[f] AMB on Facebook

ε12 and crosstalk

epsilon12

ε12 and crosstalk

Postby Malvin » July 14th, 2012, 10:28 pm

I'm a bit confused about the following:

If I understand the schematic correctly, R7 and R8 form 20K resistance between L+ and R+.

If say we use 300ohm headphones, shouldn't wee see significant crosstalk between channels? (I calculated 37db, but maybe I am missing something)? Does amplifier's output impedance matter here?
Malvin
 
Posts: 15
Likes: 0 post
Liked in: 0 post
Joined: May 11th, 2010, 8:41 pm
Country: United States (us)

Re: ε12 and crosstalk

Postby amb » July 14th, 2012, 11:56 pm

The ε12's input topology is an opamp summing amplifier (look that up on google if you want). The node at the opamp's inverting input is what's referred to as a "virtual ground" (not to be confused with virtual ground amps i.e., M³ or Mini³). The signal voltage from one channel does not "leak" into the other. The opamp "sees" not voltage swings at that point, but current changes as a result of the summing action.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ε12 and crosstalk

Postby Malvin » July 15th, 2012, 5:18 pm

Thank you Ti! I'll try to educate myself..
Malvin
 
Posts: 15
Likes: 0 post
Liked in: 0 post
Joined: May 11th, 2010, 8:41 pm
Country: United States (us)

Re: ε12 and crosstalk

Postby phaedrus » November 11th, 2018, 4:49 pm

Mornin' all :)

Sorry to necro this thread quite so badly but was about to post about this and found someone already had.

I too was concerned that this would essentially just add a 20K resistor between the channels, which I would imagine to then form a voltage divider with the grounding from the other channel and thus cause crosstalk; think the summing amplifier can be ignored in this.

So I went ahead and mocked it up in LTSpice and found the following:

This is without the protection circuit added:
WithoutProtection.PNG


This is with it applied:
WithProtection.PNG


One channel has a significant input in both cases, the other does not. I could be missing something here but I think Malvin may have been right?

My plan at this point is probably to just use two e12s unless anyone has any wisdom? :)
You do not have the required permissions to view the files attached to this post.
phaedrus
 
Posts: 5
Likes: 0 post
Liked in: 0 post
Joined: November 11th, 2018, 3:47 pm
Location: London
Country: United Kingdom (uk)

ε12 and crosstalk

Postby amb » November 14th, 2018, 1:00 am

What’s in your LTSpice simulation? I’d to see in actual diagram.

Also, the “crosstalk” is not exactly in phase with the original signal, which is curious. If it’s a simple resistive coupling between the channels, there shouldn’t be any phase shift.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ε12 and crosstalk

Postby phaedrus » November 14th, 2018, 6:20 am

Don't have it to hand right now but will post something up soon. Phase shift is probably accounted for by the output coupling capacitor present just before the output, though.
phaedrus
 
Posts: 5
Likes: 0 post
Liked in: 0 post
Joined: November 11th, 2018, 3:47 pm
Location: London
Country: United Kingdom (uk)

Re: ε12 and crosstalk

Postby phaedrus » November 18th, 2018, 8:35 am

OK, finally got around to running some more simulations. First up, here is the complete circuit. Note I've removed the actual amplifier circuits and replaced them with simple AC voltage sources, outputting via an RC network which simulates an output coupling capacitor and a 100k (fairly typical power amp?) load:

SimplifiedProtection.PNG


With one channel having a 1 volt 50hz signal and the other channel having a 1 millivolt 2200hz signal the waveform looks like this:

SimplifiedProtectionWaveform.PNG


If I disconnect the protection circuit it looks like this:

SimplifiedWaveform.PNG


...TBC in next post due to 3 attachment limit :D
You do not have the required permissions to view the files attached to this post.
phaedrus
 
Posts: 5
Likes: 0 post
Liked in: 0 post
Joined: November 11th, 2018, 3:47 pm
Location: London
Country: United Kingdom (uk)

Re: ε12 and crosstalk

Postby phaedrus » November 18th, 2018, 8:46 am

...So I reckon the issue lies here:

SimplifiedProtectionPart.PNG


Which we should be able to prove by doing this:

POC.PNG


The output of which is this:

POCWaveform.PNG
You do not have the required permissions to view the files attached to this post.
phaedrus
 
Posts: 5
Likes: 0 post
Liked in: 0 post
Joined: November 11th, 2018, 3:47 pm
Location: London
Country: United Kingdom (uk)

ε12 and crosstalk

Postby amb » November 18th, 2018, 3:44 pm

You don’t need the coupling capacitors and the resistors to ground at the ε12 inputs for this simulation. As long as your inputs are AC voltage sources with 0V DC offset, then these capacitors and resistors accomplish nothing except act as high-pass filters.

But that’s not the problem. You’re using OP07 as the opamp, not the TL082 as specified. The latter has JFET inputs with dramatically lower input bias current, and will work much better as a summing amplifier.

In the last diagram you posted, you simply disconnected the summing junction of the summing resistors from the opamp. That disables the “virtual ground” at that location, thus causing a direct signal leak from one channel to the other via the summing resistors. No wonder you have crosstalk .

But with the opamp connected, the inverting input of the opamp is a virtual ground. The input signals should have no voltage swing at that point, only current. With an ideal opamp there should be zero crosstalk, and JFET input opamps (with their very high input impedance) come much closer to that ideal. See the following article, and pay special attention to the blurb about ideal opamps and ignoring certain voltage components:

https://www.elprocus.com/summing-amplif ... lications/

If you’re still not convinced, then just use two ε12s (one per channel) to put your mind at ease.
User avatar
amb
Site Admin
 
Posts: 10379
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: ε12 and crosstalk

Postby phaedrus » December 9th, 2018, 1:05 pm

So I finally had some time to sit down and take another look at this. Mostly the shenanigans above can be explained by my forgetting to connect the output of U8 anywhere :lol: Don't simulate whilst drunk, kids. With that connected it's a lot better :oops: . Also swapped out the op amps for LT1057s as they're a drop-in, JFET input replacement for the TL082 and I got bored trying to find a spice model for that which actually worked. Still seem to get around -40db of crosstalk even then, though.

So I've tinkered around a bit, shoved a voltage follower between each channel and the summing amplifier to present a ridiculous impedance, tied the protection diodes to the positive and negative rails of the op amps so that they don't really start conducting under normal operation* and changed the protection diodes for LEDs for their improved current leakage characteristics. This does of course mean I'll have to encase said LEDs in something suitably black so that they don't turn themselves into photo diodes :D

End result? Seems to be less than -200DB of crosstalk. Not bad for an extra pair of op amps, resistors and diodes I reckon. Will get a little circuit board made up for the extra gubbins and then order an e12 when ready :)

*I was seeing interesting current changes on the output tube which didn't match up with the output voltage waveform around when these started/stopped conducting which, although they didn't seem to affect the output voltage perceptibly, I'd rather not have. Yes I'm probably mad.
phaedrus
 
Posts: 5
Likes: 0 post
Liked in: 0 post
Joined: November 11th, 2018, 3:47 pm
Location: London
Country: United Kingdom (uk)


Return to ε12 muting/protect circuit

Who is online

Users browsing this forum: No registered users and 0 guests

Quick Links

AMB audio DIY resources
AMB audio shop

cron