I recently built the Chipamp.com LM1875 kit and I'm still messing about with input caps and filter caps. I've read where lowering the value of the input cap simply moves the corner frequency up. Is that correct? And if so, how low can I reasonably go in cap value? The stock cap size is 2.2uF and, because of the small boards and consequent narrow trace spacing, I'm really limited on available caps. If I go down to, say, 1uF, I gain a lot more tweaking freedom as far as caps. My speakers don't do deep bass regardless (I estimate they go down to maybe 60-70 Hz in my room).
Hi the required capacitor value depends on input impedance we really need to see your circuit. It also determines where the bass rolls off so how much do you want? If I apply my rule of thumb 1/4pi*cr your value dictates no less than 36K ohms but this is no absolute it will work perfectly with 10K just starting to lose bass at higher bass frequencies....
The lower -3dB frequency is 1/(2*3.14*R*C). So 2.2uF gives you 3.3 Hz, which is more than enough.
if the speaker is down to -3dB @ 30Hz and you drive it with an amplifier that is also down by 3dB at the same frequency, then what you hear is 30Hz at -6dB relative to the rest of the passband.
When a 30Hz tone is played alone you probably won't be able to recognise that the 30Hz tone has been attenuated.
If however you send a mixture of 30Hz and many other tones and harmonics to the speaker you can easily hear that the 30Hz is not in the correct proportion to the other frequencies. This is what happens when we listen to music. we are comparing different tones at different levels.
Our hearing system is VERY GOOD at discriminating at what is "wrong" when comparing sounds. We spend a lifetime listening using a system that has evolved over hundreds of millennia.
move the amplifier F-3dB down an octave to 15Hz.
at 30Hz it is very roughly -1dB relative to passband.
Now listen. you will still hear that 1dB loss in 30Hz output.
I carried out this experiment many years ago and discovered that I had to move the amplifier F-3dB to more than two octaves below the lowest frequencies my speakers could put out, to make "no discernible" loss of bass output.
I adopted a decade as my simple rule of thumb.
If you want 100Hz at full level, then use 10Hz for the amplifier.
If you want 20Hz at full level, then use 2Hz for the amplifier.
BTW, that F-3dB @ 2Hz is roughly the same as F-1dB @ 4Hz.
Many commercial HiFi ampliifers claim -1dB @ ~ 4Hz.
If you were building an active speaker system and wanted the treble driver to operate down to 3kHz (F-3dB), then the amplifier should roll off @ ~ 300Hz and this reduced passband will not affect the audible output from the treble driver.
Last edited:
Once you have decided what input filters you are fitting, then you can design the passband of the amplifier. The amplifier MUST be capable of amplifying the signals that pass through the input filters.
By that calculation, a 1uF with a 22K input impedance sets the -3dB point at 7.2Hz. Would you agree that 7.2Hz with my (not very bassy) speakers is acceptable, i.e., shouldn't notice a difference?
I would say that I would hear a little loss of bass with stand mounters using that F-3dB@ ~ 7Hz
Very deep extending speakers would be quite "wimpy" if driven like that.
But there is a BIG advantage to using a higher filter frequency.
The NFB roll off and the PSU roll off, can BOTH be raised and that saves money and saves space.
set the NFB to < Filter freq/sqrt(2), i.e. NFB < 5Hz
Choose your NFB cap accordingly.
Then the PSU is set to < actual NFB freq/sqrt(2)
Very deep extending speakers would be quite "wimpy" if driven like that.
But there is a BIG advantage to using a higher filter frequency.
The NFB roll off and the PSU roll off, can BOTH be raised and that saves money and saves space.
set the NFB to < Filter freq/sqrt(2), i.e. NFB < 5Hz
Choose your NFB cap accordingly.
Then the PSU is set to < actual NFB freq/sqrt(2)
Last edited:
Everything's built and very difficult now to modify unless I start over so what if I choose to keep the NFB values as is and just modify the input cap size? No issues there I assume?
the NFB should never set the amplifier roll-off.
Doing that breaks D.Self's Law.
And has been confirmed by the capacitor testers that have posted dozens, maybe hundreds of hours work showing that filtering capacitors must be of the highest quality to minimise avoidable distortion.
Similarly the PSU must also be set low enough to allow the amplifier to perform properly.
Doing that breaks D.Self's Law.
And has been confirmed by the capacitor testers that have posted dozens, maybe hundreds of hours work showing that filtering capacitors must be of the highest quality to minimise avoidable distortion.
Similarly the PSU must also be set low enough to allow the amplifier to perform properly.
Last edited:
Hi AndrewT, what do you mean by the above line? Is this recipe for determining snubber RC ?
Thanks.
I used a 1 uf on mine mainly because of the cost and size of the film type cap. I doubt you will ever tell the difference with typical music and speakers.
Attachments
The input RC is 2u2F and 22k (48.4ms) giving an F-3dB = 1/{2PiRC} = ~3.3Hz This formula has been posted hundreds of times in this Forum !
The NFB RC is 22uF and 1k (22ms) giving an F-3dB = ~7.2Hz
In my view this does not give best performance.
The bass roll-off is in the NFB rather than the input filter.
I suggest, as I have done many times before, that you change C2 so that the input filter creates the roll-off you require.
The NFB RC is 22uF and 1k (22ms) giving an F-3dB = ~7.2Hz
In my view this does not give best performance.
The bass roll-off is in the NFB rather than the input filter.
I suggest, as I have done many times before, that you change C2 so that the input filter creates the roll-off you require.
If you go and read previous recommendations you will see that 47uF & 1k is no where near low enough.
Use Cin = 1/ (4*pi*Rin) this sets a lf rollout of 2Hz which is more than adequate considering most loudspeakers flap around uselessly at anything less than 60Hz....... Make Rf x Cf 10 times Rin x Cin as Andrew T above states this network should not be determining the roll off frequency.....
Last edited:
- Status
- Not open for further replies.