Transformer summing to mono

Design and use of the various types.
Post Reply
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Transformer summing to mono

Post by dave slagle »

Why I think it is a bad idea.

The traditional way to sum stereo to mono is to add series resistance at 10X the source Z and then make sure the load is 10X the series resistance. Some insist that using a transformer gets you around the need for series resistance and others insist that putting the secondaries in series is the proper fix.

It is my contention that all the transformer does is adds a small series resistance in the form of winding DCR and while it does work, so does a simple Y connector.

Lets start with the basics. The spice files have three options. Left is the Y-connector approach. Middle using a transformer with parallel secondaries and right the series secondaries.

Here is proof that the test method is giving the proper results. I am showing right channel only (the left is the same) and the coils have 100 ohms DCR. We see that the no transformer source sees exactly 1meg and transformer sees 1 meg + 200R and the finite inductance causes a LF corner.

Image

Now lets sum the left and right and we get no change.

Image

that is simply because both signals are identical. Watch what happens when we take the left channel and make it 1/10th the right channel.

Image

If I turned the left channel off on the no transformer the right would simply see the 100 ohm source Z of the of the left channel. Also note how 500 ohms on the transformer version is the series DCR of the 4 coils + the 100 ohms of the Left channel.

Sure the 500 ohm load of a 100 ohm source isn't such a bad thing and if you increased the coil DCR's to 225 ohms we would be at the 10X Z ratio but the important thing to note is a pair of 900 ohm resistors would do the same thing and the transformer isn't doing anything "magical"

Now lets look at the series secondary with different signal outputs from left and right.

Image

Not what I expected but the parallel option looks better.

Here are all three together.

Image

The above was the load seen by the right channel outputting 1V with the Left channel outputting 0.1V. Lets look at what the left channel outputting 0.1V sees into the 1V of the right channel.

Image

Hmmm. The left (lower output) channel doesn't much like any of the connections but it is interesting to note that the series secondary fared slightly better but.....

This was all done through AC analysis to illustrate a point. Looking at it in the time domain where you can vary both amplitude and frequency makes things ugly very quickly. Still need to get my head around that one.

dave
Get Your Fix
www.hifiheroin.com
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Series coupled vs. discrete cores.

Post by dave slagle »

In the above I assumed that all of the windings would be on the same core and it was pointed out to me that they will behave differently if two discrete transformers were used.

Indeed things look different. I lowered the load value to 10K to get some realistic numbers and the results are not what I would have expected.

I also dropped the inductance to 100hy's

Same voltage in each channel behaves as expected. (lines overlap)

Image

Dropped the left channel voltage from 1V to 0.9V to break the lines apart. Interesting to see the load value on the right went up and the load value on the left went down.

Image

dropped the Left channel to 0.5V and the pattern continues

Image

Down to 0.1V on the Left and more of the same.

Image

down another decade and the 1V source sees 20.25K as a load and the 0.01V source sees just over 100 ohms.

Image

hmmm.

dave
Get Your Fix
www.hifiheroin.com
Post Reply