[Dan Banquer]

I thought I would drop this brief observation on stereo separation to make some of you folks think about some real issues.
                      d.b.

Typical stereo separation at the listening position is usually somewhere around 8db to 10db or so I have read. This alone would appear to nullify further investigation of electronics and stereo separation, but over the years I have made some interesting observations that I would like to share with all of you.
In designing and building a few line level units I have observed that the coupled signal when doing stereo separation testing has been 90 degrees out of phase from the input signal. It also appears that it is capacitively coupled and decreases as the frequency drops, and typically 6 db per octave. In reducing this coupling I have also observed better noise immunity. For Example: If I use an “unshielded” wire to go from the back panel input to the board I will degrade stereo separation and at the same time, due to the high input impedance of the input I am connecting to I will have more noise due to pick up. If I use a piece of coax for the same connection I get an increase in stereo separation and lower noise pick up. It would appear that a higher level of stereo separation could well be more indicative of better noise immunity in this application. My own personal observation is that reducing the 90 degree out of phase coupling gives a better “image” along with reduced noise. At present I generally shoot for close to 70 db of separation at 20 kHz, I find audibility is very difficult above this. The above makes me ask the question; Is it stereo separation, or noise immunity, or both?
Recently, I was designing a line stage - headphone amp. In the design process I decided to add resistive coupling across the headphone outputs. This increased the “image” across the front of my head giving the “ illusion” of a wider soundstage. I repeated a similar experiment with my loudspeakers and found that resistive coupling levels (no phase difference between the coupled signal and the main signal) can be much higher than capacitively coupled signals before I was able to detect this. This makes sense if you think about it, because all that is being added is small amounts of mono to the “mix”. I should also note that all of the listening tests that I did on my own used CD only.
If anyone else has something substantive to add; feel free.

[_scotty_]

Dan,could you post a picture of the inside of your linestage showing the routing and physical location of the wire runs inside the chassis. It sounds like you are describing interchannel crosstalk or capacitive coupling between the channels via the wire routing technique used.  Without a picture I am conjecturing here about why the loss of stereo separation occurs.  This image
shows how I wired the input wiring in my buffer  to the selector switch.See link below. What is not obvious from the image is the spacial separation between the wires.  The back panel is the ground plane for the input signals. The jack's ground is common to the backpanel no isolating washers are in use.  The appearance  of a twisted pair
of wires carrying the signal to the selector switch is an illusion. Only the hot leg is switched. The ground leg is only connected at the jack and is floating at the other end.  This technique of twisting wire around the signal hot and connecting one end to ground is used to reject hum and add capacitance to the wire run without resorting to using shielded wire.  
http://www.audiocircle.com/image.php?id=1205
I spaced the wires as far apart as possible to avoid interchannel crosstalk and bleed through between active sources. The potential for capacitive coupling exists at the selector switch and the DACT stereo potentiometer.
The dual mono board layout helps minimize the loss of stereo separation.
The buffer performs better than it looks.. Scotty

[Dan Banquer]

Hi Scotty: First of all I would like to thank you for your post. Here's a link that shows the inside of one of my pre amps.
http://www.audioholics.com/productreviews/avhardware/SCPA1p2.php
  From what you describe it appears you have some grounding issues inside your unit and when I have time
 ( either later on today or this weekend) I will offer up some ideas that you may want to experiment with.
                     d.b.

[mgalusha]

For a recent preamp project I used some standard 24ga stranded wire running inside a braided shield. The only coax I had on hand was quite heavy and stiff and not very suitable for wiring the preamp. The RCA jacks are isolated from the chassis and the shield is grounded at the same point as the RCA's. The preamp is very quiet and I have no noise problems. It's effectivley coax.

One reason I didn't ground the input jacks was the desire for a ground lift switch. The line cord is grounded to the chassis and I have a small SPST toggle switch on the back which connects/disconnects the circuit ground from the chassis. In my system it makes almost no difference but it seemed like a good idea and was nearly free to implement.

I suspect most of my opportunity for crosstalk exists in the selector switch and volume pot due to capacitive coupling. I haven't measured the crosstalk but this has made me curious, perhaps I'll have to drag it down to the bench and see how it fares.

mike

[Dan Banquer]

"I suspect most of my opportunity for crosstalk exists in the selector switch and volume pot due to capacitive coupling. I haven't measured the crosstalk but this has made me curious, perhaps I'll have to drag it down to the bench and see how it fares. "

That's exactly what I found.  But  more of it at the Volume control due it's high impedance. What I did on the SCPA 1 was to add spacers between each channel  and then cover the spacers with copper tape and connect the copper tape to ground. This dropped the crosstalk by over 12 db and into a range that my observations suggested to me was below audibillity.  Nothing like a bit more space between high impedance networks and added ground plane to help keep things quiet. eh? Please post what you find on the bench.
           d.b.

[Dan Banquer]

Hi Mike:
    Any number of different types of coax can be found at wwww.belden.com.
What I look for for an "inside the box" coax is the O.D. (outer diameter). I think you can find a good number of choices that are less than 0.2 inches O.D.,  capacitance per meter, and shield coverage.  I think you will find plenty of choices here. Typical capacitance of most of this type of coax is 100 pf per meter or less. Shield coverage is typically 90% or better. If you can't find a local distributor for Belden,  contact me and I will get you one here in the Boston Metro Area. Now if you will excuse me I need to help scotty.
           d.b.

[Dan Banquer]

Hi Scotty:
   I have looked at the pictures of your unit and I have number of things I am going to recommend. Please bear with me as this is going to get a bit involved.
1.   Isolate the back panel RCA connectors, most of these connectors come with do come with chassis isolators so I hope they didn’t hit the trash.
2.   Most of the chassis connectors come with solder tabs for the ground connection. What I have done on a number of units is to solder a 1/8 inch flat braid connecting all of the input right channel connectors on the back panel and the same for all the left channel input connectors. This may seem a bit extreme but there is going to be a method to my madness so to speak.
3.   We now need a ground to replace the one you used to have on the back panel. Classic textbook procedure tells us we need to put this right on the perf board you are using. Strip  about 6 inches or more of wire and install it on your perf board. Connect ALL of your grounds to this wire. This will establish our center point ground connection.
4.   For the signal wires going to and from the Volume Control, ground the signal wires on the board. For all signal wires going to the back panel they should be grounded at both ends. On the board and on the back panel RCA chassis connector. For wire choices I have made some suggestions to Mike’s post that you may wish to further investigate.
5.   I’m not sure how you are grounding the chassis, so I’m going to make a few stabs in the dark here. If you are using  a two prong AC input then your perf board ground should be connected to the chassis at the ground of your unregulated supply. If you are using a three prong  AC input then please get back to me as we have a number of options.

I have outlined a lot of work for you. Take your time because this kind of stuff just isn’t done overnight. If you find yourself overwhelmed walk away for bit and come back to it later. The other thing you can now say to your self is that you have just joined the ranks of experienced professionals, and you are now in the learning process that the rest of us have gone through at one time or another. Dare I say that you will be rewarded with the final results?
                    d.b.

[mgalusha]

Dan,

I have no problem getting Belden products, in this case it was simply a matter of what I had on hand. I like your idea of adding more physical seperation to the layers in the volume control with the addtional ground planes.

I recently picked up a remote controlled relay switched attenuator kit. Since each channel will be a couple of inches from the other, capacitive coupling shouldn't be an issue.



I don't know how much all the relay contacts will impact the sound but the cost was low enough ($95) to give it a try. It provides 100 steps and 4 inputs.

I appreciate the input on shielding & grounding.

Mike

[Dan Banquer]

Let me know how the attenuator works out.
               d.b.

[mgalusha]

Let me know how the attenuator works out.

Will do. I am waiting for a new front panel to arrive from Lansing Instrument. That way if I don't like the results I haven't hacked up my existing panel. For $16.00 it was worth it to me.

[Lost81]

I recently picked up a remote controlled relay switched attenuator kit.

They have some very nice chassis at attractive prices.

I notice they are located in Hong Kong though.
How was it dealing with them and how much did shipping cost?

Thanks!


-Lost81

[mgalusha]

They have some very nice chassis at attractive prices.

I notice they are located in Hong Kong though.
How was it dealing with them and how much did shipping cost?

Suprisingly good. I ordered it off their website along with a power transformer ($6.00) and paid via PayPal. I received an email the next day asking how I wanted it shipped. It was $15 for surface shipping (4-6 weeks) or $30 for air. I choose air and received it about 4 or 5 days later. I have a total of $131 invested and it seems to be well made.

I connected everything up on the bench and hooked up a battery as a voltage source and put the meter on the outputs and ran it up and down and the voltage seemed to ramp up and down very smoothly.

The only thing I'm not happy about is that when it powers up the circuit defaults to input 2 and -88 on the level. I would rather see it default to input 1 and -99 on the input (full off) but in use this will be nearly irrelevant. I suppose one might be able to hack the eeprom code but I'm not that ambitious.

I'll post some picutres in the Lab circle when I do the installation with the new faceplate although it looks exactly like the pictures on their website.

mike

[_scotty_]

Dan, thank-you for your suggestions. The perfboard has a star ground point
for both channels after the regulators and the ground from the pot is taken to it. The circuit is isolated from the heavy charging currents in the main supply by not having a direct connection to the center tap on the transformer.  The back panel
was chosen to be the low impedance reference plane for the signal ground.    The red 12Ga wire jumpers from the star ground point on the perfboard  over to the signal ground reference plane on the back panel. The back panel is then connected by a 12Ga wire to the
to the center tap at the main supply on the Jensen caps.  The circuit is a MOSFET based  singled ended class A buffer with a cascoded active current source.  Only one MOSFET is actually in the signal path.  The other parts in the signal path excluding the wire are the jacks,selector switch,10k DaCT pot, 25,50ohm resistors, and a 680 ufd 35 volt NX Blackgate cap. There is no hum and
the noise from the system as a whole including the power amp is only discernable with your ear right next to the loudspeaker. The buffer has a better than 30mHz bandwidth with .003THD and when grounded in this fashion in this particular chassis does not oscillate.   Had oscillation, hum or noise been present some other grounding scheme would have been necessary.   Scotty

[Dan Banquer]

"The circuit is isolated from the heavy charging currents in the main supply by not having a direct connection to the center tap on the transformer. "

Hi Scotty;
     This is something I definetly would not recomend.  The center tap of the transformer is the  ground return for the circuit assuming you are using the transformers in a "bridged" configuration.  Your charging currents after the unit has been turned on are much lower than you may think.
One of the things that most of audio assumes is that if a system has a small amount of hum that appears to be inaudibe, it's O.K. That's really not the case. What that small amount of hum is telling us is that there is a ground loop. Athough pretty small, that ground loop will leave the system more susceptible to interference and give overall more noise to the system.  Thanks for your post and best of luck with the unit.
                   d.b.
P.S. I used to have that small ground loop problem: I got rid of it and I'm glad I did.

[Dan Banquer]

There is one other thing I forgot to mention.

Grounding: Easy in theory and Hell in practice.
 No wonder to me that I have lost most of hair on top if my bald head; it's a long learning process for most of us.
             d.b.