My gainclone set up and impedance mismatch?

[spacemen12]

Hi,
I have this set up:
Rudolf Broertjes' SS I/V Gain Stage -> passive preamp -> Gainclone -> Fostex fe103e in BLH (8 ohm full range speaker)

Everything is stock except on the gain stage I have two 4.7 uF in super E instead of one 4.7 uF (from http://diyparadise.com/web/index.php...id=95&Itemid=5)
. My Gainclone is Audiosector stock. My passive preamp is as follow:

There is about 20cm of wire between the preamp and the Gainclone.

I wonder is there might be a problem of impedance mismatch between the source and the amplifier. I read about the Pass DIY B1 and it look nice, but am I creating virtual problem and try to solve them? What are the symptoms of impedance mismatch?

Also is this setup can act as a low pass filter, sometime an issue with passive preamp?

Thanks a lot!

[pacificblue]

Impedance matching means that the source output impedance should be the same as the following input impedance. That is something you don't usually worry about at line level. Exceptions confirm the rule.
In your setup there is a different issue. With a 10..1000 Ohm shunt you might as well skip the 22k input resistor. How is the performance, when the LM3875 sees ~10 Ohm input impedance? And will it be the same, when it sees ~1000 Ohm?

[spacemen12]

Originally Posted by pacificblue

In your setup there is a different issue. With a 10..1000 Ohm shunt you might as well skip the 22k input resistor. How is the performance, when the LM3875 sees ~10 Ohm input impedance? And will it be the same, when it sees ~1000 Ohm?

The volume change a lot from 10 to 1000. I cannot judge for the rest, because the conditions are not equal. What can be the symptom of impedance mismatch?

[pacificblue]

Originally Posted by spacemen12

The volume change a lot from 10 to 1000.

That is the intended purpose. The negative aspect is that the DC offset will change as well. That can bring all kinds of troubles, beginning with increased distortion up to speaker destruction. Ideally you want the amplifier to see the same impedance at both input terminals. Usually that is achieved by making Rb = Ri and Rf = Rin. In an AC coupled amplifier Rin is isolated from the source by a series capacitor, so what is labeled Rin actually is Rin. In your DC coupled amplifier it is not, but Rin = Rin||Rshunt||(Rserie+R23). Rshunt is the determining factor here due to its size, therefore you get a value of ~10-1000 Ohm, when you want 22 kOhm. You should consider to convert the chipamp into an AC coupled design.

Originally Posted by spacemen12

What can be the symptom of impedance mismatch?

Mostly increased losses in the cable. In some cases that can deteriorate the sonic quality, especially, if the input impedance of the amplifier changes the frequency of the source's output filter too much, e. g. when Rin seen from the source is far too small.

http://en.wikipedia.org/wiki/Impedance_matching

[spacemen12]

Originally Posted by pacificblue

That is the intended purpose. The negative aspect is that the DC offset will change as well. That can bring all kinds of troubles, beginning with increased distortion up to speaker destruction. Ideally you want the amplifier to see the same impedance at both input terminals. Usually that is achieved by making Rb = Ri and Rf = Rin. In an AC coupled amplifier Rin is isolated from the source by a series capacitor, so what is labeled Rin actually is Rin. In your DC coupled amplifier it is not, but Rin = Rin||Rshunt||(Rserie+R23). Rshunt is the determining factor here due to its size, therefore you get a value of ~10-1000 Ohm, when you want 22 kOhm. You should consider to convert the chipamp into an AC coupled design.

Hi,
What if I put a 22k in parallel to the 10-1000 Ohm? Will it solve the problem?

Thanks

[spacemen12]

Hi,
Please ignore the previous post.

What is an acceptable DC offset?

I did this preamp to have better control at low volume. Since my room is small and my neighbors are close, I don't need high volume, and this setup was the best that I found. Is there another solution to have excellent control for attenuation between -60dB and -30dB. Here I can switch to +/- 2dB with discrete switch.

Thanks

[spacemen12]

Hi,
I measured the offset for different values of R_shunt. The DC offset increases as R shunt increases. It does vary from 3-5mV at the speaker terminals. Hopefully it is normal.

How can I test the other symptoms?

Thanks

[pacificblue]

5 mV or less is an excellent value for a chipamp. Especially for a DC coupled one that is connected to an AC coupled source.

CMRR is far from ideal, so your setup will be more sensitive to disturbances that act on the connecting cable. Try to keep it as far away from external noise sources as possible to make up for it, e. g. don't run the audio cable next to power cables, transformers, SMPS, other electric or electronic gear, etc. In most cases a distance of 20-30 cm should keep you on the safe side. The noises mobile phones tend to induce, might be more annoying in your amplifier than usually in other devices.

Originally quoted from http://www.sengpielaudio.com/calculator-bridging.htm
If the load impedance is 10 times or more the source impedance, this
is called a "bridging" impedance.
It results in maximum VOLTAGE transfer from the source to the load.

If the load impedance equals the source impedance, this is called
a "matching" impedance.
It results in maximum POWER transfer from the source to the load.

In the audio world we are happy with the maximum voltage transfer.
But we are often confronted with storytellers teaching us the myth of
power matching. Never believe that.

The worst case load impedance in your case is 47k||(10k+(10r||22k)) = ~8,25 kOhm. The source impedance is 83 Ohm. ~100:1 looks good for maximum voltage transfer.

As a rough estimate the corner frequency of the source's output filter should remain five times below the lowest audio frequency. Let us say you had fantastic speakers that were able to reproduce the lowest note of a big church organ at around 16 Hz at audible level. The filter should then have an f(-3) below 3,2 Hz. R = 1 / (2 * PI * f * C) requires R of greater than or equal to 497 Ohm. You seem to be on the safe side there as well.