Simple Solution to Heat
Simple Solution to Heat
Running at 25v to 35v SMPS power supply, off 4 LIFEPo4 cells in series (about 14v battery).
Like other folks, i noticed a very hot amp chip. Also a very hot power supply. Also got a lot of dropouts on loud passages with a high-level input.
Then, i lowered the power supply to about 16v. Now, the amp and power supply both run cool, with same volume and same SQ (on casual listening).
It makes sense the SMPS would run cooler, since it doesn't have to work as hard to convert 14v to 16vand converting 14v to 30v.
Not sure why the amp runs cooler too.
i also noticed i'm getting different noise-levels from the amp when it's input is disconnected.
I get a lot of noise running off this XL6019e1 SMPS
https://www.ebay.com/itm/253610915818
Getting little to no audible noise with this LM2587S-based SMPS.
https://www.ebay.com/itm/183350130732
Simple Solution to Heat
Running at 25v to 35v SMPS power supply, off 4 LIFEPo4 cells in series (about 14v battery).
Like other folks, i noticed a very hot amp chip. Also a very hot power supply. Also got a lot of dropouts on loud passages with a high-level input.
Then, i lowered the power supply to about 16v. Now, the amp and power supply both run cool, with same volume and same SQ (on casual listening).
It makes sense the SMPS would run cooler, since it doesn't have to work as hard to convert 14v to 16vand converting 14v to 30v.
Not sure why the amp runs cooler too.
i also noticed i'm getting different noise-levels from the amp when it's input is disconnected.
I get a lot of noise running off this XL6019e1 SMPS
https://www.ebay.com/itm/253610915818
Getting little to no audible noise with this LM2587S-based SMPS.
https://www.ebay.com/itm/183350130732
Your output filter choke core losses are lowered considerably with reduced magnetic cycling. The power for the core losses come from the power supply handling higher carrier currents.
you mean the toroid inductor on the SMPS?
is this correct? As the output V of the SMPS is increased, that pulls more current from the battery. As current into the choke maxes out, the core saturates with magnetic flux, and cannot store any additional energy. Thus, additional energy coming from the source (battery) gets wasted as heat.
But, shouldn't the SMPS be able to put out any V within it's current-rating, without saturating?
Why would this cause the amplifier to run cooler?
Thx
My reply was to the question why the class D amplifier heats less with reduced supply voltage. This is because the core losses increase rapidly as you make bigger B-H cycles in the magnetic hysteresis loop:
Since a coercive force must be applied to overcome this residual magnetism, work must be done in closing the hysteresis loop with the energy being used being dissipated as heat in the magnetic material. This heat is known as hysteresis loss, the amount of loss depends on the material’s value of coercive force.
Source: https://www.electronics-tutorials.ws/electromagnetism/magnetic-hysteresis.html
When the load current and voltage to the load (the amplifier) increase, the SMPS has to handle more power and SMPS losses increase as well. Also the SMPS choke has an increase in magnetic losses with higher current ripple. Even without entering saturation.
Yes, in a well designed SMPS you do normally not enter magnetic saturation within nominal load current. Actually, there are MagAmp designs where magnetic saturation is intentionally used for regulation and self-oscillating designs where saturation is intentionally employed.
However today, with an enormous price competition on SMPS, many (cheaper) manufacturers use magnetic components operated close to the limit such that the magnetic margins are less and core losses higher.
How would that exhibit itself? I notice no change in perceived volume, bass, or transient response when i reduce the supply from 35v to 15v.
using an 8 ohm.
According to the datasheet, overcurrent doesn't cause shutoff:
Thermal protection doesn't cause shutoff either:
The cases where shutdown occurs is:
- short-circuit across the speaker
- over-voltage
Last edited:
How would that exhibit itself? I notice no change in perceived volume, bass, or transient response when i reduce the supply from 35v to 15v.
What is at the output of an amplifier depends on what is at the input and the transfer function of the amplifier (mainly the gain). The supply voltage has little influence on the sound until the supply voltage is not sufficient for the amplifier to generate an undistorted sine-wave and you get "clipping" (distortion).
using an 8 ohm.
Fine, then it is hardly current limitation.
According to the datasheet, overcurrent doesn't cause shutoff:
Thermal protection doesn't cause shutoff either:
The cases where shutdown occurs is:
You have done your homework in the datasheet. You do not mention "shut-off" in your initial mail, you mention "dropouts". As it is you hearing the effect, your guess is better than mine.
What is at the output of an amplifier depends on what is at the input and the transfer function of the amplifier (mainly the gain). The supply voltage has little influence on the sound until the supply voltage is not sufficient for the amplifier to generate an undistorted sine-wave and you get "clipping" (distortion).
using an 8 ohm.
Fine, then it is hardly current limitation.
According to the datasheet, overcurrent doesn't cause shutoff:
Thermal protection doesn't cause shutoff either:
The cases where shutdown occurs is:
- short-circuit across the speaker
- over-voltage
You have done your homework in the datasheet. You do not mention "shut-off" in your initial mail, you mention "dropouts". As it is you hearing the effect, your guess is better than mine.
Nothing to hear. There's a dead-silent gap in the audio, which lasts for a fraction of a second, on music peaks.
The datasheet describes overvoltage and shortcircuit cause the amp the shut-off. i have not read about any other situation with this chip that would cause gaps in the audio-stream.
Since this happens only with a supply V very close to the absolute max V, overvoltage seems the likely culprit. The datasheet mentions that the TDA8932 can cause the power-supply V to go up.
Sounds like they're saying: the battery supplying the SMPS cannot deliver the current demanded by the load (TDA8932), and that causes the SMPS out V to go up. Makes sense.
Weird thing is, they say the V increase on the output of the SMPS can be caused BY overvoltage protection. I don't get that. The reverse seems more logical: that a too-high V on the output of the SMPS would trigger the over-v protection on the amp.
They say this happens in SE mode, but i'm in BTL mode. i'm guessing it's basically the same thing. Strange the datasheet doesn't mention this issue for BTL.
Terminology: Cakewalk audio notes that "drop-out" is sometimes used mistakenly to describe other behaviors.
Cakewalk - Knowledge Base - Audio Dropouts, Clicks and Pops When Playing and Recording
Ableton uses the term "dropout" to refer to silent gaps in the audio. So that seems like the right term to describe what i'm observing.
How to avoid crackles and audio dropouts – Ableton
cheers!
Last edited:
but datasheet says turn-offs happen during over-voltage, not during over-temperature.
Are they wrong?
Why less likely in BTL?
thx
I guess it depends on how much the gain is decreased - a large reduction could be indistinguishable from a mute, particularly for only a few hundred mS.
In BTL mode the current being returned to the power supply by one side of the bridge is 'eaten up' by the opposite side as its working in anti-phase. The result is the caps don't get to eat the current and hence there's no voltage boost.
In BTL mode the current being returned to the power supply by one side of the bridge is 'eaten up' by the opposite side as its working in anti-phase. The result is the caps don't get to eat the current and hence there's no voltage boost.
true, but datasheet says gain is reduced, and restored, "gradually". I'm hearing complete and instant switch to dead silence, and then instant switch to full volume. This seems much more consistent with completely shutting-down than "gradually reducing gain." Tho "gradual" can be relative-- they don't say what "gradual" means.
Going overvoltage when close to max V seems plausible. Tho they say:
So could be that. That would require the chip cools sufficiently to restart in the blink of an eye; ie, hovering right on the edge of thermal threshold. Could be that.
Thx, that's interesting. Are you saying that an infinite amount of current can be transferred from one side of the bridge to the other?
Last edited:
"infinite current" is beyond my experience except as one viewing the movie 'Forbidden Planet' where the ancient civilization called 'The Krell' had ammeters stretching out of sight, each one indicating 10X the current of its neighbour....
Fortunately we don't need to transfer infinite current, just that recirculating in the bridge. So long as the current eaten by one side is matched by that pushed out by the other, all is dandy and the supply voltage won't budge. At least, that's the theory.
I run TDA8932s at their maximum voltage (well not quite, about 35V) and in BTL and have yet to experience any 'holes' in the audio. The chips are too hot to touch comfortably for any length of time. The reason I suspect I don't get holes is my load impedance is about 32ohm by virtue of using a transformer between amp and speaker. Such a load means my peak power is way down on a typical set-up with an 8R load.
Fortunately we don't need to transfer infinite current, just that recirculating in the bridge. So long as the current eaten by one side is matched by that pushed out by the other, all is dandy and the supply voltage won't budge. At least, that's the theory.
I run TDA8932s at their maximum voltage (well not quite, about 35V) and in BTL and have yet to experience any 'holes' in the audio. The chips are too hot to touch comfortably for any length of time. The reason I suspect I don't get holes is my load impedance is about 32ohm by virtue of using a transformer between amp and speaker. Such a load means my peak power is way down on a typical set-up with an 8R load.
Theoretically, then, if one side is pulling too much current, then it will overload the other side too, no?
Very interesting! So where does the extra current go, if not to the other side of the bridge?
How does your config behave/sound at 16 v supply?
I assume you're not running off batteries. Would you expect longer or shorter battery life with the xfmrs?
thx
Last edited:
I'm unclear what 'pulling too much current means' here? Current overload will necessarily apply to both sides equally but can't see how it happens when driving a speaker.
Can't quite see what 'extra current' is applying too here. Perhaps I'm being exceptionally dense though, always possible
Not tried running all rails at 16V but did try out just the VDDA one at 16V, seemed not to change the SQ much. But it did introduce an offset which I didn't like the look of so I reverted to full rails.
Compared to running direct into speakers at lower supplies? Running higher voltage is less efficient (due to extra static dissipation) - adding the trafos reduces efficiency still further (by adding some series impedance).
Won't output trannies fix that? Sorry if already answered, I'm slowly reading through this awesome 60 page thread
I'm speaking from ignorance. I'm trying to understand the circulating current you said is pushed out by one side.
What's "all rails"? What's "the VDDA one"? Where'd you see offset?
lower supplies? I meant same supply.
Last edited:
They sure do
TDA8932 has more than one pair of power supply pins. The output stages connect to one and the signal processing stages connect to another. The VDDA pin is the latter. Offset at the outputs (both sides) but not across the speaker.
Then trafos definitely reduce efficiency. But efficiency is still plenty good enough in comparison with classAB so I feel I can afford to lose a little.