Realistically achievable to build a one off TAS3152 amp ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello,


Using 3e-Audio TPA3251 + some DAC is certainly the way to go. However, I wonder if with some dedication, it is possible to build a one off TAS3251 board (new TI chip with digital input, which is what I need).


I have access to a Fablab with CNC and chemical bath for PCB making. Other option is those PCB manufacturers.



Design would be a much simplified version of the TAS3251EVM board: only the I2S and I2C inputs and the TAS3251 surroundings.


Is this achievable for the "first time amateur", or an almost certain failure ?


Is manual soldering of a TAS3251 achievable, or not really except for experts (we have USB microscope)?



Best regards,


JM
 
I'd say designing the board (for maximum performance) would / will be much more complicated, than soldering a 0.635mm-pitch chip (which isn't even in a QFN or BGA package) :)

Manual soldering shouldn't be an issue - i've soldered 0.5mm-pitch QFP packages manually.

Of course, good soldermask on the board helps a lot, as well as good flux, and good solder-wick (for the moments when solder decides to bridge across 2-3 pins) :D
 
Last edited:
Thanks for your feedback. About the benefits of the TAS3251... Let's say that it can be appealing to some people for cost and ease.


Cost: should be similar to a TPA3251.



However, 2xI2S balanced DAC boards are about 55€, compared to 75€ for the TPA3251 board


Ease: may be the more important for a group of people with disabilities in analog electronics and higher comfort with microcontrollers. With a I2S/I2C brick, some people working on multichannel active speakers, using DSP, are comfortable.



I'm in that case, and I'm ready to trade-off some flexibility with the confidence not to spoil the implementation of the separate componants.


However, I would be happy to have your skill and be able to design and build the boards and combos I need :)


JMF
 
You’d think it’s all of the possibilities with that chip that someone would put low jitter clocks on it and try to get greater clarity than a TPA3251 driven by a more traditional DAC setup. I’m actually very curious about that chip

Those TI chips sound really good. Doubtful it sounds as good as Hypex NCore, it definitely doesn’t sound as good as my Neurochromes, but it’s really viable solution for folks with car stereo setups like mine where I emphasize clarity and quality of sound in my creation

Power is definitely an issue and class A/B rigs drain a significant amount of power
 
You’d think it’s all of the possibilities with that chip that someone would put low jitter clocks on it and try to get greater clarity than a TPA3251 driven by a more traditional DAC setup. I’m actually very curious about that chip

Those TI chips sound really good. Doubtful it sounds as good as Hypex NCore, it definitely doesn’t sound as good as my Neurochromes, but it’s really viable solution for folks with car stereo setups like mine where I emphasize clarity and quality of sound in my creation

Power is definitely an issue and class A/B rigs drain a significant amount of power


If you have hands on experience of sonic differences between the Tpa3251 and modulus86, I'm really interested to hear about it.


So difficult to make informed choices without possibility to hear the different options.


Jmf
 
Neurochrome Modulus is the best amp I’ve ever heard. Probably similar to the NCore in clarity but I have yet to hear the NCore in person. The distortion figures are silly low for those devices. More than a 100db down for the Modulus for the first harmonic

Super in your face clear amplifier.

TI TPA3250 sounds great. Still very high resolution but I haven’t extensively compared them to the Modulus amps. My Modulus amps are kind of risky. Blew out two pairs of woofers chip broke off from me not having sufficiently solid heatsink mounting. Input is DC coupled and is subject to passing DC I would presume if there wasnt sufficient decoupling from the feeding device (iPhone) in my case

This is a car installation

I suspect that the weakness in this system is the old Audison DSP I’m using. Seems to smear the highs and changing out the cheap line level cables helped. Der Kommissar by After The Fire has chimes that were smeared by it until I changed it out for better cables. He next upgrade for me is a more modern DSP

TI chip sounds very detailed still. Not as good as the Modulus but still very precise. That’s the best I can describe it. I’m in San Francisco btw
 
Follow up:

It should be noted that they both (TI and Neurochrome) sound far better than anything commercially out there. The Arc Audio XDI amp I have is a very good and powerful amp. About as clear as any other analog amplifier in that category. The grit or distortion from analog and digital are different in quality but it’s still very evident that there is unclarity. The flavor of it, digital or analog, doesn’t matter to me I just want it to go away

XDI amplifiers are arguably better in clarity than any commercial car audio amp I’ve heard in that category. The only one that was clearer was a medium end Mosconi amp. It was definitely clearer when I was listening in the shop which promoted me to look into improving my amplifiers. I didn’t think I could hear the difference but it was pretty clear in the shop that you can. Just a matter of time and exposure to different amplifiers

My first experiment into improving amplification was building the Neurochrome Modulus 86 amp. It was a dramatic improvement. I have pretty good memory of it and the detail and quality is similar to the Focal Trio6 Be series studio monitors but I could hear what I think is the sonic quality of the Focal K2 Kevlar drivers in my rig. Moving to a Focal Utopia set with the Modulus amps, I think, would bridge the gap between My K2 currently

The next improvement is trying to get a Hypex DLCP DSP to working my car. The SMPS search to power it is just impossible. It’s so hard trying to find power supplies with suitable output voltages for automotive electrical systems. This requires at least a dual rail 16 volt supply. But if I can get all of that working I can move a full digital path down towards the amp
 
Last edited:
Waouhhhh. Modulus86 in your car : that's not so usual ;-)
Thanks for the feedback on both Neurochrome and TI.
From what you say, the TI performance / burden / cost ratio seems hard to beat.
JMF

I'm kind of crazy like that.

BTW most car have their tweeters badly misaligned from the woofers. They need to form a normal that shoots from a plane formed by both tweeters/woofers so that it tilts upward to the listener. It's normally too low in stock configurations. Tweeter and midrange orientation is critical to prevent a null or time misalignment at the crossover frequency. I put mine as forward on the dash as I could. Really made a huge difference

The car is basically my audio electronics lab. I spend a lot of time in it so the time investment is worth it.

Neurochrome Modulus 86 was my first adventure into amplifiers after I did a bunch of research and was impressed by the analysis thread. Very happy I did. Tom Christensen is a design level professional in the field of op amps and knows a lot about circuit validation and passive components.

My target now is to make my car sound as much like the Focal Trio6 Be series studio monitors. Absolutely the clearest things I've ever listened to. Heard them for the first time at NAMM 2017.

I found out about the TI chips from a booth at NAMM btw and decided that they were worth looking at after reviewing their distortion characteristics
 
The TAS3251 is mainly a TPA3251 with a PCM5242 in on Chip. If you look into the datasheet, the power amp does still have analog inputs, so the DAC -> ADC connection is still extern. I actually don't see the benefit of this IC. o_O


Doctormord,


If I decides to follow the TPA3251 path, which DAC would you advise to benefit from the balanced inputs ? In TI range, it seems that it is mainly PCM5242 and PCM5252 that provide the balanced outputs. and their package seems really difficult for hand soldering :-(


(Abraxilato LingDAC is another option, but obviously some work to get built)


By the way, looking deep in the TAS3241 datasheet, it happens that the included DSP is a reduced version from the PCM5242 one. Especially it seems that there is no way in the TAS3251 to delay samples from one channel, which is useful for time alignment (I need it for my project) when it is included in the dedicated chips.


JMF



JMF
 
...
By the way, looking deep in the TAS3241 datasheet, it happens that the included DSP is a reduced version from the PCM5242 one. Especially it seems that there is no way in the TAS3251 to delay samples from one channel, which is useful for time alignment (I need it for my project) when it is included in the dedicated chips.
JMF
JMF

Raspberry Pi has two i2s inputs (header, HDMI). You could potentially delay the data with a simple FIFO written in a tight real-time loop in software after reading in the data from an i2s input board.

The TAS3251 is an interesting chip but the bring-up on it is heavy. It's a very complicated device and requires a lot of heavy lifting to get something basic working. Unless you're planning on making a ton of these devices, you're better off with the pricey evaluation board just for basic DSP/DAC hacking.

We can keep talking about it though :)
 
The TAS3251 is an interesting chip but the bring-up on it is heavy. It's a very complicated device and requires a lot of heavy lifting to get something basic working. Unless you're planning on making a ton of these devices, you're better off with the pricey evaluation board just for basic DSP/DAC hacking.

We can keep talking about it though :)


This is what triggers my curiosity in a world that changed a lot and where making PCB is more easy than some years ago, but where soldering may be more difficult than passthrough components.


TI is doing a pretty good job, providing all infos (and gerbers) of their evaluation boards, BOM and so on. When I see the difficulty to tweak existing BOMs, it seems sometime easier to do it well from ground up (correct coil, correct power supplies, the inputs and config you want...). I have bad memories with the FX-D802 where corners have been cut in different parts of the design, no way to access I2S...



But sure that I should not under estimate the work. And I'm a bit disappointed about the uDSP reduced functional set?


JMF
 
This is what triggers my curiosity in a world that changed a lot and where making PCB is more easy than some years ago, but where soldering may be more difficult than passthrough components.
...
But sure that I should not under estimate the work. And I'm a bit disappointed about the uDSP reduced functional set?
JMF

I’m happy they put any kind of DSP there. It’s not for them to put memory for a delay in their chips. It’s for folks like us to program it up via some embedded computer

You can always start a collaborative PCB design using PCB construction tools and work on it. Layout can be critical for the demanding component handling high frequency though

That’s the thought. The OSH (Open Source Hardware) folks are more than willing to make your PCBs btw. Good folks
 
I did the programming job for my current setup, with the DSP aand samples delays in a STM32 Nucleo board. Works fine.


However if you want to use the TAS3251 in active speakers, with one channel for the woofer and the other for the tweeter using the internal DSP, then you have a issue as you input one channel, but high/low separation is in the chip and you have no mean to delay the High or the Low :-(


JMF
 
Why not delay it using two i2s lines ?


Can be done, I agree. That's why I selected the Stm32 nucleo boards.


However, the vast majority of devices, like RPi (excluding tweaks/twists like using HDMI out) have only 1 I2S exposed. So would be much more versatile to have everything onboard in the DSP/amp. frustrating to be so near from the "ideal" solution.


JMF
 
Well, you should post the code and plans on github or something like that. I would be interested in building such a device.

TAS3251 is going to awhile I suspect but I can't wait for folks to come up with low jitter clock boards for it. It would be really interesting as you wouldn't need a DSP anymore as much of the filtering ability has been pushed onto the chip

It would be a good replacement for the TPA3251 I'm think about getting next or TPA3255
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.