what could be the reasons for High freq roll off after 14khz

[rhythmsandy]

I have a lm4702 2sc5200 1943 pair amp. I have measured in room using a RTA with berdynamic mic.. we were playing through sony laptop which sound ashtonishingly good but we have seen a rolloff from 14khz to 20khz

im using vifaxt25sc ring radiator using a first order crossover but used a first order crossover with 28 .47uf in parallel to achieve the capacitor.

is the rolloff because of the possibilities..

1. the amp has a 7uH inductor approx ( 10 turns of 5mm dia 1mm thick coil ) at the output
2. the capacitors in parallel are making effective inductance because of leads ( infact they have to get reduced coz they are in parallel)
3. the mp3 320kbps format might not have enough HF information
4. Do room acoustics play role ( since we have kept the mic on axis at 1m)

i would like to know that is the inductor at the otuput makes any difference?
what if its removed? any problems? could the rolloff be happening because of that?

 

[rhythmsandy]

I found couple of things..

1. the capacitor datasheet specifies that there will be rolloff after 100khz
2. I have found the inductor value as .18uH and checked what kind of imp it would give at different frequencies I found that..
the imp at the following frequencies..
at 1khz its 1.13
at 10khz its 11.13
at 15khz its 16.96
at 18khz its 20.36
at 20khz its 22.62

is it the reason why its attenuating the upper frequencies...

 

[rhythmsandy]

oops sorry the measurement is done using pink noise with presounus live 16.4.2 so its not related to mp3s..

 

[Bare]

Erm Measured OR heard ?
IMO genuinely unlikely you actually heard the problem.. IF.. over 45 yrs old and a Human Male

 

[rhythmsandy]

measured...

 

[rhythmsandy]

i found where the prob is coming frm... its coming from the RL network on the output when i connect the output before the inductor everything is working perfect all the hf information and transparency opened up... what values of RL recommended?

what if the RC network is removed..

 

[TANDBERGEREN]

What values did You have there?
Linkwitz-Riley-network usually consist of a 10 Ohm resistor in series with a 0uF1 film capacitor. Different values and configurations are also seen, but I guess about 98% of amplifiers use this one.

I would not recommend to remove the filter.
Would give "infinite" bandwiith out with all problems that would occur in Your speakers and amp.

 

[rhythmsandy]

Ive tried removing both the filters just straight but I dont see any problems its playing fine and clear. Do you mean that if I increase the volume then the amp would oscillate?

anyway there is a NFB so why will be the bandwidth infinite?

 

[lineup]

You can most times remove the RL filter. The one with coil.
But you are playing with fire if you remove the RC filter.
Sooner or later there will be smoke.
Keep 10R in series with 100nF

 

[ezavalla]

Linkwitz-Riley-network usually consist of a 10 Ohm resistor in series with a 0uF1 film capacitor. Different values and configurations are also seen, but I guess about 98% of amplifiers use this one.

It is a Zobel network, not a Linkwitz-Riley one

 

[rhythmsandy]

but I would like to know what value of the inductance to be there as output zobel as current value is about 1uH with 10ohm 2W resistor and a .1uf with 10ohm resistor in series is shunted at the output.

I have a rolloff starting from 14khz what should be the values of the inductor so that the amp will not oscillate at higher frequencies..

or can i just eliminate the zobel as you said 98% of the amps dont have it.

what if the output is connected before the inductor? will the amp oscillate as its RLC circuit?

 

[multisync]

Most amplifiers that use an ouput inductor have a resistor in parallel with the inductor usually 10 ohms if your amplifier is the same the resultant impedance will never be higher than 10 ohms.

measure the amplifier frequency response with an 8 ohm resistor load. Is it flat to 20k?

Measure your amplifier frequency response with your speakers connected, use a low power level, you don't want to blow your tweeter. Is it flat? measure the voltage at the tweeter terminals? is it rising as the frequency goes from 20-20,000?

check your arithmatic XL=2*pi*F*L

=6.28 *20,000* .000018

= 2.26

in parallel with a 10 ohm resistor = 1.8 Zohms

if you made a measuring mistake with the inductor it would be much less.

 

[rhythmsandy]

im measuring the voltage across the output and its giving as 10mv i believe its the DC offset... but no voltage output as you have specified at the output of the speaker connectors...

 

[rhythmsandy]

sure there is a rolloff atleast by listening since its major change so what is the solution now since this amp would be going for high power application in PA

 

[rhythmsandy]

i have connected the speaker cable directly before the inductor it still worked fine without any problems but at moderate levels but no idea how it sounds at higher level but just have a doubt that will that blow off anything?

i dont have a cro thats a major problem..

 

[TANDBERGEREN]

It is a Zobel network, not a Linkwitz-Riley one

Damned, You are right.



Linkwiz-Riley we use in filters, as in active crossovers, speaker crossovers and like theese.

I am truly embarrased

 

[rhythmsandy]

did anybody faced such prob before? whats the solution now?

 

[TANDBERGEREN]

Waiting for an answer to what the value of the ZOBEL-networks components were.
I can't really understand such rolloff with correct values there.
One could also give the output inductor a check.

Ah, You have posted that too.
.18uH is it measured, or have you just picked one from a supplier?
Tried with 3,9Ohm in paralell?
Shuld push the roll-off even higher, but 14Khz is extremely low.
What kind of feedback is used?

 

[rhythmsandy]

here is the schematic you can see the output stage..
Imageshack - lme49810.jpg

 

[TANDBERGEREN]

C12 and C13 is in place, and with the correct values?

I am thinking there must be a feedback problem here, or does the R31 get hot as H***?