D
Deleted member 148505
One thing that affects the THD+N the most (assuming that the layout and output inductors are good) is the higher noise floor brought about by the post switching regulators. When I used LM317 as the regulator I can only get 0.004% THD+N sound card measurement at 8 ohms 10W on my non PFFB board.
I saw Dr Mordor used LM7812/LM7912 on his older boards which I think will get poor noise floor results (Higher THD+N). As a tip, I think a minimum of LM2940 + LM2990 post regulators will get Dr. Mordor's boards THD+N on par with the EVM.
Split supply rail for opamps is good idea as you will eliminate the 4 input capacitors.
I saw Dr Mordor used LM7812/LM7912 on his older boards which I think will get poor noise floor results (Higher THD+N). As a tip, I think a minimum of LM2940 + LM2990 post regulators will get Dr. Mordor's boards THD+N on par with the EVM.
Split supply rail for opamps is good idea as you will eliminate the 4 input capacitors.
Last edited by a moderator:
Engineers are such an odd breed, demanding to see each other's "stuff" and insisting that their own measurements are smaller!
I have used a LT3045 (best LDOs as far as noise) on critical rail and almost as good (LDOs) for rest of rails. Each rail has a dedicated LDO in my design.
Thats one element on why we are able to achieve our measurements
Also , based on test data, split rail supply has a distinctive advantage in terms of noise. My first proto was based on single rail (opamps) but since then I moved on to dual rails.
Note that my front end is nothing like TI datasheet.. I am using buffers and FDAs..I chose my buffers for EMI rejections ..
This amp was available yesterday and almost pulled the trigger. Anybody have experience with this one? I am looking for a complete unit.
https://www.amazon.com/dp/B08GQL6BJ...olid=1DLMUTCMBP11F&psc=0&ref_=lv_ov_lig_dp_it
https://www.amazon.com/dp/B08GQL6BJ...olid=1DLMUTCMBP11F&psc=0&ref_=lv_ov_lig_dp_it
This certainly does not comply with any safety standards.
Is there a nameplate on it?
Is there a CE marking?
With a metal case housing, protection classII is mandatory.
Adequate double insulation of inside wiring and other pre-cautions.
This includes a batch on it with the protection class II symbol.
Without out that batch, automatically protection class I must be assumed.
That requires definitely protection earth grounding.
Which is non-existent.
Any questions else?
Is there a nameplate on it?
Is there a CE marking?
With a metal case housing, protection classII is mandatory.
Adequate double insulation of inside wiring and other pre-cautions.
This includes a batch on it with the protection class II symbol.
Without out that batch, automatically protection class I must be assumed.
That requires definitely protection earth grounding.
Which is non-existent.
Any questions else?
Last edited:
Completely agree... also, no PE connection when there are obviously Y caps on the PSU board means un-controlled EMI levels
and -in any case- another deviation from proper implementation conditions of this PSU which is genuinely designed to operate with PE connected to it.
Kal.
Why don't you create a new thread in the Class D subforum for your project? It will make for a better experience for everyone here, I think (in a good way - don't take this as any sort of insult).
cdsgames,
Gain of 10dB for TPA3255 on your design means 3.16x of voltage gain,
your maximum output power at 8 ohms will only be 30.6W
while 30W @8ohms output is perfectly adequate for music listening, this is
quite a bit lower than the original TPA3255 spec of 195W in BTL mode.
It is not surprising that you are getting excellent distortion figures at low gain.
The TI PFFB scheme used 5.57dB as feedback from the 21.5dB spec loop gain
and achieved approx 5dB of distortion reduction, at the expense of 5dB lower
voltage gain. Since you lower the gain to 10dB with 11.5dB feedback,
I would expect you to have lowered the distortion by at least 11.5dB as well.
This is a common design trade off between gain and amount of feedback to
apply, you pick the operating point within the available open loop gain:
higher gain/lower FB (less error correction) or
lower gain/higher FB (more error correction),
it all depends on what you want to optimize.
In addition, using low noise components and good layout practices will
lower the noise floor and add more to the improvement too. I see you are
doing that already.
Ok I will create it and update there.
As a last tip..
The cap on Avdd , I have used multiple caps and by checking with scope I have taken the noise data. I have found that 220uF is not enough and more (around 1000uF ) capacitance does decrease the noise floor of Avdd.
This info can be used for any design/mod of TPA3255.
Sacar , I have designed the AMP with a front end that has selectable gain . So you can have gain of 10db (min) , 13db , 17db and 20db (total)
Further I have designed my LLC with selectable voltages so you can remove (or add 2-3 db to all of the above)
As a last tip..
The cap on Avdd , I have used multiple caps and by checking with scope I have taken the noise data. I have found that 220uF is not enough and more (around 1000uF ) capacitance does decrease the noise floor of Avdd.
This info can be used for any design/mod of TPA3255.
Sacar , I have designed the AMP with a front end that has selectable gain . So you can have gain of 10db (min) , 13db , 17db and 20db (total)
Further I have designed my LLC with selectable voltages so you can remove (or add 2-3 db to all of the above)
Last edited:
