (i.e. pop music and such), one can figure VRMS = ¼ VPEAK (or even lower). Your VPEAK simply can not exceed 15 volts. Period. That's what power-supply "rail clipping" is all about. It sounds really nasty.
So if your music program source is 100 millivolts (common for PCs and such), and if the LM380's fixed gain is 50 (unless other external measures are implemented to mute it), then the 100 mV × 50 = 5.0 V VRMS. But the peaks might be 4× higher or attempt to go to 20 volts.
Guess what that'd do.
Clippity clippity clip.
TWO solutions, both of which almost aren't worth doing. Get a bigger "near max spec" power supply of ± 22 volts. Get a much larger reservoir capacitor set for the power supply (10,000+ μF) so that the average and peak loads can be met. And don't play the thing quite at clipping. Its bad for your ears, anyway.
As in really bad, over long periods of time.
Just saying,
GoatGuy
All I can suggest is that you think about what you said you're driving with, and what the capability of LM380N chip is. According to LM380 | Mid/High-Power Class D Amplifiers | Audio | Description & parametrics it is specified to be able to handle ± 10 to 22 volt supplies, and to source 1.5 amps to the load.
You've got the dual chipped board hooked to ±15 V at 0.5 amps. This - to me - means “clipping is going to happen” at elevated levels.
One of the reasons your Sennheiser headphones headphones gave minimum output is because (if I remember right) they're also pretty high impedance. 32 Ω I remember - for the older ones. Many other headphones are either 16 Ω or 8 Ω.
In any case, the clipping current will be defined both by what your power supply can deliver (enhanced by the possible on-printed-circuit-board power supply reservoir capacitors), but for continuous SINE waves, it would be
E = IR
I = E/R
I = 15 ÷ 32 = 0.5 amps
I = 15 ÷ 16 = 0.9 amps
I = 15 ÷ 8 = 1.8 amps
See what's going on there? Current limits of the LM380 _will clip_ above their internal protection limits. You'll hear that. Likewise, the higher impedance Sennheisers will simply be drawing less power. You'll hear that too. Quieter operation.
Lastly, for music program sources (i.e. pop music and such), one can figure VRMS = ¼ VPEAK (or even lower). Your VPEAK simply can not exceed 15 volts. Period. That's what power-supply "rail clipping" is all about. It sounds really nasty.
So if your music program source is 100 millivolts (common for PCs and such), and if the LM380's fixed gain is 50 (unless other external measures are implemented to mute it), then the 100 mV × 50 = 5.0 V VRMS. But the peaks might be 4× higher or attempt to go to 20 volts.
Guess what that'd do.
Clippity clippity clip.
TWO solutions, both of which almost aren't worth doing. Get a bigger "near max spec" power supply of ± 22 volts. Get a much larger reservoir capacitor set for the power supply (10,000+ μF) so that the average and peak loads can be met. And don't play the thing quite at clipping. Its bad for your ears, anyway.
As in really bad, over long periods of time.
Just saying,
GoatGuy
Goatguy,
I think you are correct, it has been some while since I have used headphones and they were always ear bleeding levels when I was DJ'ing. This setup works but I feel it runs out of steam too early. What else are people using these days? The only headphone amp that blew me away was Musical Fidelity X Cans V2.
With regards to components they are on hand.
I think you are correct, it has been some while since I have used headphones and they were always ear bleeding levels when I was DJ'ing. This setup works but I feel it runs out of steam too early. What else are people using these days? The only headphone amp that blew me away was Musical Fidelity X Cans V2.
With regards to components they are on hand.
I don't see the power supply as a problem. After all, <7.5 Vp into 40 ohms is <200 mA, peak.
I hadn't really looked at the LM380 before, but better late than never I guess. It's pretty much a beefed-up version of the venerable LM386, using the same kind of level-shifting 0-V-included input stage (input coupling cap not required but nice to have for pots) and output DC offset setting. Complete with the same kind of high noise levels and miserable distortion, I assume. At the end of the day, it's a pretty lousy (if interesting) circuit. Current gain just about gets the job done, there isn't an awful lot of loop gain (or PSRR), output stage biasing could do with a bypass cap. It could get pretty close to the rails but only on higher-impedance loads. The 1.3 A peak output current spec sounds impressive but I bet it can only sink that much and source a fair bit less. I mean, how much current could the VAS ever run at if the entire ICs draws but 7 mA, and how much beta would the single npn follower still have? Like 4 mA, and 200?
40 ohms, 100 dB/mW - what sort of Sennheisers are these?
You'll find just about a gazillion different approaches in the Headphone Systems section of this forum, depending on which spot in the complexity - performance - power consumption triangle strikes you fancy. NE5532 with BD139/140 Class A or AB buffer? Same with IC or beefy opamp buffer? LM1876 with some loop gain drained off (needs higher and preferably bipolar supply though)? Something entirely discrete? PCB done entirely from scratch or pre-made? What sort of budget do you have, and what kind of parts?
I hadn't really looked at the LM380 before, but better late than never I guess. It's pretty much a beefed-up version of the venerable LM386, using the same kind of level-shifting 0-V-included input stage (input coupling cap not required but nice to have for pots) and output DC offset setting. Complete with the same kind of high noise levels and miserable distortion, I assume. At the end of the day, it's a pretty lousy (if interesting) circuit. Current gain just about gets the job done, there isn't an awful lot of loop gain (or PSRR), output stage biasing could do with a bypass cap. It could get pretty close to the rails but only on higher-impedance loads. The 1.3 A peak output current spec sounds impressive but I bet it can only sink that much and source a fair bit less. I mean, how much current could the VAS ever run at if the entire ICs draws but 7 mA, and how much beta would the single npn follower still have? Like 4 mA, and 200?
40 ohms, 100 dB/mW - what sort of Sennheisers are these?
You'll find just about a gazillion different approaches in the Headphone Systems section of this forum, depending on which spot in the complexity - performance - power consumption triangle strikes you fancy. NE5532 with BD139/140 Class A or AB buffer? Same with IC or beefy opamp buffer? LM1876 with some loop gain drained off (needs higher and preferably bipolar supply though)? Something entirely discrete? PCB done entirely from scratch or pre-made? What sort of budget do you have, and what kind of parts?
Last edited:
The LM380 is a very old IC and couldn't take advantage of some of the later design techniques or processing in later ICs. National/TI has or had an applications data sheet dated from 1972 for it. Even with good layout, under some conditions, I've had oscillation from the bottom side of the waveform (often called "bottom side fuzzies"). This is most likely due to an improperly compensated CFP in the quasi comp output stage.
The upper transistor uses no bootstrap current source and output swing is limited. Unless the output quiescent voltage was set a bit less than 1/2 the supply voltage, clipping can be asymmetrical. Other chips used a bootstrap cap or like with the TDA2003, a better designed circuit for good output swing into low impedance loads.
The upper transistor uses no bootstrap current source and output swing is limited. Unless the output quiescent voltage was set a bit less than 1/2 the supply voltage, clipping can be asymmetrical. Other chips used a bootstrap cap or like with the TDA2003, a better designed circuit for good output swing into low impedance loads.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.