A different take on the Cohen/Green mic pre topology

jdbakker

Well-known member
Joined
Nov 24, 2005
Messages
1,431
Reaction score
27
Location
Amsterdam, The Netherlands
diffgreen.PNG

(this is just a concept; phantom, supply decoupling, RFI filtering and the like still need to be added)

The differential op-amp is a TI OPA1632 or similar. This part will keep the common-mode voltage at its output at Vocm, which is grounded in this design. Here also lies the main difference with the familiar Cohen/Green input stage: where the Green will reproduce any common-mode voltage at its input at the output (or Acm = 1 for all settings of the gain network), this configuration should have an Acm of 0.

The trouble with that is that the emitter current of the input transistors will be modulated by any common mode signal present at the input. For perfectly matched transistors this should have no negative impact, but I'm not entirely sure what will happen when the transistors are mismatched.

Thoughts?

JDB
[need to sim this, but can't find any good diff amp models]
 

jdbakker

Well-known member
Joined
Nov 24, 2005
Messages
1,431
Reaction score
27
Location
Amsterdam, The Netherlands
[quote author="mediatechnology"]I
need to sim this, but can't find any good diff amp models
You could probably build it faster but I'm old-fashioned that way.[/quote]
Normally I would just build it, but the OPA1632 is a 150MHz amp in a SOIC package, which is a bit ... challenging to prototype properly.

JDB.
 

Samuel Groner

Well-known member
Joined
Aug 19, 2004
Messages
2,935
Reaction score
74
Location
Zürich, Switzerland
I considered this approach as well but didn't do any further research on it (I didn't even realise the CM modulation problem yet). For simulation or breadboard you could use three ordinary opamps to build a fully differential one.

Samuel
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
19,917
Reaction score
1,235
Location
Hickory, MS
--deleted---

IIRC there might have been a RC across the collectors for stability when followed by a conventional opamp of the period.

JR


EDIT... oops... I misread the schematic...

The resistors at the top of the schematic are driven by the output signal so in fact the feedback loop is closed around the input devices and this should be low distortion like the other similar designs... not like the soundcraft I mentioned.
 

bcarso

Well-known member
Joined
Feb 20, 2005
Messages
4,055
Reaction score
4
Location
San Fernando Valley, Los Angeles
[quote author="JohnRoberts"]This looks pretty close to an old Soundcraft preamp topology. [/quote]

Also a bit like a UREI schematic I saw ages ago.

But utilizing one of these new diff in/out TI amps I think is a good idea---I'll be interested to see where it goes. It should save space at least.
 

jdbakker

Well-known member
Joined
Nov 24, 2005
Messages
1,431
Reaction score
27
Location
Amsterdam, The Netherlands
[quote author="Samuel Groner"]For simulation or breadboard you could use three ordinary opamps to build a fully differential one.[/quote]
How? I am not aware of a three-opamp replacement for a diff in/out amp. (I do know about the 3-opamp instrumentation amplifier, but that's a totally different beast).

I suppose one could take a regular Green pre and connect the input stage feedback resistors to the (differential) output of the pre -- but that places op-amps in op-amps' feedback loops, and that usually leads to instability.

JDB.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
19,917
Reaction score
1,235
Location
Hickory, MS
The data sheet is not very revealing of the internal schematic. It may be a unconventional topology that does not translate directly to several opamps cobbled together, since each opamp will have compensation while the compound structure may only use one such compensation point for the whole circuit.

YMMV

JR
 

Samuel Groner

Well-known member
Joined
Aug 19, 2004
Messages
2,935
Reaction score
74
Location
Zürich, Switzerland
I am not aware of a three-opamp replacement for a diff in/out amp.
From AES preprint 6294 (omit the four feedback resistors for the basic amplifier): differential_opamp.gif

For a practical realisation you'll probably need local compensation for each opamp (the 100 pFs in your schematic provide two third of that allready).

I remembered wrong that I haven't studied the discussed topology any further. In fact I've found a thoroughly simulated, fully discrete implementation on my hard disk. It comprises some sort of dual CM feedback loop which might have solved the collector current modulation problem, but I'll need to look at it in detail again.

Samuel
 

Samuel Groner

Well-known member
Joined
Aug 19, 2004
Messages
2,935
Reaction score
74
Location
Zürich, Switzerland
I quickly checked my design and it didn't solve the modulation problem.

Perhaps it is possible to design CM-driven current sources to the emitters which provide the necessary current to drive the feedback resistors?

Samuel
 

bcarso

Well-known member
Joined
Feb 20, 2005
Messages
4,055
Reaction score
4
Location
San Fernando Valley, Los Angeles
[quote author="JohnRoberts"]The data sheet is not very revealing of the internal schematic. It may be a unconventional topology that does not translate directly to several opamps cobbled together, since each opamp will have compensation while the compound structure may only use one such compensation point for the whole circuit.

YMMV

JR[/quote]

I did something a while back with a separate loop to center the output voltage of a bridge-tied load power amp, which otherwise was lacking a reference (it was a single-supply design). I suspect the considerations for the TI part are similar and the solution comparable.
 

Carl_Huff

Well-known member
Joined
Nov 30, 2004
Messages
145
Reaction score
1
Location
near LA, CA USA
I have been using the THS4131 (cousin to the OPA1632) in audiophile projects for the last year or so. I have found it to be an outstanding product! Nelson Pass licensed his SUSY technology to TI for these products. I know of a high end audio manufacturer that uses them as filters in their 'well reviewed' D/A converter. The boys at Twisted Pear Audio www.twistedpearaudio.com offer some nice PCBs and kits for DIYers that are well done and inexpensive to buy.
 

bcarso

Well-known member
Joined
Feb 20, 2005
Messages
4,055
Reaction score
4
Location
San Fernando Valley, Los Angeles
If you knew SuSy like I know SuSy....

It seemed the last time I looked at this (quite a while ago) the claims about noise* reduction didn't wash. But if TI elected to license it then maybe I was hasty in reaching that conclusion.

Thanks for those links Carl.


EDIT: *that is, random circuit noise, versus supply ripple/noise etc.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
19,917
Reaction score
1,235
Location
Hickory, MS
[quote author="bcarso"]

I did something a while back with a separate loop to center the output voltage of a bridge-tied load power amp, which otherwise was lacking a reference (it was a single-supply design). I suspect the considerations for the TI part are similar and the solution comparable.[/quote]

I recall scratching out a three transistor differential to perform a similar function in level shifting a differential signal sitting up at phantom voltage (DC coupled) down to a dfifferential output referenced to zero

3-put.gif


caveat.. this is a simplified and unproven schematic so might have errors.
but two differential audio signals get level shifted to a different DC level.

JR
 

mikep

Well-known member
Joined
Feb 18, 2006
Messages
450
Reaction score
2
Location
Philadelphia
I stumbled on a slightly different twist of this myself. the emitter current thing is troubling, seems like a source of distortion even without CM signals but Id have to think about that more. you could make sure that no CM signal is present by using an input transformer. better yet, feed the CM signal foreward to the Vocm pin with a buffer. what is the BW of the Vocm pin? not stated in the datasheet.

the 4131 looks like a killer part.

mike
 

jdbakker

Well-known member
Joined
Nov 24, 2005
Messages
1,431
Reaction score
27
Location
Amsterdam, The Netherlands
[quote author="Carl_Huff"]The boys at Twisted Pear Audio www.twistedpearaudio.com offer some nice PCBs and kits for DIYers that are well done and inexpensive to buy.[/quote]
I suspect you are referring to their IVY I/V converter, right? That particular design highlights one of the problems with diff I/O op-amps: you can only set the CM voltage at the output, whereas for a DAC you'd need to set the voltage of the virtual summing node at the input. From a quick glance it looks like they have an input voltage of ~2.5V, whereas the TI PCM179x really wants to see a summing node at 0V for best performance (according to TI engineers and the data sheet).

JDB
[and that's why I finally decided against the OPA1632 as an I/V stage for the PCM1794 in my DAC. With component variations and all I ended up needing an output CM servo to set the input CM voltage, and that was just a bit too icky to fly]
 

jdbakker

Well-known member
Joined
Nov 24, 2005
Messages
1,431
Reaction score
27
Location
Amsterdam, The Netherlands
[quote author="Samuel Groner"]
I am not aware of a three-opamp replacement for a diff in/out amp.
From AES preprint 6294 (omit the four feedback resistors for the basic amplifier): differential_opamp.gif[/quote]
I don't have access to AES papers, but I suspect AES 6294 is essentially the same as this paper, right? In that case the line right below that picture is significant: When tried out in practice, this circuit will most strongly resist any attempt made at stabilising it. This matches my limited experience with this circuit; have you managed to stabilize it?

In any case, it'll probably be less work to dead-bug an OP1632 than to build a workalike from common op-amps.

Thanks,

JDB.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
19,917
Reaction score
1,235
Location
Hickory, MS
[quote author="mikep"]I stumbled on a slightly different twist of this myself. the emitter current thing is troubling, seems like a source of distortion even without CM signals but Id have to think about that more. you could make sure that no CM signal is present by using an input transformer. better yet, feed the CM signal foreward to the Vocm pin with a buffer. what is the BW of the Vocm pin? not stated in the datasheet.

the 4131 looks like a killer part.

mike[/quote]

If I understand your concern. the common variant of these hybrid transistor-opamp designs where you wrap individual opamps around each transistor, will keep the current constant in each emitter and pass any CM voltage through at unity gain to both opamp outputs where it will be cancelled by the following differential stage.

JR
 

Carl_Huff

Well-known member
Joined
Nov 30, 2004
Messages
145
Reaction score
1
Location
near LA, CA USA
jdbaker, in response to your question ...

My experience has been using Twisted Pear's 'Twisted X' driver PCB. At $13.50 it is an inexpensive and convenient way to explore differential circuits. I used it recently as a balanced buffer in front of a TI PGA2320 attenuator when prototyping a recent project. I was very satisfied with the results.
 

Latest posts

Top