TI TPA3100D2, Amazing Class D Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Member
Joined 2008
Paid Member
Just want to bring this class d chip to light. It all started when I came across a review for the TBI Millenia G3 at TNTAudio - a little known amp that has been out since 2008 with very little attention or fanfare. In that review it beat out the Trends Tripath and the Dayens Ampino. Almost a year passes and now another audio site covered the amp last month also giving a stellar review. Spurred on by that review a tour is under way at Audio Circle with everyone (including myself) being stunned by the sound quality.

So I took a look inside the Millenia G3 and found a pretty simple layout with the Texas Instruments TPA3100 at it's heart. There is ZERO information on the web about this chip aside from TI's website. So in true diy form I ordered the eval board from Digikey ($149) and have been playing with it. It does about 10w at 12v up to 32w at 24 for 8 ohm.

I have had the TK2050s, IR2092 (Classdaudio amps, love them) based amps, SimpleSE tube amp, Geddes-approved Pioneer D912 - all kinds of amps and this little TPA3100 is a contender. And yes by contender I mean I think it sounds better than all of them. Downside: It's low power, topping out at around 32w at 8ohms. But it's a powerful 32watts. I did no modding aside from bypassing the chintzy speaker posts. Sounds incredible with batteries but also great with a simple linear PS. I have a Meanwell SP150 coming to try as well. So who knows what would be possible if some of diyaudio's sharper minds took a look!

I myself am not an electronics expert, I just like to build things so I leave it to to others to sort out the details. But I felt it was important to bring this unknown chip to this forum because it is really, really, REALLY good.


An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
Just curious, where do you get the figure of 32W into 8R with a 24V supply? The datasheet says 21.8W @ 10% THD.

OK, now I see you were quoting the specs for the Millenia G3.

Those specs are for the TPA3106D2 though, so either they're lying about the specs, or you mistakingly thought the TPA3106D2 in the unit was a TPA3100D2 . :scratch:

BTW, if you try to crank the volume on that eval board using a 24V supply you will run intro problems. It's only recommended to run it at 18V for an 8-ohm load (see the datasheet for the IC). Higher than that requires added cooling.
 
Last edited:
Seems like I've had a "duh" moment here. The TPA3100D2 is just a dual channel version of the TPA3106D2. The power difference is only due to the thermal limits of the IC package.

Does the Millenia have any sort of heatsink over the IC? For the price of that amp, you think they would have made it dual mono with two 3106D2.
 
Last edited:
Member
Joined 2008
Paid Member
Seems like I've had a "duh" moment here. The TPA3100D2 is just a dual channel version of the TPA3106D2. The power difference is only due to the thermal limits of the IC package.

Does the Millenia have any sort of heatsink over the IC? For the price of that amp, you think they would have made it dual mono with two 3106D2.

No heatsink on the Millenia. I don't know about the other TPA boards, but this one does something special.
 
Member
Joined 2008
Paid Member
Hi Wushulu
Can you elaborate more on the sound difference between this board and the TK 2050.
regards
kp93300

It's been a long time since I had the TK2050 but my impressions from that time was in line with many others: impressive but with a slightly analytical (?) quality. Hard to articulate. I have also heard Virtue Audio implementation and overall the same. I preferred the IRS2092 boards to the 2050 by a wide margin. YMMV.

As for the TPA/Millenia itself again hard to describe - the link to the audio circle thread has other people's feedback, which may be more helpful. Closest I can say is you feel like you are listening to a master tape - low level information retrieval, a kind of you are there experience. Sorry after a while language gets in the way. I could say even MORE transparent, revealing, organic, deep, natural, etc. but that's not helpful really. I just know that what I hear is not just good, but great. Partly why I posted here so maybe the technical geniuses around here can pick it apart and share thoughts.

Hard to believe it was made for televisions...:eek:
 
Last edited:
No heatsink on the Millenia. I don't know about the other TPA boards, but this one does something special.

Hmm.... well, I don't believe in magic.

I think TI knows best and they derated the power output of the 3100D2 for good reason. With only the thermal pad of the IC transfering heat to the PCB, there's no way that little IC can sustain full output on both channels with a 24V supply.

Stressing an IC to the limits of its SOA is somethig I would expect from some cheap bargain amp maker, not a boutique audio company.

Oh well, glad I can make my own dual mono pair for less than $50.... $500 for a chipamp is too extreme for me.
 
Member
Joined 2008
Paid Member
Who said anything about magic. Please indicate references to magic. And I don't know what the rest of the rambling has to with anything. You're too awesome to be fooled. You could make a better/safer amp than TBI. That's great news. Anything constructive about the chip or design itself?
 
Hi Wushulu,
Any difference in the sound between the evaluation board and the TBI implementation.?
I felt my battery powered Trend audio with TA2024 and some mods in the input and power supply caps on board is very transparent already.
I am always willing to try new ideas.
You post are intriguing and tempted to buy one if not for the FEDEX fee of US 40 !
The TI chip is very cheap though.
thanks

kp93300
 
Member
Joined 2008
Paid Member
Hi Wushulu,
Any difference in the sound between the evaluation board and the TBI implementation.?
I felt my battery powered Trend audio with TA2024 and some mods in the input and power supply caps on board is very transparent already.
I am always willing to try new ideas.
You post are intriguing and tempted to buy one if not for the FEDEX fee of US 40 !
The TI chip is very cheap though.
thanks

kp93300

They sound very similar...
 
So, are these the easy and relatively low cost mods that need to be made to this board?

1. Change input RCA's?
2. Change speaker output posts?

And runing the board off 24V batteries?

Wushuliu, is there any update on running this board off the Meanwell? How does running the board off batteries and a linear PSU compare?
 
Member
Joined 2008
Paid Member
So, are these the easy and relatively low cost mods that need to be made to this board?

1. Change input RCA's?
2. Change speaker output posts?

And runing the board off 24V batteries?

Wushuliu, is there any update on running this board off the Meanwell? How does running the board off batteries and a linear PSU compare?

The RCAs are a problem. Maybe someone can figure it out: They have differential inputs with jumpers by them for unbalanced. I have no idea what balanced rca's exist that could be used in the first place. Anyway, I tried to bypass the rca inputs but nothing worked. I tried to solder connections directly to the pads which normally works fine but there is something to how the differential input is set up that I am missing because nothing works. So I have to plug an rca into the jacks and run wires from the plugs to the chassis inputs, creating an unneeded extra connection.

If someone can explain how to avoid having to use the stock rca that would be great.

As for binding posts, up to you.

Power supply is also a matter of choice. Batteries work great and last a good while. I use a Meanwell SMPS and it sounds terrific. Honestly I can't detect enough of a difference to stick with batteries. Probably the bateries are lower noise, slightly quieter background but convenience wins and I am impressed with the Meanwell.

I do want to add that having compared with the TBI, the TB still sounds amazing and slightly better than what I have with the devlopment board. I am using a DCB1 buffer pushed to 800mA and in certain ways it sounds as good or better but the TBI really does have a very clean, low noise profile to my ears and a more holographic 3-D delivery. I can only guess this has to do with the Opamp OP2889 buffer. Hopefully smarter minds can explain why this may be.
 
Last edited:
Member
Joined 2008
Paid Member
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.