Has anybody some suggestion for the circuit in first post?
I think it is fairly accurate.
Maybe a few details ......
I think it is fairly accurate.
Maybe a few details ......
Hi lineup
I see you have done the recommendation of TOM in post to and implement it. fine
I am not a chip amp expert but i would prefer the LM3886 not in the TF housing. the LM3886T/NOPB is for me the better choice because of heat "contact" to the heat sink..etc.
LM3886T/NOPB
kr
chris
I see you have done the recommendation of TOM in post to and implement it. fine
I am not a chip amp expert but i would prefer the LM3886 not in the TF housing. the LM3886T/NOPB is for me the better choice because of heat "contact" to the heat sink..etc.
LM3886T/NOPB
kr
chris
Last edited:
The input RC filter with 1k and 480p is around 300khz.
The lower band of AM is in the khz range.
Why such high bandwidth?.
Wouldn't something like 2k and 680p ( 100khz ) be enough ? And help with RF more ?.
Tho I also have way higher freq bandwidth, 220p and 1k on my ohter amp, 700khz ish . And I have no problem wirh RF, weird sounds.
The lower band of AM is in the khz range.
Why such high bandwidth?.
Wouldn't something like 2k and 680p ( 100khz ) be enough ? And help with RF more ?.
Tho I also have way higher freq bandwidth, 220p and 1k on my ohter amp, 700khz ish . And I have no problem wirh RF, weird sounds.
Last edited:
Okay, let me try to make an honest contribution.
I think it's to the point of doing a PCB layout. Make sure the high current carrying areas are wide and short, or even whole pours of copper. Post the layout and we'll comment on it.
It has everything to do with your point.
Oh yeah? What is it that I don't understand about Ohm's Law?
Oh, OK. So you use a term that has an established definition in a way that is incongruent with the established definition only so you can later claim that you were misunderstood and then move the goal post. Got it!
So there are those inverted commas again. What are you describing with the term 'pre-clipping' (in quotation marks – that's what those are called)? Is this another one where you'll move the goal posts after I spend time making my argument? Or are you now using quotes for emphasis?
Don't think so.
Yes. I even know that Zout and DF are linked by Zload, so they really are two ways of looking at the same impedance. Are you aware of that?
Now you're just making stuff up. I suggest you go study how amplifiers work. Maybe design a few of them.
Let me try one more time:
The LM3886 is guaranteed to be able to deliver 7 A of output current.
According to its data sheet, the LM3886 can swing to within 4 V of the power supply rail with a 4 Ω load and within about 2.6 V with 8 Ω load. The 1.4 V difference in output swing explains some of why the LM3886 delivers less than twice the output power into 4 Ω as it does into 8 Ω. The difference is caused by the open loop output impedance of the LM3886 and any current-dependent nonlinearities in the output devices. No surprises there.
What this means is that an LM3886-based amp running on ±30 V will be able to provide 26.0 V (peak) output swing into 4 Ω and 27.4 V (peak) output swing into 8 Ω. That's 85 W and 47 W, respectively, so pretty close to the 65-70 W, 40-45 W obtained in reality. Note, however, that the clipping voltages in the data sheet assume DC operation, whereas the power figures are for operation with THD+N below 0.1 %.
So let's take this LM3886 amp powered by ±30 V, attach it to an 8 Ω nominal speaker that dips to 4 Ω at some frequency. Let's adjust the input signal to the amp such that the amp provides 20 V (peak) into the speaker at a frequency where the speaker's impedance is 8 Ω. I hope that we agree that this is well within the capabilities of the amplifier, yes?
In this scenario the amplifier will provide 20 V (peak) into 8 Ω, which is 20/8 = 2.5 A (peak). Again, well within its capabilities, yes?
P = 20^2/(2*8) = 25 W will be dissipated in the speaker. Before you claim that I don't know that P = E^2/R, keep in mind that E in that equation is the RMS voltage, whereas the 20 V I just used is the peak voltage. RMS = peak/sqrt(2) for a sine wave.
Now, let's change the frequency to the frequency where the speaker has its 4 Ω dip. We leave the amplitude the same. Assuming that the output impedance of the LM3886 is significantly smaller than 4 Ω, this means we still get 20 V (peak) at the output of the amp, correct? According to Ohm's Law, the amp now provides an output current of 20/4 = 5 A. Those are still within the amp's capabilities, right?
P = 20^2/(2*4) = 50 W will be dissipated in the speaker. Again, recall that the 20 V is the peak voltage and not the RMS voltage.
If the amplifier is operated within its current and voltage limits and it has a reasonably low output impedance (thus, high damping factor) it will be able to power a speaker even if the impedance of the speaker varies as function of frequency.
I suppose your next argument will be that if the speaker 'sees' 50 W at the frequency where its impedance is 4 Ω and 25 W at frequencies where its impedance is 8 Ω, it must sound louder at the frequency where its impedance is 4 Ω. That's not the case in reality. What is the case in reality is that the speaker is less efficient at some frequencies than it is at others, so even though its impedance varies and it, thus, dissipates more power at some frequencies than it does at others, all of that is captured by the efficiency and the frequency response of the speaker. After all, the frequency response of the speaker is measured at a constant voltage not at a constant power.
Tom
It doesn't matter greatly. If you soldered the LM3886T to the heat sink, you would get a better thermal connection for sure. But if you use a silicone pad like most of us you'll find the thermal difference between the two to be minimal.
The only time I've seen a difference in performance between the LM3886T and LM3886TF is at heat sink temperatures of 100+ ºC. No person in their right mind would run a heat sink that hot. At more common heat sink temperatures of 50-60 ºC there's no difference in performance between the T and the TF.
It's the same die inside the package.
Tom
To avoid too much attenuation and phase change at 20 kHz.
I don't think I've ever had AM radio creep into circuits. I have had FM (both 88-108 MHz and VHF TV) creep in. Not to mention GSM cell phones. That 200 Hz frame rate of GSM is really annoying.
If attenuation is what you're after, multiple RCs or an LC filter is what you want.
2 kΩ would show up in the noise contributors. And I'm not sure even 2 kΩ, 680 pF provide enough RF attenuation. They might.
Tom
I added 22uf and 100uf and changed the coupling capacitor C1 from 22uf to 4.7uf. The problem still persists.
Is there any chance that the large area of the heatsink at negative potrential creating any interferance to the circuit?
Last edited:
Do the lights in your house also flicker when your domestic heavy load appliances turns ON? I had similar issues , little tightening of terminal screws in the AC mains fuse box sorted this out
Have you observed any differences by shorting the RCA inputs or supply voltage drop across the LM3886 or the power LED blinking?
https://www.circuitbasics.com/design-hi-fi-audio-amplifier-lm3886/
Tom Chr is credited in the supply bypass cap value selection,
also preventing radio interference is also explained neatly
Last edited:
The heat sink should be grounded to the chassis. I have occasionally seen that pop up in THD measurements, but it's not been a very repeatable effect.
If heavy loads switching on/off causes RF interference the cause is more likely to be arcing of the switch. If there's arcing in the fuse box you have different issues. Been there. Done that. 🙂 Some douche forgot to tighten the neutral connection to one of the circuits in my house. Worked great with LED lights but not very well when I plugged in a 1 kW heat gun.
Tom
Won't that make the whole metal chasis negative? LM3886T's bare metal tab without the mica washers -> heatsink -> grounded to chassis
Or am I getting something wrong?
I never connect bare metal tab to heatsink. Always use insulator.
Unless you're making a "double insulated" amplifier you should connect chassis, any part of it, to safety ground. So you need to use an insulator between the metal tab and the heatsink. Of just use a TF package.
I was under impression that T package heat dissipation would be better without those insulators, else I see no benifit in metal package