Hi friends,
I've got my LM3886 stereo system running now for a test and I re-checked the Overture Design Guide spreadsheet after double-checking the supply voltages.
I'm not sure if I should be worried about the supply voltage being +- 38.5V, which leaves 3.5V headroom. Should I be going for a smaller output voltage on the secondaries on the transformer (currently 25V)?
It also seems to think that the heatsink is going to cause me trouble. I've felt the heatsink and it feels lukewarm after 30 minutes of listening. Does this mean the design doc is wrong or I am not outputting 100W ?
I've filled in my values in the spreadsheet and attached the screenshot below.
EDIT: I've got thermal paste on one chip (the other will be done in the next few days), and no insulation washers. I've got the TF package, the one without the metal tab on the body.
I've got my LM3886 stereo system running now for a test and I re-checked the Overture Design Guide spreadsheet after double-checking the supply voltages.
I'm not sure if I should be worried about the supply voltage being +- 38.5V, which leaves 3.5V headroom. Should I be going for a smaller output voltage on the secondaries on the transformer (currently 25V)?
It also seems to think that the heatsink is going to cause me trouble. I've felt the heatsink and it feels lukewarm after 30 minutes of listening. Does this mean the design doc is wrong or I am not outputting 100W ?
I've filled in my values in the spreadsheet and attached the screenshot below.
EDIT: I've got thermal paste on one chip (the other will be done in the next few days), and no insulation washers. I've got the TF package, the one without the metal tab on the body.
Attachments
Last edited:
Sorry to burst your bubble, but you won't get 100 W out from a single LM3886. While that Overture Design spreadsheet is good, it does not take the SPiKe protection circuit into account. You get the maximum amount of power from an LM3886 into a 4 Ω load with a ±32 V power supply. Unless you are using a regulated supply, I recommend designing for ±30 V, which means you'll need a 2x22 VAC power transformer.
It sounds like a good thing that the heat sink is only lukewarm after 30 minutes of listening (at reasonably loud volume I assume). I don't see an issue there.
Tom
It sounds like a good thing that the heat sink is only lukewarm after 30 minutes of listening (at reasonably loud volume I assume). I don't see an issue there.
Tom
No worries Tomchr, you're not bursting my bubble. Didn't think it was 100W, hence I doubted that the spreadsheet was accurate.
In terms out loudness, if I put max volume on the PC line out, then it gets pretty loud at 50% volume on the amp, such that I don't really want to push it more for risk of annoying my neighbours (apartment complex). So I'm not really disappointed with the output power at this stage (may want more power if I buy a house).
The heatsinks seem fine, lukewarm, but wasn't at very loud volume. Will do more testing. Hence again doubting what the spreadsheet was warning me about the heatsinks.
Regarding the transformer, I've got 25V dual secondaries, but the output voltage is +-38.5V, which is a * 1.54 scalar. Not sure if my mains is just high. Presumably dual 22V secondaries would be +-33.88V ?
In terms out loudness, if I put max volume on the PC line out, then it gets pretty loud at 50% volume on the amp, such that I don't really want to push it more for risk of annoying my neighbours (apartment complex). So I'm not really disappointed with the output power at this stage (may want more power if I buy a house).
The heatsinks seem fine, lukewarm, but wasn't at very loud volume. Will do more testing. Hence again doubting what the spreadsheet was warning me about the heatsinks.
Regarding the transformer, I've got 25V dual secondaries, but the output voltage is +-38.5V, which is a * 1.54 scalar. Not sure if my mains is just high. Presumably dual 22V secondaries would be +-33.88V ?
Relatively normal for a unregulated power supply.
Voltage will be higher with no signal or idle because there is little current being used.
Once more current is being drawn the voltage will start to drop.
You might have 38 volts at idle.
But once you are using more current close to the rating of the transformer
voltage will sag down to the more assumed 35 volts.
If you ever get to that point. really depends on the speakers.
Most speakers are rated thermal, which is usually much much higher than when they reach actual distortion.
Of course it varies depending on the enclosure and actual mechanical ability's of the speaker.
In a generalized manner a " 60 watt Speaker" might actually reach 10% distortion
with only 15 to 20 watts.
And actual normal listening levels before the neighbors get mad might only be 8 to 12 watts.
Voltage will be higher with no signal or idle because there is little current being used.
Once more current is being drawn the voltage will start to drop.
You might have 38 volts at idle.
But once you are using more current close to the rating of the transformer
voltage will sag down to the more assumed 35 volts.
If you ever get to that point. really depends on the speakers.
Most speakers are rated thermal, which is usually much much higher than when they reach actual distortion.
Of course it varies depending on the enclosure and actual mechanical ability's of the speaker.
In a generalized manner a " 60 watt Speaker" might actually reach 10% distortion
with only 15 to 20 watts.
And actual normal listening levels before the neighbors get mad might only be 8 to 12 watts.
50% volume? If you are speaking about the volume potentiometer, that does not say anything about the actual output. Do not rely on the position of the knob, what matters is the input voltage and the gain. Typically amps reach max power at 2/3 of the volume pot scale, the rest is reserve for really low input signals. If you really want to know how much of the maximum power you are using, look at the specs of the amp schematics (gain) and how much your source puts out. The latter is quite hard to get because most mainboards and cheaper sound cards do not give such informations. You can still measure the input and calculate the power with the gain. It will still be an estimate unless you know the impedance response of your speakers, but far closer than the volume dial.
The idle voltage of a transformator is usually higher than with a load, the voltage sags to the nominal voltage with current drawn. That is completely normal. After the rectifier and the caps the voltage is 1,41x higher ( SQR(2) ) than the secondary transformer voltage. The LM3886 can work with a maximum of 84V or +/-42V DC according to the datasheet.
That means, with the +-38.5V you are close to the maximum voltage but the power grid got fluctuations. Usually it is a maximum of not more than 10% but that depends on where you live, very remote locations may have a higher diversion from that, power generators too. You don't exceed the 42V unless the grid goes over 16,17% of the nominal voltage, which should be enough. And the LM3886 will not be destroyed if the voltage gets higher, it will likely shut down, damage may occur at ~20% over voltage (roughly over probably +/-50V).
The danger is pretty low because that's still the voltage of the IDLE transformator. You can put a 2nd load on the trafo (bulb, resistor etc) which pulls the voltage down. Or just use the amp and look how much the voltage goes down.
However, the amp will likely get pretty hot on high power use, a biger heatsink or a 'silent' fan can handle that. I like to use the latter, the silent fans are pretty cheap and you can adjust the speed of them with the voltage (except for the PWM ones).
If the heat sink is just getting lukewarm at neighbor-annoying volume levels you don’t have anything to worry about. That maximum power dissipation from the spreadsheet will never be sustained with music - and to get it that high you literally need to have it turned up so loud and garbled that you can’t even tell what song is playing anymore. And when you do that, the power supply probably *will* drop to +/-30 volts. With real world music turned up loud but not distorting the average power being sent to the speaker is a handful of watts (single digits) and power dissipation in the amp about 2x that. If your heatsink can handle say 15 watts of thermal dump indefinitely without burning your hand on it. then you’re probably good to go. If this were a guitar amplifier you’d want a bigger heat sink or a fan.
If you design the amp to deliver a sine wave at the worst case operating point for dissipated power, you'll end up with a very conservative (and expensive) design both for the heat sink and the power supply. If you account for the crest factor of music, you can dial back the design quite a bit. I go into great detail here: Taming the LM3886 Chip Amplifier: Thermal Design – Neurochrome
Tom
Tom