A Simple AC-Coupled Balanced Line Reciever

Where we discuss new design ideas and modern spins on old ones.

A Simple AC-Coupled Balanced Line Reciever

Postby mediatechnology » Fri Nov 16, 2012 5:22 pm

I needed an AC-coupled, high impedance balanced line receiver with exceptional low frequency common mode rejection.

I wanted to use a bipolar input op amp with low-value film capacitors for a "non-electrolytic" approach.
This requires a high impedance input to reduce capacitor matching requirements or bootstrapping a relatively low impedance to make it look larger.

One of the concerns having large value bias resistors are noise, due to input noise current and offset due to high bias current.
Whitlock's InGenius papers make the case for low value bias resistors bootstrapped to look like high value ones: http://www.jensen-transformers.com/an/ingenaes.pdf
Then I remembered T-bias and what it provides Mic inputs: viewtopic.php?f=6&t=191&p=2404

Image
AC-Coupled Line Input with T-Bias Network to Raise Input Common Mode Impedance

A high differential input impedance is not required. The LF cutoff for differential signals is <8 Hz.
R1 and R2 provide differential termination.
R3 provides a DC return for op amp bias currents, both DC and noise, as well as a bleed path for the capacitors. (Bleeders on the LH side are not shown.)

Normally huge DC offsets will develop across a 1M R3 and they do here.
The input noise currents also convert to a relatively large voltage across R3.
(Do note that the source impedance in series with the reactance of Cin are also in parallel with R1+R3.)

What makes the circuit practical is that these current-induced voltages appear in common mode.

Common mode rejection is realized by the THAT1240. (I needed unity gain otherwise a 1246 would have been used.)
Thus, the noise current developed across R3 is canceled and the DC offset, also in common mode, is virtually eliminated.

Because R3 is so large in value and the common mode impedance high, capacitor matching requirements are greatly relaxed.
This circuit provided -86 dB common mode rejection using 10% film caps regardless of frequency from 10 Hz - 1 kHz and beyond.
There is not any measurable noise or output offset penalty with R3=0 Ohms or R3 = 1M Ohm.

I developed this circuit for the DC-coupled preamp where the "-" input is tied to the floating Vcm supply and the "+" input to the 1510 output.
I had previously used a single-ended BIFET amp and decided that additional common mode rejection was needed.
An added bonus is that a bipolar op amp could be used along with film capacitors.
http://www.waveulator.com
http://www.ka-electronics.com
http://www.linkedin.com/in/waynekirkwood
“Logic will get you from A to B. Imagination will take you everywhere.”
User avatar
mediatechnology
 
Posts: 1826
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas

Re: A Simple AC-Coupled Balanced Line Reciever

Postby ricardo » Tue Nov 27, 2012 5:50 pm

Bit nervous of using LME49860 as voltage followers in this application.

It appears to be a deluxe version of LM4562 which often exhibits latching behaviour on startup in single supply applications.

In TL07x type topologies, this is often accompanied by phase reversal on overload. Bad choice for followers or LF filters that might be overloaded.

But both NI devices are rated down to +/- 2V5 so I can't imagine they haven't checked this.

Have you done some severe overload tests? Not that the intended application does this but I'm thinking of general purpose use.
ricardo
 
Posts: 119
Joined: Sun Apr 10, 2011 7:24 am

Re: A Simple AC-Coupled Balanced Line Reciever

Postby mediatechnology » Tue Nov 27, 2012 6:23 pm

Have you done some severe overload tests?


Well yes and no.
You may have a point with that part in particular. I'm looking for a good bipolar follower with fairly low CM distortion...
A 5532 driven from a low Z source might do just as well.

However: I've popped 48V through 2 uF (from the DC-coupled preamp output phantom on/off and fault tests) into the LM49860 without rail protect diodes and only 100R series resistors (to prevent the previous stage from oscillating with a scope load) and it seemed to do OK. That would seem to be the king of overloads vs a signal overload.

In the interest of safety I am adding bridge to rail protection diodes at the op amp input and 100R current limiting in each input leg since the previous stage slews from 0 to +48V pretty quickly.

I did push a LM4562 beyond 34V and it does not like it.
(The LM49860 is made from a higher voltage process and may be more robust?)

EDIT: One other thing I have in my test circuit that got popped with 48V are 2K feedback resistors in the follower.
Those 2K's are there to provide some post-preamp gain making that stage a 0 to +10 dB INA.
It might be that the latchup is reduced by having some series resistance in the feedback leg.
I recall seeing that in a 5532/34 datasheet I think.
I'll try this as pure followers again and see if I can make it latch-up and why...
http://www.waveulator.com
http://www.ka-electronics.com
http://www.linkedin.com/in/waynekirkwood
“Logic will get you from A to B. Imagination will take you everywhere.”
User avatar
mediatechnology
 
Posts: 1826
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas

Re: A Simple AC-Coupled Balanced Line Reciever

Postby ricardo » Wed Nov 28, 2012 3:57 am

5532 followers are good if the THD is OK for you. In Jurassic times, I couldn't get 20kHz 22dBu dist. off the noise floor with SOTA gear of the day. (Radford LDO + THD set & Sound Tech Intermod set.)

I did manage to kill one but as that killed the series resistors, diode protection bridge & PSU too, I figured that trying to drive a small power amp down its throat was a bit unfair. :twisted:
ricardo
 
Posts: 119
Joined: Sun Apr 10, 2011 7:24 am

Re: A Simple AC-Coupled Balanced Line Reciever

Postby mediatechnology » Wed Nov 28, 2012 4:35 am

The resulting CM distortion from the use of followers in this circuit may cancel in the following differential stage.
Since the input never sees the outside world in the circuit I intend to use it in the drive will always be low Z.

Good that it took a power amp to take out the 5532. :D
http://www.waveulator.com
http://www.ka-electronics.com
http://www.linkedin.com/in/waynekirkwood
“Logic will get you from A to B. Imagination will take you everywhere.”
User avatar
mediatechnology
 
Posts: 1826
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas

Re: A Simple AC-Coupled Balanced Line Reciever

Postby mediatechnology » Wed Nov 28, 2012 6:59 am

I tried to latch-up the LME49860 but couldn't though I did manage to kill one.

LME49860/5532 test circuit.
Diode clamps to rails at + input.
Follower.
Using Rfb =2K or Rfb = 0 no latchup.
No latch-up with 100R in series with 2 uF and a 30V pulse.

No series R or Rfb (both 0R) and a +30V pulse through 2 uF took out the LME49860. RIP LME49860.

It's a pretty extreme test IMHO and the 100R (maybe less?) series R to protect the device doesn't really hurt performance that much since the CM impedance is a MegOhm.

The JRC5532 passed the 30V test with Rin and Rfb = 0. Yeah.
http://www.waveulator.com
http://www.ka-electronics.com
http://www.linkedin.com/in/waynekirkwood
“Logic will get you from A to B. Imagination will take you everywhere.”
User avatar
mediatechnology
 
Posts: 1826
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas

Re: A Simple AC-Coupled Balanced Line Reciever

Postby ricardo » Wed Nov 28, 2012 8:32 pm

mediatechnology wrote:No series R or Rfb (both 0R) and a +30V pulse through 2 uF took out the LME49860. RIP LME49860.
Thanks for this Wayne. But I think we're both going to rot in hell ! Mea maxima culpa.
__________________________

All those JRC/NJM OPAs show the same internal circuit on their datasheets. Its the simple PNP i/p topology which goes back to Raytheon RC4136 and used by a lot of other people too. eg MC33078 and LM833. I can well believe they have this process down to a fine art and now use the same circuit for a wide variety of different apps.

eg their NJM4562 is most certainly NOT LM4562. Though many important performance specs are similar, it doesn't have the latching of LM4562

But I find it hard to believe JRC5532 uses this circuit rather than the tried & tested Philips topology. Comments?
ricardo
 
Posts: 119
Joined: Sun Apr 10, 2011 7:24 am

Re: A Simple AC-Coupled Balanced Line Reciever

Postby mediatechnology » Thu Nov 29, 2012 5:24 am

The ON MC33079 series and the LP counterpart are a nice good-performing PNP input OPA http://www.onsemi.com/pub_link/Collateral/MC33078-D.PDF

But I find it hard to believe JRC5532 uses this circuit rather than the tried & tested Philips topology. Comments?


Are you sure those PNP inputs that you've sometimes seen in 5532/34 datasheets is real?
I had ON Semi correct a 20 year-old schematic error this summer. They and Philips both showed PNP inputs: viewtopic.php?f=12&t=454&start=2
http://www.waveulator.com
http://www.ka-electronics.com
http://www.linkedin.com/in/waynekirkwood
“Logic will get you from A to B. Imagination will take you everywhere.”
User avatar
mediatechnology
 
Posts: 1826
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas

Re: A Simple AC-Coupled Balanced Line Reciever

Postby ricardo » Thu Nov 29, 2012 6:50 am

mediatechnology wrote:Are you sure those PNP inputs that you've sometimes seen in 5532/34 datasheets is real?
I had ON Semi correct a 20 year-old schematic error this summer. They and Philips both showed PNP inputs: viewtopic.php?f=12&t=454&start=2
Philips/TI & loadsa other 5532/4s have NPN i/ps as is right & proper.

I've never had my paws on NJM/JRC 5532/4 so dunno if they use the genuine Philips topology or the PNP i/ps as R4136, MC33079 and their other stuff.

5532/4 have very high input currents so swapping with a opposite polarity input stage might rub a lot of electrolytics the wrong way.
ricardo
 
Posts: 119
Joined: Sun Apr 10, 2011 7:24 am

Re: A Simple AC-Coupled Balanced Line Reciever

Postby mediatechnology » Thu Nov 29, 2012 7:06 am

Philips/TI & loadsa other 5532/4s have NPN i/ps as is right & proper.


Not necessarily. Actually the PNP input error crept into a Philips NE5534 (only) schematic sometime in the 1990's. ON simply copied it.

Image
Philips NE5534 Schematic Error 1994.

The early Signetics and Philips schematics were right. I had to send ON a copy of the TI 5534 schematic to make my point.

JRC does use NPN inputs in it's 5532/5534 and 2114 varient.

Image
http://www.waveulator.com
http://www.ka-electronics.com
http://www.linkedin.com/in/waynekirkwood
“Logic will get you from A to B. Imagination will take you everywhere.”
User avatar
mediatechnology
 
Posts: 1826
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas

Next

Return to Pro Audio Design