Hello everyone!
I was wondering if it is possible to run TPA3255 in a composite amp configuration? LM4562 is something that everyone seems to be using as error correction opamp, so I figured will be nice combo.
Where can I find resources on that kind of information?
Thanks in advance,
Pavlo
I was wondering if it is possible to run TPA3255 in a composite amp configuration? LM4562 is something that everyone seems to be using as error correction opamp, so I figured will be nice combo.
Where can I find resources on that kind of information?
Thanks in advance,
Pavlo
It is possible to lay another loop around class - d amplifiers if you knao how to do it. You might improve PSSR this way for instance. If you want to reduce THD things get much more complicated however. Natural sampling PWM amplifiers suffer from an effect that is called ripple aliasing. If things go wrong you might even increase THD with a higher feedback factor. The TPA3255 does actuallly have an internal feedback network that is reducing the effects of ripple aliasing.
Laying an additional loop around a switching amp is always more demanding than around a linear amp.
Regards
Charles
Laying an additional loop around a switching amp is always more demanding than around a linear amp.
Regards
Charles
Hi,
I did this using TPA3118 BTL based basic boards in order to have a even more complex circuit i.e current feedback audio amps.
Provided that it's carefully done in terms of phase/loop gain & with the proper voltage/current protections while using the lowest gain at the tpa3118 level, it's working perfectly!
I bought some tpa3255 boards to do the same. still not have had the time to test it as the boards need substantial modifications... but I'm quite confident 😉
Another project on my desk, (among a lot of others.. so many men on the moon so little time...) I also want to explore a different path, i.e hybride A/D circuit, i.e the Class D stage does the main power amplification job, (to maintain efficiency) with a limited BW, and an additional conventional analog amp, with significant lower power but higher BW does the 'fine' tuning of the output signal but adding dedicated current to both cancel/minimize digital commutation spikes and improve signal resolution. Has anyone already tried this?
Thx
Cheers
N.
I did this using TPA3118 BTL based basic boards in order to have a even more complex circuit i.e current feedback audio amps.
Provided that it's carefully done in terms of phase/loop gain & with the proper voltage/current protections while using the lowest gain at the tpa3118 level, it's working perfectly!
I bought some tpa3255 boards to do the same. still not have had the time to test it as the boards need substantial modifications... but I'm quite confident 😉
Another project on my desk, (among a lot of others.. so many men on the moon so little time...) I also want to explore a different path, i.e hybride A/D circuit, i.e the Class D stage does the main power amplification job, (to maintain efficiency) with a limited BW, and an additional conventional analog amp, with significant lower power but higher BW does the 'fine' tuning of the output signal but adding dedicated current to both cancel/minimize digital commutation spikes and improve signal resolution. Has anyone already tried this?
Thx
Cheers
N.
Last edited:
There are many topologies with small linear amplifiers combined with switching amplifiers. One of the most recent and successful attempts is the Devialet amplifier. You may find other examples when you search for switch mode assisted linear amplifier (SMALA). There was one sugggestion by a Prof from Munich which was feeding the correction signal form the small linear correction amp via the cap of the output filter.
There is also an old JAES article on feedforward error correction that was suggesting a combination of switching amp and a small linear amp as an example.
Did you measure THD and IMD of your nested feedback loop ?
Regards
Charles
Hi Charles,
Thanks for the info.
I already had a thorough investigation of Devialet's concept (I wanted to work in their R&E dpt... ;-) ! ) but their systems are way too sophisticated + not really for my ears, being a classical music lover + musician, sound engineer.... for me the quest for the 'natural' sound is the ultimate goal...
My idea of an hybrid amp is still just in my head, but I already thought of something similar to what you describe of the Prof approach... but knowing the devil side of using loops... simple ideas require long design/test/adjustment sessions...
As for THD and IMD measurements, I didn't.
I checked with the 200MHz (TX2445) BW analog oscilloscope that the system is stable (no oscillation) feeding the amp with pure sinusoidal audio frequencies, also checked 'visually' the feedback error signal 'shape', the noise floor etc
But most importantly I checked with my ears at ' human' sound level, without and with feedback connected to Focal 3 way column speakers in bi-amp configuration, with active LF/ MH-HF preamp/filters.
I compared this to my 'old' MOS current feedback analog amp config, ( >> 100 kHz BW ...) and it's very acceptable for me. (again at reasonable sound level; the MOS amp is able to deliver 2X200W per channel... not really needed & quite greedy... times are changing... ).
Cheers
N.
PS: I also have a parallel project, to build a 'linearized' switched powersuply i.e with active noise reduction ...
Thanks for the info.
I already had a thorough investigation of Devialet's concept (I wanted to work in their R&E dpt... ;-) ! ) but their systems are way too sophisticated + not really for my ears, being a classical music lover + musician, sound engineer.... for me the quest for the 'natural' sound is the ultimate goal...
My idea of an hybrid amp is still just in my head, but I already thought of something similar to what you describe of the Prof approach... but knowing the devil side of using loops... simple ideas require long design/test/adjustment sessions...
As for THD and IMD measurements, I didn't.
I checked with the 200MHz (TX2445) BW analog oscilloscope that the system is stable (no oscillation) feeding the amp with pure sinusoidal audio frequencies, also checked 'visually' the feedback error signal 'shape', the noise floor etc
But most importantly I checked with my ears at ' human' sound level, without and with feedback connected to Focal 3 way column speakers in bi-amp configuration, with active LF/ MH-HF preamp/filters.
I compared this to my 'old' MOS current feedback analog amp config, ( >> 100 kHz BW ...) and it's very acceptable for me. (again at reasonable sound level; the MOS amp is able to deliver 2X200W per channel... not really needed & quite greedy... times are changing... ).
Cheers
N.
PS: I also have a parallel project, to build a 'linearized' switched powersuply i.e with active noise reduction ...
Note that the TPA is inverting output, and you might need to add another opamp to "double invert".
Another thing that comes into my mind to note is the TPA is running in single rails, which i have no idea which PINs is the COM pin for the TPA, while your opamp most likely is built DC coupled, this increased the difficulty to composite the FB loop too I assume.
Another thing that comes into my mind to note is the TPA is running in single rails, which i have no idea which PINs is the COM pin for the TPA, while your opamp most likely is built DC coupled, this increased the difficulty to composite the FB loop too I assume.
Hi !
thanks for the warning, you are right. this also means that in differential BTL mode, the inputs become somehow differential, inverted vs the outputs.
I anyway would advise to use an additional opamp to master the input impedance, BW and phase of the feedback loop, with capacitive coupling.
There is apparently no access on the chip to the COM voltage.The input signals are to be fed in through a capacitor at the input pins,with a max voltage directly at pin referred to ground of -0.3 V / +7V signals.
And yes, as for a lot of single chip class D amps it's single railed . The outputs are by design, 'floating' around PVDD/2, hence capacitors in SE mode, and DC protection integrated in the device.
So the feedback loop can not be DC coupled, but to simplify it's design I use dual rail DC supply for the opamps, paying attention to the above limitations.
This all is easy on paper, much trickier in really.. hence my first trial with the TAP3118, with much lower voltages/power, and carefully chosen protection parameters allowed by the device.
Cheers
N.
thanks for the warning, you are right. this also means that in differential BTL mode, the inputs become somehow differential, inverted vs the outputs.
I anyway would advise to use an additional opamp to master the input impedance, BW and phase of the feedback loop, with capacitive coupling.
There is apparently no access on the chip to the COM voltage.The input signals are to be fed in through a capacitor at the input pins,with a max voltage directly at pin referred to ground of -0.3 V / +7V signals.
And yes, as for a lot of single chip class D amps it's single railed . The outputs are by design, 'floating' around PVDD/2, hence capacitors in SE mode, and DC protection integrated in the device.
So the feedback loop can not be DC coupled, but to simplify it's design I use dual rail DC supply for the opamps, paying attention to the above limitations.
This all is easy on paper, much trickier in really.. hence my first trial with the TAP3118, with much lower voltages/power, and carefully chosen protection parameters allowed by the device.
Cheers
N.
I forgot also to metion that there is an application note from TI explaning additional post filter feedback loops to improve the overall performance. It'all done with passive components, but it gives a lot of elements as for stability etc... https://www.ti.com/lit/pdf/slaa788
Cheers
N.
I believe the dual rail operated MA5332MS will be an easier to implement with the composite feedback loop. But the residual noise is at the high site @ 250uV with the referenced designs.
Hi!
The 5532 is a classic; probably the opamp the most commonly used in a lot of audio devices. It's from 1979 !!... in those days it was 'fabulous' compared to the existing ones in terms of Gain X BW and noise... unbeatable price vs quality-performance ratio....
But I always considered its performances as OK for 'average' low power audio boards in the range of several volts peak to peak. I saw it used in a lot of RIAA turntable preamps, and there, you can immediately hear that the noise performance is just not good enoug, even with discrete differential J-FET stage added in front of it as often seen....
I know some will oppose that the background noise of the vinyl +TT rumble covers it, but nonetheless if you manage to have a decent TT, it makes a difference and then you can distinguish between carefully engineered vinyl masters and poor ones. (I have 2 LPs of the same recording of a Gershwin concerto, from the same brand, but made with 2 decades difference. Tthe second one has a substantial higher high frequency BKG noise...).
For this type of circuit, the more expensive THAT1512 opamp is a lot better even more, quite astonishing when used as the first stage for the phono preamp.... quite competitive vs discrete component ones.
But as an acceptable/affordable alternative for my ears to the 5532 , the LME49720, an improved version of the LM4562 (National Semiconductor legacy). It's BW/Gain and most importantly noise figures are respectable. Pin to pin compatible with a lot of dual opamps such as the 5532. That's the one I usually use for line level audio signal treatment such as the feedback loop discussed here.
Cheers
N.
Hi
sorry I read MA 5 532 instead of MA S 5332...! my brain being used to juggle with this reference..
Cheers
N.