Fully differential I/O amplifier

[ojg]

So according to TI, these new fully differential I/O amplifiers, like OPA1632 is the best thing since sliced bread, and when I look at their datasheet I tend to agree.

The deal is this: "Differential output reduces even-order harmonics and minimizes common-mode noise interference."

So I start thinking, why haven't we seen this before? Smart people must have thought of this long ago. And if so why aren't there any discrete power amplifiers out there based on this? Or maybe there is but I haven't found them yet.

Anyone knows? Is it something worth pursuing?

Here is the diagram of the THS4131 amp:

[sss]

Originally posted by ojg
So I start thinking, why haven't we seen this before?

i guess we all have seen this before , its good for balanced inputs etc but thats the only application i can think off

[anatech]

Hi ojg,
Push pull amplifiers (tube & transistor) do this. Tube types have been doing this since before many of us were born. Of course, as sss pointed out, the studio standard balanced out does this too. There are only some situations where this is a benefit, like long cable runs or really noisy (electrically) enviroments.
-Chris

[ojg]

Originally posted by anatech
There are only some situations where this is a benefit, like long cable runs or really noisy (electrically) enviroments.

Why is "reduction of even-order harmonics and minimized common-mode noise interference" only a good thing when you have long cable runs?

Back to my point. These amplifiers seem (to me at least) different from the regular balanced output drivers such as
DRV135 (see attachment)

The THS4131 looks rather normal at the input, a regular diff input stage with folded cascode. The peculiarity is that there are two output buffers in opposite phase. There are two sets of feedback networks one from each buffer to each side of the diff input stage. In addition there is a "servo" to cancel common-mode errors. This is also different from a normal bridged amplifier.

If this has been used in tube circuits before I am interested in seeing a schematic (even though I know nothing about tube design...)

[anatech]

Actually, the even order harmonics are cancelled when the opposite phases are combined ... like in an output transformer of a tube amp. The same occurs in the balanced input which could be a transformer or differential pair.
Otherwise, you have the same distortion spectrum (no cancellation).
A long cable run is an application of the noise cancelling property of a balanced output. I should have made that more clear.
-Chris

[thanh]

This technique probadly can only reduce and not cancel distortion. Its THD is 0.000003%
I love zeroTHD.I think perhaps phase problem cause aliving distortion.Can you understand me? My english is not good!

[Petter]

The basic topology is very similar to Pass X as I have posted here before.

Petter

[x-pro]

I've already posted on this subject a couple of weeks ago:

OPA1632 and THS4131

Cheers

x-pro

[ojg]

x-pro:
I saw your post earlier, but thought the subject was worthy of more discussion!

Petter:
I need to study the X circuit closer, but I thought it used two amplifiers in a "sort of" bridged mode? Similiar but not quite the same? There's been many X threads in the Pass forum but which one discussed the theory behind this?

anatech:
I know the even harmonics are cancelled when you combine the opposite phases. If that is all there is to it then the DRV135 and other bridged amplifiers should have equally impressive specs, but they don't.

Also noteworthy is that the OPA1632 has only 78dB of open-loop gain, very low by audio standards. (OPA134 has 120dB!) So the low distortion is not due to large amounts of NFB.

I have a feeling that the trick is having the feedback from each output buffer return to a single differential input pair, but I can't prove it...

[x-pro]

Originally posted by ojg
Also noteworthy is that the OPA1632 has only 78dB of open-loop gain, very low by audio standards. (OPA134 has 120dB!) So the low distortion is not due to large amounts of NFB.

More important that this 78 dB gain is the same almost over all audio frequency band, but "120 dB of gain" in OPA134 is only valid for very low frequencies- at 10 kHz it has only about 60 dB gain - almost 20 dB lower than THS4131 (or OPA1632)

x-pro

[pmkap]

The specific differences (by ommission) between the THS4131 and the OPA1632 appears to be the common range input offset, +-6v vs +-3V.
This might be a result of a different architectural method of dealing with the Ocm pin,
The THS41x1 uses a rather 'trick' method, that might be absent in the newer OPA.
http://www.diyaudio.com/forums/showt...4435#post14435

[ojg]

Originally posted by x-pro
More important that this 78 dB gain is the same almost over all audio frequency band, but "120 dB of gain" in OPA134 is only valid for very low frequencies- at 10 kHz it has only about 60 dB gain - almost 20 dB lower than THS4131 (or OPA1632)

Indeed!

I think Petter led me onto the right track. I have now read Pass' whitepaper and his patent on Su-Sy and I can see that there are clearly similarities in how the distortion from each output buffer is combined in the input stage. What also contributes to low distortion is the near perfect matching that TI can achieve between the output stage transistors in an IC.

Question is: Are the implementations different enough that they don't infringe on Pass' patent?

[x-pro]

Originally posted by pmkap
The specific differences (by ommission) between the THS4131 and the OPA1632 appears to be the common range input offset, +-6v vs +-3V.
This might be a result of a different architectural method of dealing with the Ocm pin,
The THS41x1 uses a rather 'trick' method, that might be absent in the newer OPA.
http://www.diyaudio.com/forums/showt...4435#post14435

Not offset but maximum rated input differential voltage. I suspect it has nothing to do with the Ocm (which is controlling the output). There is one more notable difference - current noise on OPA1632 is about 3 times lower that that on THS4131. Perhaps they change the input stage a bit to get lower noise and as a result got lower max diff. voltage. Otherwise these two do appear identical... .

x-pro

[Nelson Pass]

Since TI has pointedly ignored my offer of license, my
attorney is evaluating whether it's worth the effort to
chase them or not.

[john curl]

Lots of luck! ;-)

[Jocko Homo]

If anyone is deserving, they are.

Unless it costs too much $$$.

Oh, hell.....do it anyway.

Jocko

[Jocko Homo]

I designed a preamp, back in........'88(?) that looked just like this.

Except it used JFETs in the front-end, and I ran it open loop.

Sounded pretty good............

But since you asked about power amps........

Check out the LM6313, if it is still on National's site.

(Thanks to Charlie Hansen for pointing this out several years after I made the first one..............)

Jocko

[ojg]

Originally posted by Jocko Homo
If anyone is deserving, they are.

Unless it costs too much $$$.

That's the problem with huge companies, they can simply ignore patents from small companies because their army of lawyers know how to drag the case out on technicalities (no pun intended), making a lawsuit far too expensive.

IMO, patents by small companies are only effective against other small companies, such as between hifi manufacturers.

[Dave]

You don't have to look too far to find fully differential amplifiers from other manufacturers,

http://cache.national.com/ds/LM/LM4895.pdf

AD8139

LT6600-20

LTC1992

Would these also infringe on Nelson Pass's Patent? I guess there isn't enough information in the datasheets to say for sure.

[Nelson Pass]

The LT6600-20 and the LTC1992 clearly do not from what I
can see.

The AD8139 and LM4895 might, but it would hinge on the
technicality of folded cascoding, which are not revealed in
the datasheets. They do appear to apply SuperSymmetry
otherwise.

[grataku]

Originally posted by Nelson Pass
Since TI has pointedly ignored my offer of license, my
attorney is evaluating whether it's worth the effort to
chase them or not.

I got the feeling they'd probably start by fighting to throw the patent out and you'd lose, realistically speaking.

[Nelson Pass]

Originally posted by grataku
I got the feeling they'd probably start by fighting to throw the patent out and you'd lose, realistically speaking.

Maybe, Maybe not. More likely they would weigh the costs of
legal action and the possibility that they would be picking up
my costs versus a paltry license fee.

They would also weigh the probability that I would spend the
money, and if they did any research they would discover that
we have spent as much as $500,000 on legal action before -
and prevailed.

Invalidating a patent would be very expensive for them to try,
and I don't think they could recover their costs.

Also, you discount the sympathy that the little guy often gets
from the courts when battling behemoths.

All told though, legal action is unpleasant business, and I
personally haven't ever seen anybody go through it with a
smile, even when they win.

pass:/ likes to win anyway.

[grataku]

Originally posted by Nelson Pass


All told though, legal action is unpleasant business, and I
personally haven't ever seen anybody go through it with a
smile, even when they win.

pass:/ likes to win anyway.

Not as unpleasant as buying a house in NJ. I am closing in a few days and I am starting to wish I had a "family" to help me look after my interest, if you know what I mean...

[Nelson Pass]

Looking to get into waste management, are we?

[dimitri]

Since TI has pointedly ignored my offer of license, my attorney is evaluating whether it's worth the effort to chase them or not.

Dear Nelson, may I ask you

--does USPat 05376899 cover the amplifier with balanced inputs and balanced outputs, as shown?

--You wrote that it will cover two CFB amps connected as bridge http://www.diyaudio.com/forums/attac...&postid=387359
But will your patent cover this circuit if R1=0 (no coupling resistor at all)?
Quote: “An amplifying circuit having identical inverted folded-cascode amplifier stages coupled together through a coupling resistor at the positive terminals of the respective stages”.