Hi, I've built a power amp using LM3886. The sound is a bit too up front and less bass and power. I'm following the original circuit from National Semiconductor website. I would like to know is there any improvement could be made? One more thing, I would like to know is there any benefit if I parallel tow LM3886 together? Will it be sounding more control, more powerful and more bass? By the way, I'm feeding the LM3886 with +28vdc and -28vdc, will it be enough for 2 LM3886? I hope you guy can help me, thank you.
Try pushing the power input to +- 35V DC. Bigger caps, say 6,800uF for 2 channels. You can also try to parallel two LM3886s (ie PA100 configuration), which will push around 100W into 4 ohms. The power limit of around 50W into 8 ohms will remain, limited by power supply voltages. For more power into 8/4 ohms, try the BPA configuration. What is your speaker impedance?
It's about thermals on your voltages rails and speaker impedance. For 4 ohm +/-28V is what National recommends. For 8 ohm it is +/-35V. You can run a little higher if running music since it is not as demanding as a sine wave. If you get a big heat sink and keep it cool enough you can run even higher. One amp I built have bridged channels using LM3886. They were fine with music but would over heat with sine waves after a few minutes. Probably not the best design but it was another fun project so not worried.
Caps will affect sound quality. Be sure the input cap is large enough in value to get good bass as well as the feedback cap (what National usually calls Ci) is also large enough. You can also parallel nice poly caps with these to help improve sound. Having enough rail capacitance will also help the bass, driving 4 ohms needs more than 8. Other ideas spread around these forums but those are the basic. If you have no DC from the source then you can remove both caps for best sound quality. Just be sure you know you don't have DC or it will get amplified to the load by the amount of gain you have, if Ci is removed. If Ci is there only unity DC gain.
-SL
Caps will affect sound quality. Be sure the input cap is large enough in value to get good bass as well as the feedback cap (what National usually calls Ci) is also large enough. You can also parallel nice poly caps with these to help improve sound. Having enough rail capacitance will also help the bass, driving 4 ohms needs more than 8. Other ideas spread around these forums but those are the basic. If you have no DC from the source then you can remove both caps for best sound quality. Just be sure you know you don't have DC or it will get amplified to the load by the amount of gain you have, if Ci is removed. If Ci is there only unity DC gain.
-SL
Today I've found another transformer with 25v-0v-25v AC and +-38vdc. I've connected it to my LM3886 power amp and the result is the bass is tight and lower bass is more crontrol, but the same problem is still there, upper bass is still less than what I've expected. And 1 more thing is the LM3886 is quite hot, without any input signal, the heatsink temperature is around 43c, 5c more than the previous transformer. Is it normal? I also have changed the Cl cap from 22uF to 33uF. I think I'll add more e-cap and see what is the result.
A scope would be very handy. Having the heat sink at 43C means nothing without knowing the ambient temperature and the heat sink thermal resistance (C/W). The LM3886 will burn 4 - 5W with no input depending on supply voltages. So if you know the thermal resistance of the heat sink, ambient room temperature and temperature of the heat sink then you can figure if it is in the right range of power dissipation. If it is higher then oscillation, as already noted, if the most likely culprit.
-SL
-SL
Hi,
what is the range of Iq from the spec sheet?
What are your input and NFB filter turn over frequencies?
The PSU caps (22mF in total), what is on each supply rail and are the rails supplying one or two or more chipamps?
Keep in mind that 4ohm speakers require twice the current compared to 8ohm speakers (and twice the smoothing capacitance) and at these higher currents chipamps start to struggle unless specifically built to get the best out of them.
what is the range of Iq from the spec sheet?
may understate the dissipation requirement.
What are your input and NFB filter turn over frequencies?
The PSU caps (22mF in total), what is on each supply rail and are the rails supplying one or two or more chipamps?
Keep in mind that 4ohm speakers require twice the current compared to 8ohm speakers (and twice the smoothing capacitance) and at these higher currents chipamps start to struggle unless specifically built to get the best out of them.
A LM3886 will have around 50mA per rail. There is not a lot of varation in supply current with supply voltage so power dissipation varies with supply voltage. So for +/-35V rails and 50mA per rail you get 3.5W of power dissipation. Remove the caps and increase DC offset increasing current a little more and a tad higher. If oscillatin then the supply current should be noticeably higher and hotter LM3886 results.
-T
-T
Today I've borrowed a scope from my friend and since I'm new to this kind of gadget, I don't know how to rectify the oscillation with using the scope. Can you please give me some example? Since the LM3886 is built-in thermal protection and speaker short protection, will it still dangerous to drive 4ohm speaker with around +-38vdc? When the LM3886 reach the heat level it can handle, it will turn off automatically, so if using 4ohm speaker and high vdc, it will not spoil the LM3886, just will turn on the thermal protection faster only, right?
Hi,
use the scope to examine the output of the amp.
The sine wave or square wave should have no oscillation on any part of the signal.
You will need a dummy load that can take a fair bit of continuous power.
Try to work quickly through a series of input voltages from about 1/100 max power (1/10 max voltage) upto about 110% of max input voltage. Once above 3/10 voltage you need to be quick because the amp and the load start getting hot very quickly. A switched attenuator is much better here than turning a pot. That way you can momentarily switch up to say 50% voltage, read the screen and switch straight back down to 5% voltage, reset the signal voltage and repeat.
Try a different frequency and run through the voltage sequences again. 1kHz, 5kHz, 10kHz, 20kHz, both sine and square. 20kHz may make the Zobel smoke! Bad for it's lungs, give it a good telling off.
Then try adding a small cap to your dummy load (100nF //8r) and do the voltage runs and frequency runs again.
Then step through a range of caps 220nF, 470nF, 1uF, 1u5F and finally 2uF. repeating the voltage and frequency runs for each cap resistor combination.
If you amp is good for 4ohm loading then you should substitute 4r for 8r as your load.
Finally, check the output at 80% to 90% of maximum input into half load resistance. ie. into 4r for 8ohm amp and into 2r for a 4ohm amp.
This is a long process, to do and even longer waiting for it to cool down. Check temperatures regularly. Doing high power tests with a sink that is already hot might (will) destroy the output stage.
The elimination of any oscillation will come later. Give the experts a call for advice after you have identified the problem (frequency, waveform, signal ampltude, load condition).
Good luck.
use the scope to examine the output of the amp.
The sine wave or square wave should have no oscillation on any part of the signal.
You will need a dummy load that can take a fair bit of continuous power.
Try to work quickly through a series of input voltages from about 1/100 max power (1/10 max voltage) upto about 110% of max input voltage. Once above 3/10 voltage you need to be quick because the amp and the load start getting hot very quickly. A switched attenuator is much better here than turning a pot. That way you can momentarily switch up to say 50% voltage, read the screen and switch straight back down to 5% voltage, reset the signal voltage and repeat.
Try a different frequency and run through the voltage sequences again. 1kHz, 5kHz, 10kHz, 20kHz, both sine and square. 20kHz may make the Zobel smoke! Bad for it's lungs, give it a good telling off.
Then try adding a small cap to your dummy load (100nF //8r) and do the voltage runs and frequency runs again.
Then step through a range of caps 220nF, 470nF, 1uF, 1u5F and finally 2uF. repeating the voltage and frequency runs for each cap resistor combination.
If you amp is good for 4ohm loading then you should substitute 4r for 8r as your load.
Finally, check the output at 80% to 90% of maximum input into half load resistance. ie. into 4r for 8ohm amp and into 2r for a 4ohm amp.
This is a long process, to do and even longer waiting for it to cool down. Check temperatures regularly. Doing high power tests with a sink that is already hot might (will) destroy the output stage.
The elimination of any oscillation will come later. Give the experts a call for advice after you have identified the problem (frequency, waveform, signal ampltude, load condition).
Good luck.
I've chcked everything, everything seem in perfect condition. The scope showed nothing wrong with the power amp. Today I've use the amplifier to drive a pair of 8ohm speakers for 2 hours with very high level, as my surprise, the temperature reading was 65C and stop there, can't go up anymore. So do you still think that my power amp has problem or the problem not because of the oscillation?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.