Low pass filter design for TDA1541

[Zodiac]

Hello all,

Please could you give me a hand designing a suitable LPF for use with a TDA1541 dac with non os, going into a solid state amp?

I've been playing around with switchercad and trying to understand butterworth filters but my brain cannot understand what the simulation is showing me!

I think I will need a 3rd order filter, and I was trying to simulate this in switchercad but couldn't figure out what values of L and C to use.

Looking at the circuit below, the left most part is a resonant circuit to compensate for sin(x)/x, using Q spoiling resistor. The 10uf is a nominal coupling capacitor. The circuitry to the right is the RC lowpass input to my solid state amp and the 75k load.

Attached is simulation excl lpf:

[Zodiac]

Here is the circuit simulation using switchercad

[ashok]

I used a second order pi filter and it sounds fine.
R1 is the I/V converter resistor. R1 and R2 cobine to give you a 50 ohm load to the DAC. In my case I have an SRPP tube stage after the filter. It works very well. The inductor is a small ready made coil about as big as a 1/4 watt resistor. In fact it looks like a resistor.
You can tweak C2 and R2 to change the response.
Have fun.
Cheers.

[ashok]

It's attached here.

[Zodiac]

Thanks ashok, how would I use a pi filter in this design? The current to voltage resistor is 1.5k, as it will be used in a transistor based I/V design.

Also, I was playing around with the Butterworth eqns and came to something more useful. Implementing a three pole filter with values of:

----L1---------L3------
..............}
..............}
.............C2
..............}
-------------------------

L1 600mh
C2 100pf
L3 400mh

gives me a -3db at 30khz, as shown in the simulation below. Note this simulation DOES NOT include the RC input filter of my amp (see first post for schematic)

[Zodiac]

Again

[Zodiac]

But look what happens when the RC input filter of the amp is included (see first post for details). The nice filter effect above is completely messed up!

I think maybe I'm going to have to test this with REAL WORLD components....

[ashok]

Hi Zodiac,
Be careful here. The resistor at the DAC output for I/V conversion should be small. If it is too large it will generate large amounts of distortion. Many people on the Net have put up posts on this and I have seen some FTT's of the result. The conclusion was that the resistor should be 100 ohms or lower for low distortion. Ideally the DAC should see a virtual ground like the input of an inverting amplifier stage with voltage shunt feedback.

I think your 1.5K on the DAC might be too high.
I will check out on the 3rd order filter. In my DAC there is an 8x upsampling chip and so a 2nd order would do. I guess you have a NOS DAC. Remember that there are many other things in your signal chain that will remove the HF hash. Might not be necessary to cut it down too much as long as it does not generate suprious products in the electronic chain.
Cheers.

[Werner]

Rin must be very small indeed, necessitating active gain later on.

The circuit below is for a (much louder) TDA1543, but can be scaled to 1541 duties. Simulate it, have a look at group delay up to 20kHz (bridge the DC coupling cap first), and accept that the first bunch of images up to 40kHz do not really need that much attenuation anyway as they are correlated with the music's treble contents, not the bass.

By playing with the ratio of positive feedback you can tune in peaking at 20kHz to compensate for Sin(f)/f, but I'm not sure this is a good thing as phase response will suffer.

[Zodiac]

Thanks ashok and werner for your replies.

The reason the resistor is so large is that it is used in the very last part of the I/V stage below. Looking at the diagram (rbroer's design BTW) R4 is the I/V resistor. Sin(x)/x compensation therefore needs to be put in series with this load.

The last filter response I put up was quite good, but I have another question. The inductor values needed to drive a 75k load are very high. If I put some kind of buffer (e.g emmitter follower??) between the ouput of the I/V stage and the amp would that change things? And how would this be implemented?