I recently built a headphone amplifier based on a USB to I2S chip, a PCM1792A DAC, a OPA4134 and a TPA6120A2. This was a fun learning experience, and very satisfying as this headphone amplifier sounds just fantastic.
The PCM1792A has differential outputs for each channel, and these outputs are current. There is an I/V stage (The OPA4134), which also has differential outputs, and this is finally fed to the TPA6120A2. No capacitors in the signal path, other than the LPF required by the DAC. You can find this circuit in the TPA6120A2 datasheet, but for your convenience I'll attach it here:
Now on to the questions:
1) Could I replace the TPA6120A2 in this circuit with LM3886 in the same configuration? Replacing Rf/Rg resistors to set gain, of course. Will this give me a fantastically high SNR amplifier with excellent noise rejection at the input?
2) Instead of one LM3886, can I use two of them in "bridge"? In this case I already have the two signals 180 degrees out of phase. Could I use two chipamps in the same configuration (both inverting or non-inverting), and connect each to a speaker's leg? Will this give me 100W @ 8 ohms? Will this also have high SNR and high noise rejection?
The PCM1792A has differential outputs for each channel, and these outputs are current. There is an I/V stage (The OPA4134), which also has differential outputs, and this is finally fed to the TPA6120A2. No capacitors in the signal path, other than the LPF required by the DAC. You can find this circuit in the TPA6120A2 datasheet, but for your convenience I'll attach it here:
Now on to the questions:
1) Could I replace the TPA6120A2 in this circuit with LM3886 in the same configuration? Replacing Rf/Rg resistors to set gain, of course. Will this give me a fantastically high SNR amplifier with excellent noise rejection at the input?
2) Instead of one LM3886, can I use two of them in "bridge"? In this case I already have the two signals 180 degrees out of phase. Could I use two chipamps in the same configuration (both inverting or non-inverting), and connect each to a speaker's leg? Will this give me 100W @ 8 ohms? Will this also have high SNR and high noise rejection?
I would set up the LM3886 as a differential amp and use it to provide power to your speaker.
You could use two non-inverting LM3886 amps (use 0.1% resistors for setting the gain) if you really want a bridge configuration.
You'll get the best performance with the single LM3886 configured as a diff amp, though. That's my prediction anyway.
~Tom
You could use two non-inverting LM3886 amps (use 0.1% resistors for setting the gain) if you really want a bridge configuration.
You'll get the best performance with the single LM3886 configured as a diff amp, though. That's my prediction anyway.
~Tom
I'm simulating with LTSpice, using just an opamp as the output (i changed the output current, though). A single opamp works like the circuit in the photo, but two of them don't work right:
what I did was use two differential input opamps, and swap the input polarity in one. When measuring the outputs of the output stage, the outputs are effectively 180 degrees out of phase. But when i connect them through a 8 ohm resistor (to simulate a speaker), they clip when supply is +/-35V. They do work when the supply is +/-70V. This is impossible for the LM3886, of course.
The problem is, of course, that I don't understand how I should wire this
the LM3886 isn't very good as a amp compared to the TPA - but its aiming for a lot more power
another consideration is the LM3886 has a minimum gain requirement too
in all the LM3886 or other power chip amp makes a poor diff amp compared to properly chosen, applied op amps of the TPA6120 generation
even the TPA isn't the best diff amp due to the big difference in the +,-input as a CFA
you can take the large Vswing of a op amp diff output from the DAC and use it to drive power chip amp with a little distortion advantage - have to use "noise gain compensation" to meet the LM3886 min gain while using a lower gain for the audio signal
most go more modular - build power amps in separate boxes so you can mix and match source
another consideration is the LM3886 has a minimum gain requirement too
in all the LM3886 or other power chip amp makes a poor diff amp compared to properly chosen, applied op amps of the TPA6120 generation
even the TPA isn't the best diff amp due to the big difference in the +,-input as a CFA
you can take the large Vswing of a op amp diff output from the DAC and use it to drive power chip amp with a little distortion advantage - have to use "noise gain compensation" to meet the LM3886 min gain while using a lower gain for the audio signal
most go more modular - build power amps in separate boxes so you can mix and match source
I'm simulating with LTSpice, using just an opamp as the output (i changed the output current, though). A single opamp works like the circuit in the photo, but two of them don't work right:
what I did was use two differential input opamps, and swap the input polarity in one. When measuring the outputs of the output stage, the outputs are effectively 180 degrees out of phase. But when i connect them through a 8 ohm resistor (to simulate a speaker), they clip when supply is +/-35V. They do work when the supply is +/-70V. This is impossible for the LM3886, of course.
The problem is, of course, that I don't understand how I should wire this
Schematic?
~Tom
Schematic here:
I think I might have found the problem to my previous question. I was giving my amplifier an input of 1Vpp, but the output clipped. I was doing it wrong: the differential amplifier ampifies "twice as much". So, for a gain of 20, I was effectively getting a gain of 40 (since the formula is Rf/Rg * (V2-V1) - in my case V2-V1 was 2V when the input was 1V, because of the differential amplifier). I reduced the LM3886's gain to 10 and I now get the expected output through the resistor.
EDIT: Disregard the part above. It's correct but the output voltage isn't enough.
This is what I'm getting at the output:
The output is clipped to 20Vpp, in each amplifier, while both have 70V supply (+/-35V). Looking at inverting input of the amplifiers there's this:
Which is, I suppose, the amplifiers fighting against each other.
Why does this happen?
I think I might have found the problem to my previous question. I was giving my amplifier an input of 1Vpp, but the output clipped. I was doing it wrong: the differential amplifier ampifies "twice as much". So, for a gain of 20, I was effectively getting a gain of 40 (since the formula is Rf/Rg * (V2-V1) - in my case V2-V1 was 2V when the input was 1V, because of the differential amplifier). I reduced the LM3886's gain to 10 and I now get the expected output through the resistor.
EDIT: Disregard the part above. It's correct but the output voltage isn't enough.
This is what I'm getting at the output:
The output is clipped to 20Vpp, in each amplifier, while both have 70V supply (+/-35V). Looking at inverting input of the amplifiers there's this:
Which is, I suppose, the amplifiers fighting against each other.
Why does this happen?
Last edited:
Seeing 6ohms rating alerts me to look for the effective impedance of each section.
Is the bass section a 6ohm, is the mid section a 6ohm, is the treble section a 6ohm?
The section that takes most of the passband has the most influence on the effective load seen by the amplifier.
eg. a 2way with a 4ohm bass mid and a 10ohm treble could draw total current with a white noise signal that indicates an average 6ohms impedance.
But the Bass mid is the predominant load seen by the amplifier for almost all the high level, longer duration signals.
Had the impedance test been done with pink noise the avaerage impedance may have come out a bit lower than measured by the white noise test.
The amplifier behaves (heats up) as if it was driving a 4ohms speaker and not a 6ohms speaker.
A bridged amplifier driving a 6ohms speaker is seen by both halves of the amplifier as a 3ohms load.
But if it's a 4ohm bass mid then the amplifiers see 2ohms.
Having been alerted, you must find out what's inside.
Is the bass section a 6ohm, is the mid section a 6ohm, is the treble section a 6ohm?
The section that takes most of the passband has the most influence on the effective load seen by the amplifier.
eg. a 2way with a 4ohm bass mid and a 10ohm treble could draw total current with a white noise signal that indicates an average 6ohms impedance.
But the Bass mid is the predominant load seen by the amplifier for almost all the high level, longer duration signals.
Had the impedance test been done with pink noise the avaerage impedance may have come out a bit lower than measured by the white noise test.
The amplifier behaves (heats up) as if it was driving a 4ohms speaker and not a 6ohms speaker.
A bridged amplifier driving a 6ohms speaker is seen by both halves of the amplifier as a 3ohms load.
But if it's a 4ohm bass mid then the amplifiers see 2ohms.
Having been alerted, you must find out what's inside.
Last edited:
Then I'd have to take it apart and look at what's inside. The manual doesn't specify. This speaker is 3-speakers but 2-way only (two subwoofers and a tweeter).
two bass mid and one treble immediately tells me the manufacturer is lying when specifying 6ohms.
It is almost certainly a pair of 8ohms drivers that become paralleled at some frequency to make a bass shelf that cancels the baffle step loss.
It probably means the amps see 2ohms. That is a bridged amplifier killer !
It is almost certainly a pair of 8ohms drivers that become paralleled at some frequency to make a bass shelf that cancels the baffle step loss.
It probably means the amps see 2ohms. That is a bridged amplifier killer !
which is what you should have done to develop the brief before designing the amplifier.
I took the woofers out and this is what I found:
Two of these in parallel, a crossover (2 caps and 2 coils), and the tweeter.
Should I really proceed to measure the impedance of the speaker?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.