It means that the hum to your speakers is at a level that is -70dB below the supply ripple level. So -70dBv below 200mVp-p is ? 63μVp-p
In line with
Vn Output integrated noise 20 Hz to 22 kHz, A-weighted filter, Gain = 20 dB
–80 dBV or 65 μV
The chip sees a 8 ohm load per channel, not your power supply.
If you go with above #'s, 5 V / 85mA = 58.8 ohm
TPA3118 Icc Quiescent supply current mA
SDZ = 2 V, No load or filter, PVCC = 24 V = 32-50 mA
In line with
Vn Output integrated noise 20 Hz to 22 kHz, A-weighted filter, Gain = 20 dB
–80 dBV or 65 μV
The chip sees a 8 ohm load per channel, not your power supply.
If you go with above #'s, 5 V / 85mA = 58.8 ohm
TPA3118 Icc Quiescent supply current mA
SDZ = 2 V, No load or filter, PVCC = 24 V = 32-50 mA
Too much for this brain.
What I really need to know is; as I design a linear supply for this board, how much supply ripple is too much?
Example: In general terms, if the supply voltage swings from 14.5v to 15.5v (a full volt of ripple) what will that mean with respect to the TPA3116/8 performance?
Your ears are the final test.
If TI tests with 200mV p-p then I'd say that is a good target.
You can play with this too
Online Calculator .:. Linear Power Supply Designer
Member
Joined 2006
Member
Joined 2006
The internet is full of poorly educated opinions/statements...
I can't find any specific literature on the measurements of lead acid batteries, but assuming they're comparable to alkaline/lithium/mercury batteries (and I can't think of any reason they'd be different) they're pretty damn quiet. Check out figures 2/3 on page 3 on this paper:
http://tf.nist.gov/timefreq/general/pdf/1133.pdf
Note the LM317 waaaay up at the top of figure 2, 60dB-80dB worse than the batteries. Think a LT "low noise" regulator is 60-80dB better?
I can't find any specific literature on the measurements of lead acid batteries, but assuming they're comparable to alkaline/lithium/mercury batteries (and I can't think of any reason they'd be different) they're pretty damn quiet. Check out figures 2/3 on page 3 on this paper:
http://tf.nist.gov/timefreq/general/pdf/1133.pdf
Note the LM317 waaaay up at the top of figure 2, 60dB-80dB worse than the batteries. Think a LT "low noise" regulator is 60-80dB better?
Member
Joined 2006
Simple Voltage Regulators Part 1: Noise
Perhaps this one...?
http://www.diyaudio.com/forums/showthread.php?s=&postid=187556&highlight=#post187556
I went through that battery, SMPS, linear ps in the previous T amps round but with no clear conclusion .
Will try yours this time and also with Jeff's suggestion
Perhaps this one...?
http://www.diyaudio.com/forums/showthread.php?s=&postid=187556&highlight=#post187556
I went through that battery, SMPS, linear ps in the previous T amps round but with no clear conclusion .
Will try yours this time and also with Jeff's suggestion
Last edited:
As an example of the minimal power supply requirements for these class "D" amps,
I am using a class 2 AC/DC adapter, 12.5V/800mA/7W to power my portable radio using a TPA3100D2.
The DSO measures 260mV p-p of ripple, with no music,
At a good (near uncomfortable) volume using 4ohm Dynaudio Gemini speakers, 15' 20AWG the DSO measures peaks that maximize at around 1V.
With that amount of gain and eff speakers, if I pause the source, I only hear with my ear right beside the tweeter, the faintest bit of noise, but no hum at all.
I can actual detect the amp clipping a bit, that is why a higher voltage ~24V is the best thing to use.
I have a 4700uF/35V bulk ecap & 2 of 220uF for local decoup on-board. The power-pack lucky to have 1000uF.
So that is it, you really do not need a linear voltage regulator, using one, actually defeats the purpose of using a high efficiency class "D" amp in the first place.
The lin reg will dissipate more power than the amp will, which really does not make too much sense since it really buys you nothing other than heat, expense. It also limits dynamic range.
Enjoy your TI class "D" amps everyone
I am using a class 2 AC/DC adapter, 12.5V/800mA/7W to power my portable radio using a TPA3100D2.
The DSO measures 260mV p-p of ripple, with no music,
At a good (near uncomfortable) volume using 4ohm Dynaudio Gemini speakers, 15' 20AWG the DSO measures peaks that maximize at around 1V.
With that amount of gain and eff speakers, if I pause the source, I only hear with my ear right beside the tweeter, the faintest bit of noise, but no hum at all.
I can actual detect the amp clipping a bit, that is why a higher voltage ~24V is the best thing to use.
I have a 4700uF/35V bulk ecap & 2 of 220uF for local decoup on-board. The power-pack lucky to have 1000uF.
So that is it, you really do not need a linear voltage regulator, using one, actually defeats the purpose of using a high efficiency class "D" amp in the first place.
The lin reg will dissipate more power than the amp will, which really does not make too much sense since it really buys you nothing other than heat, expense. It also limits dynamic range.
Enjoy your TI class "D" amps everyone
Last edited:
I'm going to use this (I found all the parts in a bin I picked up a couple years ago):
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Anyone know why LTspice does not factor in the voltage increase due to rectification?
In the above circuit, the source is 13.7vac (the real measured output of my unloaded 12v transformer). LTspice simulates the circuit pushing out a nice, steady 26.x vdc.
A few days ago I was reading about rectification and something about increased voltage factor of sq. rt. 2 (1.414). I ignored this, because the LTspice sim ignored it.
After work today, I actually built the circuit. To my surprise, the circuit pushed 37.x vdc.
So remembering that 1.414 factor, I got out the calculator and, of course:
13.7v * 1.414 = 19.3718
19.3718 * 2 = 38.7436
Subtract diode losses, and you get 37.x vdc.
So why the heck doesn't LTspice know this?
In the above circuit, the source is 13.7vac (the real measured output of my unloaded 12v transformer). LTspice simulates the circuit pushing out a nice, steady 26.x vdc.
A few days ago I was reading about rectification and something about increased voltage factor of sq. rt. 2 (1.414). I ignored this, because the LTspice sim ignored it.
After work today, I actually built the circuit. To my surprise, the circuit pushed 37.x vdc.
So remembering that 1.414 factor, I got out the calculator and, of course:
13.7v * 1.414 = 19.3718
19.3718 * 2 = 38.7436
Subtract diode losses, and you get 37.x vdc.
So why the heck doesn't LTspice know this?
Oh, and one very cool "side effect" of all of this that I noticed voltage taken between c2 and c3 is (of course) exactly half of the total circuit voltage - around 18.8vdc perfectly flat.
While it seems like a bridge rectifier and a smoothing cap would be a simpler solution to get ~18v, that solution results in big ripple unless the caps are really big and really low ESR.
The circuit above provides ~18v with zero ripple, and uses caps with an ESR of .235 (but makes the transformer work harder).
While it seems like a bridge rectifier and a smoothing cap would be a simpler solution to get ~18v, that solution results in big ripple unless the caps are really big and really low ESR.
The circuit above provides ~18v with zero ripple, and uses caps with an ESR of .235 (but makes the transformer work harder).
Last edited:
Interesting.
LTspice assumes that people know this. I suppose my lack of knowledge is the issue, not LTspice.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.