I took a look at the data sheet. They've done all the distortion plots with a 22 kHz BW so it peaks at ~15 kHz and then drops off. I suspect if you did it with a more normal 80 kHz BW, the distortion would be very high.
For a cheap portable HP amp, this might do, but I dont think its real hi-fi.
I bought a DAC recently which has a very good sound on its line output. But, it uses one of the low voltage TI headphone drivers and I was shocked at the difference between it and my iPod, and then again from the iPod to a proper HP amp.
The thing with HP amps, you can hear shortfalls in the sound immediately.
I've used the LME49600 + LME49710 - thats a wicked combo, but granted, its not portable.
For a cheap portable HP amp, this might do, but I dont think its real hi-fi.
I bought a DAC recently which has a very good sound on its line output. But, it uses one of the low voltage TI headphone drivers and I was shocked at the difference between it and my iPod, and then again from the iPod to a proper HP amp.
The thing with HP amps, you can hear shortfalls in the sound immediately.
I've used the LME49600 + LME49710 - thats a wicked combo, but granted, its not portable.
TPA6120 has been the actual topic of recent activity great, with audio specs in its datasheet - and it is a DSL driver chip so you can look at its alias part number THS6012 datasheet for distortion plots to MHz (but only starting at 100 kHz)
the 2 data sets show a ridiculously linear performance of the chip by itself
the LME buffer's distortion plots are never shown outside of another high gain feedback loop - I would bet its performance is not as good as the TPA6120
the 2 data sets show a ridiculously linear performance of the chip by itself
the LME buffer's distortion plots are never shown outside of another high gain feedback loop - I would bet its performance is not as good as the TPA6120
Last edited:
to calaculate Zout of the amp you need the outer feedback loop "excess loop gain", which varies with frequency
for 1 kHz and that config you have >3000x, so you have less than 10 milliOhm output Z from the amp
by 20 kHz the excess gain drops ~20x so the Zout @20kHz ~100 milliOhm
for 1 kHz and that config you have >3000x, so you have less than 10 milliOhm output Z from the amp
by 20 kHz the excess gain drops ~20x so the Zout @20kHz ~100 milliOhm
Well, I dunno 'bout that. I've had a pretty hard time putting my finger on the sonic differences between a barefoot Clip+ and a FiiO E11 when using my trusty HD590s - they do seem to exist (I prefer the E11) but in a direct ~level-matched comparison they always come out as pretty darn subtle. ("Unfortunately" I mostly seem to be using cans that are notoriously easy to drive for one reason or another - these at about 100 ohms nom and 200 ohms max and ~110 dB/V, HD580s, Soundmagic E10s. And I'm a rather quiet listener to boot, estimated 55-60 dB SPL typ, somewhat higher on classical. Guess I should drag out those DT231s.) The E11 is more noisy when using more sensitive loads though. I can also hear the difference between 47R, 22R and ~0R of Zout with the '590s.
That brings us to the interesting part - the question of Why? What kind of cans did you use? What kind of levels do you tend to listen at? And which model TI headphone driver are we talking about (they've got like a gazillion these days, plus the NatSemi lineup)? (They say curiosity killed the cat, I know. Makes me awfully glad I'm not a cat.)
Usually non-subtle differences have quite non-subtle causes as well. Unmatched listening levels, differing output impedance (via interaction with headphone impedance response), audible hiss, non-negligible ground return resistance that degrades xtalk, even channel imbalance (threshold of audibility is ~0.3 dB), and finally gross crossover distortion. Most of these should show up in basic loaded measurements using a good-quality splitter cable. Most of them are also related to the properties of the load in some way.
Last edited:
no, you really need some EE basics, elementary feedback theory
http://seniordesignlab.com/sd_docs/Links/Op_Amps_For_Everyone.pdf is often recommended
unfortunately there are few project amps/kits with TPA6120, even fewer with multiloop feedback
it is very ambitious to jump into such an advanced project without much more of the basics to guide you in designing/building/debugging/testing
in the multiloop amp the outer feedback components determine the overall gain of the amp
if you want more than Av =5 then you change the outer loop feedback
http://seniordesignlab.com/sd_docs/Links/Op_Amps_For_Everyone.pdf is often recommended
unfortunately there are few project amps/kits with TPA6120, even fewer with multiloop feedback
it is very ambitious to jump into such an advanced project without much more of the basics to guide you in designing/building/debugging/testing
in the multiloop amp the outer feedback components determine the overall gain of the amp
if you want more than Av =5 then you change the outer loop feedback
Last edited:
TPA6120A Composite Schematic.
Thank you for your help jcx.
I'm uploading the schematic of the TPA6120A using the composite topology.
If there is an error or problem in the schematic, please let me know.
The gain of this amplifier design is 5.64 v/v.
Regards.
Alfredo Mendiola Loyola
Lima, Peru
Thank you for your help jcx.
I'm uploading the schematic of the TPA6120A using the composite topology.
If there is an error or problem in the schematic, please let me know.
The gain of this amplifier design is 5.64 v/v.
Regards.
Alfredo Mendiola Loyola
Lima, Peru
Attachments
The Opa2134 has an open loop gain of 50dB at 20khz.
My close loop gain is 5.64 v/v = 15 dB
Excess loop gain = 50db- 15db = 35db = 56.23 v/v
So my output impedance is 10 Ohm / 56.23 = 0.177 Ohms.
The 10 Ohms is the resistor at the output of the TPA6120.
Is my calculus ok?
source: (Page 13-4)
http://www.analog.com/static/imported-files/application_notes/AN-356.pdf
Regards
Alfredo Mendiola Loyola
close, with the TPA set with its local feedback to Av=+2, it adds 6 dB to the outer loop gain which you get to count towards the outer loop gain too
the next thing to know is what effect does it have on your headphone - typically < 1-2% level vs frequency variation is found to be imperceptible, with considerably more roll off near the 20 Hz and 20 kHz audio frequency extremes allowed
the next thing to know is what effect does it have on your headphone - typically < 1-2% level vs frequency variation is found to be imperceptible, with considerably more roll off near the 20 Hz and 20 kHz audio frequency extremes allowed
Last edited:
in its operating range the output Z of the OPA134 even open loop is likely around 100 Ohms so its fine driving the TPA +input
if anything I would add a Zobel of 50 Ohms in series with ~200pF to gnd to terminate the TPA +input at >10s of MHz, well beyond where we can rely on the OPA134
if anything I would add a Zobel of 50 Ohms in series with ~200pF to gnd to terminate the TPA +input at >10s of MHz, well beyond where we can rely on the OPA134
- Status
- Not open for further replies.