Help me with impedance balancing

Advertise with us
shop deals on amazon
Shop deals on ebay
GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
shot

shot

Well-known member
Joined
Nov 28, 2008
Messages
669
Location
Zagreb, Croatia
Hi,

I'm in the end of the racking job on some Beag preamp modules. I'm done with PSU, recapping and metalwork, and unit sounds fine. But as a last thing I'd like to balance it's outputs doing impedance balancing.

What would be the right way to do this? I'm confused since I've read some different articles on it and got an advice that confuses me even more.

I got advice from a local guy to just stick 100R in between ground and cold on xlr and be done with it. That seems to me not to be the right way, since I don't know actual output impedance of the unit and 100R is a blind guess...

I got an (real wild diy!) idea to throw a 5K potentiometer in the same spot and run my output wire deliberately near the power xfmr so it picks up some hum, and then monitor the input on my analyzer until I tweak the pot to the point of the max cmrr. Later I replace the pot with a measured value. Would that work?

Or should I replicate C32/R55 network (or more?) on the cold side grounding it? See the attached schemo...
(notice there's three outputs, I'm using only the first one)

:)

best,

Luka
 

Attachments

  • BeagEME011sch.jpg
    BeagEME011sch.jpg
    77.3 KB · Views: 94
Usually for opamp output, where the NFB is usually huge, the output resistance of the opamp as the resistance inside the feedback loop (R65 in your case) are ignored. That stage you could call 0Ω for most practical applications. C33/R56 is what you want to match. I wouldn't drive the external world from an EQ stage, but that's how it's done so let's use it. I would add in series with C33 a resistor, 100Ω is fine, maybe less is ok too. 33Ω is used by jense, using a 2W resistor with 40 wraps of wire on it makes an OLI, which is just that, a resistor parallels with an inductor. Something like 33Ω and 20µH. This gives low output impedance at audio freq but not so low (about 33Ω) at much higher freq, making the opamp not worry about capacitance for long cables or RF noise injected at the output. In your case the output stage seems to be prepared for driving some capacitance (R65 inside the FB loop is used for that) Then you probably don't need the OLI.

So, matching C33 and R56 would be all what you need. You should also care where you are going to connect it, that could be tricky too! Close to the opamp decoupling should be the appropriate place, but then the decoupling point shouldn't be too dirty or it could became a problem... If you are ok with the place the output is leave it there and use RC network. Looking for a better place to connect it may improve something, less leakage to the PS from the opamp, maybe less distortion but hardly measurable. The problem as I mentioned if the point is corrupted by other signals, could put that dirt in your output...

JS
 
Thank you Joaquins and Ian! I was hoping you'll chime in with some food for thought!

I already read the SOS article that Ian suggested. That's where I got my (not so good) idea of sticking the potentiometer and finding the best possible cmrr...

I'll go with the C33/R56 on the cold side to ground. But it will be a problem how to physically put it in there. This is the module with four channels boxed into a metal cage. It is very tightly packed. It was a nightmare when I went to disassemble it to gain access to boards in order to recap few lytics!
I think I'll first try to mount them on the XLR on the back of the case. I'll experiment with what ground to connect it to. This module has two separate audio grounds and chassis ground. Two audio grounds must be joined for the unit to work, and I've joined them at the connector (original connector, not the xlrs I've added on rack case)
I must go to work now! Duh! I'll try it tonight when I get home!!
Thanks mates!
 
shot said:
Hi,

I'm in the end of the racking job on some Beag preamp modules. I'm done with PSU, recapping and metalwork, and unit sounds fine. But as a last thing I'd like to balance it's outputs doing impedance balancing.

What would be the right way to do this? I'm confused since I've read some different articles on it and got an advice that confuses me even more.

I got advice from a local guy to just stick 100R in between ground and cold on xlr and be done with it. That seems to me not to be the right way, since I don't know actual output impedance of the unit and 100R is a blind guess...
Click to expand...
Agreed. According to the schemo, the actual impedance of the output is a very low resistance (about 1 mohm in series with an inductor - which value is to determine - in series with a 2uF cap. Clearly the 1mohm (or whatever value results from calculation) can be omitted. The inductance value would be about 2 to 10uH. The effects of not compensating this inductance will not be seen below 100kHz. So, in the end, you would be done with a single 2uF cap.
If you want to be serious about it, you should match the caps.
In fact, it is quite difficult to properly match the caps, so the most common scheme consists in making the series resistance dominant, by increasing it via a simple resistor; this has also the added benefit of protecting the output from shorts and increasing stability.
The output of an opamp shows a dominant reactive term (equivalent to an inductance) due to the drecreasing amount of NFB correleated with the decreasing open-loop gain. In order to calculate it (in fact simulating it is easier), one needs knowing the type of IC and the associated values of the FB loop.
I got an (real wild diy!) idea to throw a 5K potentiometer in the same spot and run my output wire deliberately near the power xfmr so it picks up some hum, and then monitor the input on my analyzer until I tweak the pot to the point of the max cmrr. Later I replace the pot with a measured value. Would that work?
Click to expand...
No. The most important term that needs to be balanced at 50Hz is the capacitor (C32). The 5k potentiometer would do nothing.
Or should I replicate C32/R55 network (or more?) on the cold side grounding it?
Click to expand...
That would be it. Note that the value of the 22k resistor is absolutely not critical there. It has virtually no effect on CMRR; it just makes sure the cap sees a discharge path.
 
You should check out this thread. http://groupdiy.com/index.php?topic=56345.msg724344#msg724344
 
abbey road d enfer said:
In fact, it is quite difficult to properly match the caps, so the most common scheme consists in making the series resistance dominant, by increasing it via a simple resistor; this has also the added benefit of protecting the output from shorts and increasing stability.
Click to expand...

If I understand you correctly, this means I should add another resistor on the end of the signal, right?
What value I should use? Less than 1K or greater?
And that value is to be mirrored also on the cold side for balancing, right?
 
abbey road d enfer said:
The output of an opamp shows a dominant reactive term (equivalent to an inductance) due to the drecreasing amount of NFB correleated with the decreasing open-loop gain. In order to calculate it (in fact simulating it is easier), one needs knowing the type of IC and the associated values of the FB loop.
FOR RFI IMMUNITY, the dominant effect for ALL balanced outputs is LAYOUT.  You need to have all these extra bits laid out symmetrically so the stray capacitances are the same on both legs.

Having 100p COG/NPO ceramics AT the XLRs between p2 & 3 and p1 helps to swamp you layout assymmetries.  Remember the aim of the exercise is not to maintain some mythical balance.  IT'S TO REJECT INTERFERENCE.  (For a microphone output, these should be 10n or greater.

No. The most important term that needs to be balanced at 50Hz is the capacitor (C32). The 5k potentiometer would do nothing.
Yup.  Buy loadsa electrolytics & sort them.

If you can afford the LF rolloff, using films will be more reliable in the long term.

ANY twiddle pots make things worse for RFI cos the assymmetry they introduce.
Click to expand...
Click to expand...
Click to expand...
Click to expand...
 
shot said:
abbey road d enfer said:
In fact, it is quite difficult to properly match the caps, so the most common scheme consists in making the series resistance dominant, by increasing it via a simple resistor; this has also the added benefit of protecting the output from shorts and increasing stability.
Click to expand...

If I understand you correctly, this means I should add another resistor on the end of the signal, right?
Click to expand...
Correct.
What value I should use? Less than 1K or greater?
Click to expand...
  Anything between 22r and 100r should be adequate; for a number of reasons (some being rather irrational  ;) ), 47r seems to be the most commonly adopted value.
And that value is to be mirrored also on the cold side for balancing, right?
Click to expand...
Correct.
As ricardo wisely pointed out, stray capacitances are rarely balanced, so putting a pair of discrete caps that become dominant is a thoughtful addition.
As to the pairing of electrolytics (which is as temporary as their operating life), you have to remember that it's the absolute difference of impedance that is paramount in determining the LF CMRR; as a consequence, making the nominal Z as small as possible is more important than pairing them exactly. Use the largest practical value; that will alsooptimize distortion.
 

Similar threads

Nidare
Confusion about output driver and signal levels
2
Replies
20
Views
918
Newmarket
Newmarket
S
v672 (and more general) input impedance
Replies
0
Views
211
sea mouse
S
S
“H / Balanced-T” Attenuator structure
Replies
19
Views
895
Graham Nystrom
Graham Nystrom
S
6V6 Output Transformer NFB resistor mod / value
Replies
14
Views
683
ruffrecords
ruffrecords
V
Repurposing a swinging input EQ PCB for an API 553 based EQ
Replies
1
Views
1K
Vyvyan
V

Latest posts

Back
Top