Hmm, but AVDD is derived from VDD as well, so the gate drivers aren't "isolated" then. To my understanding, the RCs are ripple/noise reflection filters.
Interesting find.
Interesting find.
No question that film caps provide better THD performance than MLCC's, if the voltage swing on the cap in the circuit is enough to cause the uF's to change with the audio signal.
Try repeating that plot at lower test frequencies and see how bad it gets.
Try repeating that plot at lower test frequencies and see how bad it gets.
Well, seems like you hit the trap. 😉
Two of these 3 amps use MLCC, one is using foil.
THD vs frequency is another story. MLCC (usually) tend to increase at low frequencies, yes.
Two of these 3 amps use MLCC, one is using foil.
THD vs frequency is another story. MLCC (usually) tend to increase at low frequencies, yes.
Last edited:
Hi Christian,
I was trapped like gmarsh. I was sure the Foil cap was the winner on this bench.... Can you do a test with Tantal cap ?
Well, they might be winners in a different amp situation. The unanswered question is, if these foil used is not suitable due to inner construction (stacked vs. wound) and/or geometrical size and location regarding noise pickup.
Regarding differential dc-bias, this is ~zero for the interlink coupling as the AFE output and TPA32XX input are biased at ~6V, so the influence not the biggest concern.
I can't test tantal for several reasons. 1st, the measurements are done in a lab in denmark, i sent the amps to. 2nd, i don't use tantal in any way due to:
https://www.sourceintelligence.com/what-are-conflict-minerals/
Regarding differential dc-bias, this is ~zero for the interlink coupling as the AFE output and TPA32XX input are biased at ~6V, so the influence not the biggest concern.
I can't test tantal for several reasons. 1st, the measurements are done in a lab in denmark, i sent the amps to. 2nd, i don't use tantal in any way due to:
https://www.sourceintelligence.com/what-are-conflict-minerals/
oops... I missed a couple of posts here. Sorry for going back a little.
I can't remember if you already said. Maybe you could recall us the capacitors references you used ? As you say, all caps are not born equal.
TPA325x input bias is AVDD/2 = 7.8/2 = 3.9V
Opamp bias is 6V. The "interlink" caps should see 2.1Vdc bias.
While we talk about dc-blocking caps, no one here seems to use TI's suggested electrolytics (Panasonic EEE-FK1C100R) eventhough they prove to have very good results on the EVM, are cheap and small.
question to the builders : What are your reasons to use other caps ?
ps: Yuan-Jing announced some months ago a TPA3251 board at $38. Status was "coming soon". Today status changed straight to "sold out" 😀
probably impossible to meet the anounced price point !
interesting.doctormord said:
I can't remember if you already said. Maybe you could recall us the capacitors references you used ? As you say, all caps are not born equal.
uhmm.... sure about that ?
TPA325x input bias is AVDD/2 = 7.8/2 = 3.9V
Opamp bias is 6V. The "interlink" caps should see 2.1Vdc bias.
While we talk about dc-blocking caps, no one here seems to use TI's suggested electrolytics (Panasonic EEE-FK1C100R) eventhough they prove to have very good results on the EVM, are cheap and small.
question to the builders : What are your reasons to use other caps ?
have you noticed that in their latest design, the TPA3255 EVM, 3R3 resistor was replaced by a 10µH inductor. Seems that TI is learning as well 😉5th said:
ps: Yuan-Jing announced some months ago a TPA3251 board at $38. Status was "coming soon". Today status changed straight to "sold out" 😀
probably impossible to meet the anounced price point !
Yeah, not completely zero to be correct. (But 2.1Vdc bias for a 50V (quality X7R) device is negligible.
Because electrolytics. (But i can't speak for others)
Will get back to you later.
Data for the MLCC:
Capacitors | details for GRM21BR71E105KA99# | Murata
Have a look at dc-bias and ac voltage rms.
Capacitors | details for GRM21BR71E105KA99# | Murata
Have a look at dc-bias and ac voltage rms.
I wonder why nobody has started to sell the TPA3251 or the TPA3255 yet? I am talking about companies like yuan-jing.com. I think Sinewave has the TPA3251 but I kinda want the TPA3255 🙂
yuan-jing.com withdrew it's bid for unknown reasons. I agree - it's strange. We are many waiting probably....
//
//
I have just ordered Sinewave TPA3251, I don't thin the TPA3255 is out on the market other than samples.
Yes, I emailed Sinewave and they told me they can't get supply for TPA3255 now. Is TPA3251 stock on low supply too? And strange thing is no other company or maker that sell TPA3251 based amplifier.
Correct, they told me they need to find a reliable source for the TPA3255 and they do sell the TPA3251. However, I want (not need) more power. I never know what kind of speaker I will buy in the future.
I agree with you. Very strange that there is NO 3251 makers out there other than Sinewave and their price is little high (at least to me).
Is the cost of TPA3251/3255 chip A LOT more than the 3116?
TI published few days ago an appnote to design ClassD LC output filter.
The TPA3251 EVM is used to measure about 20 different inductors.
Worth reading
http://www.ti.com/lit/an/slaa701/slaa701.pdf
App notes are published out of order 🙂
This one is about selecting the most linear inductor. The best (Wurth) is large and expensive.
A previous recent appnote explained how to add a Post filter feed back (PFFB) on TPA3251 to lower the non-linearities of inductors. One of the acclaimed benefit being that a less performing cheeper inductor could be used with better overall amplifier results.
Has anyone an idea for the optimal compromise for a coil selection with PFFB ?
The TPA3251 EVM is used to measure about 20 different inductors.
Worth reading
http://www.ti.com/lit/an/slaa701/slaa701.pdf
App notes are published out of order 🙂
This one is about selecting the most linear inductor. The best (Wurth) is large and expensive.
A previous recent appnote explained how to add a Post filter feed back (PFFB) on TPA3251 to lower the non-linearities of inductors. One of the acclaimed benefit being that a less performing cheeper inductor could be used with better overall amplifier results.
Has anyone an idea for the optimal compromise for a coil selection with PFFB ?