22uF of what voltage rating? I'm guessing there weren't much uF left due to DC bias. As for the TPA3132D2 (TPA3116 in QFN package) I use in my layout, there isn't any hiss at all. (400-1200kHz dead silent) I even don't use any 10R+330pF snubbed nor LC filter. Haven't tried without but I use a simple RC-Filter on the AVCC-pin. (10R 1uF, increasing the resistance is not a good idea there)
Additionally there a 1k Rs right in front of the input pins to block HF coming from GSM (forming an RC with the help of stray capacitance).
I'd consider to check your AVCC decoupling first.
PCBs avail, mainly 0402 smd.
Last edited:
Yeah. Remove the input coupling caps and see if the hiss goes away - audio will obviously go away too.
If hiss stays, it could be something like terrible decoupling, wrong value cap on GVDD. If it goes away, I'd finger the input circuitry.
I couldn't find the one I'm using in a 110V version but this would probably do the job; 1pcs 60W 110V Professional Electric Adjustable Temperature Soldering Iron Tool | eBay.
For 230V users I have good experience with this particular one, Pro Adjustable Constant Temperature 60W 220V US Soldering Iron Tool | eBay
Thanks for the tips! I'll look into that!
Doctormord, is that rc filter (10R 1uF) between pin 17 and pin 18 then?
For 230V users I have good experience with this particular one, Pro Adjustable Constant Temperature 60W 220V US Soldering Iron Tool | eBay
Thanks for the tips! I'll look into that!
Doctormord, is that rc filter (10R 1uF) between pin 17 and pin 18 then?
Last edited:
Doctormord, after some looking I found the answer from Abraxalito. (page 300 tpa3116 thread.) I will try this mod first. Did you try different values on that filter?
For SMD soldering, my best investment was magnification (I already had a good soldering iron):
http://www.amazon.com/Donegan-Optical-OptiVISOR-Headband-Magnifier/dp/B0068OSIIS
http://www.amazon.com/Donegan-Optical-OptiVISOR-Headband-Magnifier/dp/B0068OSIIS
those look really uncomfortable, what's wrong with using cheap drug store glasses
I know I know it's not industrial chic
YIhaw Solderstation http://youtu.be/uPRSA0ub2uY
who cares guy, it's 15 buy a couple of real tips for 20 more http://youtu.be/uPRSA0ub2uY
Last edited:
A key benefit of the glasses is the range of magnifications. I find them comfortable and often leave them on for hours. I don't consider them chic...indeed they are the fast track to a a bad hair day. They do get more use than my Mantis stereo magnification system.
My drug store glasses are for reading and less demanding work. The are one of the many 'gifts' that come with middle age.
My drug store glasses are for reading and less demanding work. The are one of the many 'gifts' that come with middle age.
That's what I said. 🙄 10R+1uF, same like 1R+10uF (same freq). Your millage may vary, it depends on the layout. (You won't try anything greater than 100R, making things worse)
470uF is absolutely not necessary, but Abraxalito is also using some 470000uF for PVCC.. 🙄🙄🙄 Not my business.
I should have mentioned they are hinged so you can just push them up and out of the way.
Yay decoupling simulation. Trying various parts and package sizes now to achieve a low wideband impedance.
Black curve (unlabeled) is the ugly one. When capacitor #1 is above its series resonance and capacitor #2 is below, you effectively get a L and C in parallel which form a tuned circuit, creating a high impedance at their resonance. Ideally these don't end up anywhere where the amp will put a harmonic...
Black curve (unlabeled) is the ugly one. When capacitor #1 is above its series resonance and capacitor #2 is below, you effectively get a L and C in parallel which form a tuned circuit, creating a high impedance at their resonance. Ideally these don't end up anywhere where the amp will put a harmonic...
so the black line is the parallel combination of all the decoupling caps listed, in ohms at the chips power pin? I reckon the pcb has some losses to smooth all those woop-ti-doos out at least > 10 MHz
Last edited:
Yep. I'm currently ignoring distributed PCB plane capacitance (hard to calculate that without special tools) and assuming 1nH/mm PCB trace inductance.
hi gmarsh,
Is there any merit in using TPA3118 in the board?
Without heatsink, the caps can be placed really close to the pins and changing caps may be easier..
I presume you have also include the avcc pin isolation as described in thread 2992.
regards
kp93300
Is there any merit in using TPA3118 in the board?
Without heatsink, the caps can be placed really close to the pins and changing caps may be easier..
I presume you have also include the avcc pin isolation as described in thread 2992.
regards
kp93300
I'm using TPA3118 currently. TPA3116 means I'd have to move the decoupling further away as you're saying, and it also means I'd have to find a suitable heatsink.
TI told me in the past that PVCC and AVCC absolutely must be the same voltage - and I'm concerned that if you add a RC filter on AVCC, PVCC can be significantly greater than AVCC when you first apply power due to the RC time delay.. hopefully not breaking anything. I'll check with TI on this one first, then consider it.
TI told me in the past that PVCC and AVCC absolutely must be the same voltage - and I'm concerned that if you add a RC filter on AVCC, PVCC can be significantly greater than AVCC when you first apply power due to the RC time delay.. hopefully not breaking anything. I'll check with TI on this one first, then consider it.
You'd better decouple it. (I.e 10R 1uF) There are some discussions about that at the TI forums.
Hey Santa! 😉
Are you piled-in deep my Eastern friend?! 😀
(gmarsh way be under a metre ... uh, 3 feet of snow! Perfect day for DIY.) 😀
Funny you should say that... when I coined the phrase, 'wiener' [Wiener, 'we have a wiener / winner', hasslehawk] I wanted an all-feature, 'properly' designed PCB utilizing the excellent class-D / digital design knowledge here on Diyaudio.
Then I wanted to compare The Wiener to all the other 3116 designs as sort of a 'benchmark' (measurements and listening).
So I definitely want the 3116 to make comparisons as accurate as possible. The 3116 can be swapped in your design, correct?
As far as heatsinking ... I guess you guys just Rock! more than I do ( 🙁 ) because the 3116 doesn't even get warm under 19V and only makes a small heatsink warm at 24V.
For power testing with a sine wave maybe but ... what's your experience?
I think your board with the close decoupling (3118) with a 3116 installed and a small (or 'shaved' to clear the caps) heatsink would be ... perfect!
Did you add all the functionality? Power output adjust ( see what it's 'clipping' sounds like ), fault pin detect / reading?
What about that double diode circuit from one of those TDA729X threads?
[All I can find is a one diode circuit, I'm sure I saw a 2-diode version ... my Google-fu has failed master.... ]
It has 2 (exotic; I recall extra 'wings' on the diode symbol, schottky, barrier?) diodes, one reversed in series with the two ps pin. It keeps the pins a max of one diode drop difference. Choose the diodes to keep them as close as needed?
Cheers,
Jeff
Last edited:
Can you give me search term to find those discussions at TI forums about tpa3116/18? Or do you mean diyaudio threads, then no need.
Yeah, I can't find any posts on TPA3116/3118 AVCC decoupling on the E2E boards, other than when I asked about running different voltages on the two pins several months ago.
I just have PLIMIT tied to GVDD. Makes sense for some applications, didn't really see the need for hooking it up here.
In my experience with the Boominator card, the TPA3118 will happily put out 50Wx2 RMS (20v peak into 4 ohms) at 1KHz without any issues, at room temperature - gets a bit warm to touch, but silicon can handle far hotter temperatures than fingers. If you're worried about heat or want high temperature operation, mount the card to a metal plate with standoffs and put a Berquist gap-pad between the card and the plate.
Nah, Boston got the worst of it and we got probably 10cm here. Shovelled that off the driveway, then it rained and took care of the rest. It's been a pretty damn easy winter, I guess all that styrofoam I burned when I was a pyro kid paid off 🙂
Hey Santa! 😉
Are you piled-in deep my Eastern friend?! 😀
(gmarsh way be under a metre ... uh, 3 feet of snow! Perfect day for DIY.) 😀
Yeah, the TPA3116 and TPA3118 have the same pinout and either can be installed on this board. Only difference between the two chips is the thermal pad on bottom versus top - both chips have the exact same silicon die so I wouldn't expect any difference in performance/sound.
You'll definitely need to shave/modify a heatsink, pretty much every part around the TPA (decoupling caps, bootstrap caps, etc) are all taller than the HTSSOP package itself so it'll be a close fit. I also haven't provided any means to mechanically attach a heatsink to the board. I'll just say if you use a TPA3116, you're on your own 🙂
I just have PLIMIT tied to GVDD. Makes sense for some applications, didn't really see the need for hooking it up here.
In my experience with the Boominator card, the TPA3118 will happily put out 50Wx2 RMS (20v peak into 4 ohms) at 1KHz without any issues, at room temperature - gets a bit warm to touch, but silicon can handle far hotter temperatures than fingers. If you're worried about heat or want high temperature operation, mount the card to a metal plate with standoffs and put a Berquist gap-pad between the card and the plate.
IMO it wont be "proper" not to fully support both 3116 and 3118 with an external heat sink that is available somewhere. IDK seems worthwhile to try to make it happen. seen a long skinny one somewhere.
It was on the wish list from day one! not sure yer baseline soldering underneath the chip sounds too newbee easy either, can we used some leftover artic silver or sumptin?
It was on the wish list from day one! not sure yer baseline soldering underneath the chip sounds too newbee easy either, can we used some leftover artic silver or sumptin?
Last edited:
- Status
- Not open for further replies.