990 OpAmps
meej...@my-deja.com
directly?
thanks
jrg
Sent via Deja.com
http://www.deja.com/
BASSMANCP
http://www.mil-media.com
They make an all j-FET direct replacement
that blows the doors off of any other 990's
out there. You can also get the regular ones that John Hardy makes for himself
and Jenson at:
www.jensentransformers.com or at the
John Hardy web sight which I don't
have handy. All are about $75.00
Chris
John W. Hardy
I manufacture the 990 op-amp as designed by Deane Jensen,
and later versions (990A and 990C) as designed by Deane
Jensen and Steve Hogan. These are the op-amps (990C) that I
use in all of my mic preamp products, including the M-1,
M-2, MPC-500C (for MCI 500 series consoles), MPC-600 (for
MCI 600 series consoles), MPC-3000 (for Sony MXP-3000 series
consoles) and Jensen Twin Servo (R) 990 Mic Preamp. The
Jensen Twin Servo was originally made exclusively for Deane
Jensen, who then sold them to his customers. I now sell them
directly to customers.
The 990 can directly replace a 2520 (same dimensions and
pinouts), but each application should always be analyzed for
performance differences, possible circuit modifications,
etc. I have heard of several instances where summing amp
circuits were significantly quieter when a 990 was used in
place of a 2520. The sonic quality will probably be
different, with the 990 being less colored.
Note that Millennia Media offers what they call a "990" (as
mentioned in another message), but it is completely
different from the 990 that Deane Jensen designed. The
Jensen circuit uses bipolar transistors, while the Millennia
Media circuit uses Field Effect transistors, making it a
very different circuit. Frankly, I find it strange that they
call it a "990", but life is strange.
The 990 prices range from $46.95 (original circuit) to
$49.95 (990C) in quantities of 1-24 pieces (NOT "about $75"
as erroneously reported in another message). If you need a
data package or have any questions, please let me know.
Thank you.
John Hardy
The John Hardy Co.
meej...@my-deja.com
together a 990 back in '79 at API but the quiescent current was much too
high. Any specs on the MM? Slew rate?
thx
fred_f...@my-deja.com
opamp design (the MM product is my circuit design) at
www.forsselltech.com. Use the Products link.
The slew rate is > 60v/us.
Fred Forssell
In article <9325tt$dua$1...@nnrp1.deja.com>,
Bill Whitlock
Deane Jensen's patented (but made public domain shortly after issue by
Deane) 990. One of the main reasons the "real" 990 sounds so sweet is that
it has a pair of small inductors in the emitters of the input differential
pair - this is a tremendous advantage which overcomes the otherwise nasty
tradeoffs of stability, noise, and gain-bandwidth in any op-amp design.
Another important feature of the 990 is intentionallyrather high quiescent
current in the output stage. This keeps operation of the output stage Class
A until output currents reach some 15 mA. In applications where the load is
over about 1.5 k-ohm, operation is entirely Class A. The real 990 is
definitely *not* "just another op-amp".
Bill Whitlock
Jensen Transformers, Inc.
<meej...@my-deja.com> wrote in message news:9325tt$dua$1...@nnrp1.deja.com...
Tonebarge
The John Hardy Company
Address: P.O. Box AA631
Evanston; IL; USA
Phone:847-864-8060
Fax:847-864-8076
Regards,
TB
--
"Measure twice, cut once."
Fletcher
> You can check out the specs and some test data for my latest discrete
> opamp design (the MM product is my circuit design) at
> www.forsselltech.com. Use the Products link.
If you're using the new one in the 'CS-1'...it sounds phenomenal!
--
Fletcher
Mercenary Audio
TEL: 508-543-0069
FAX: 508-543-9670
http://www.mercenary.com
"this is not a problem"
David Collins
> fred_f...@my-deja.com wrote:
>
> > You can check out the specs and some test data for my latest discrete
> > opamp design (the MM product is my circuit design) at
> > www.forsselltech.com. Use the Products link.
>
> If you're using the new one in the 'CS-1'...it sounds phenomenal!
It is, and they do.
DC
--
Dave Collins Entropy just isn't what it used to be!
johnl...@my-deja.com
"John W. Hardy" <johnwel...@home.com> wrote:
> Note that Millennia Media offers what they call a "990" (as
> mentioned in another message), but it is completely
> different from the 990 that Deane Jensen designed. The
> Jensen circuit uses bipolar transistors, while the Millennia
> Media circuit uses Field Effect transistors, making it a
> very different circuit. Frankly, I find it strange that they
> call it a "990", but life is strange.
Hi John,
The 1" x 1" op-amp package and pinout has for decades been used by a
number of manufacturers under a variety of names. Some may disagree, but
in recent years I believe the term "990" has become synonymous with this
package -- largely in part to your work and fine repuation. We thought
about calling it an MM-2520 (per API), but went with the "MM-990"
designator because we felt more people were familiar with Deane's
original moniker. Please take this as an unqualified compliment.
John La Grou,
Millennia Music & Media Systems
http://www.mil-media.com
Fletcher
>
> The 1" x 1" op-amp package and pinout has for decades been used by a
> number of manufacturers under a variety of names. Some may disagree, but
> in recent years I believe the term "990" has become synonymous with this
> package -- largely in part to your work and fine repuation. We thought
> about calling it an MM-2520 (per API), but went with the "MM-990"
> designator because we felt more people were familiar with Deane's
> original moniker. Please take this as an unqualified compliment.
>
>
Of course there were only 9,997 other available numbers between 1-10,000
besides "990" and "2520"...so I guess the options were a bit limited...
Greg Kenny
Bill Whitlock wrote:
> First, as John Hardy points out, the MM version is not the same design as
> Deane Jensen's patented (but made public domain shortly after issue by
> Deane) 990.
Is the original Deane Jensen 990 schematic available?
Thanks,
Greg Kenny
Bill Whitlock
"Fletcher" <Flet...@mercenary.com> wrote in message
news:3A574653...@mercenary.com...
> Of course there were only 9,997 other available numbers between 1-10,000
> besides "990" and "2520"...so I guess the options were a bit limited ...
Thanks Fletcher, I couldn't have said it better myself. The 990 was designed
by the late Deane Jensen and was simply packaged so it could retrofit the
2520 modules. If the only thing that's the same about the MM design is the
package, why not 2520? Anyway, the honest-to-goodness original 990 design
by Deane Jensen (which, among other contributions to audio, resulted in
Deane's Lifetime Achievement award from Mix magazine) is made by John Hardy
with our blessings.
Bill Whitlock, president
Jensen Transformers, Inc.
Bill Whitlock
the "990 Paper." It's free, of course. We can't e-mail it or post it at our
website because of copyright considerations.
"Greg Kenny" <g...@storm.ca> wrote in message
news:3A5762D0...@storm.ca...
Luke Kaven
are excellent sounding units. But I'm curious as to why the package should
dictate the model number as opposed to the underlying circuitry. That would
be like calling every opamp in a certain DIP case a "741". Its hard for me
to imagine the "990" referring to anything but Deane Jensen's circuit
design, as published in the original AES paper.
<johnl...@my-deja.com> wrote in message
news:93796a$kgq$1...@nnrp1.deja.com...
BASSMANCP
Their sight has a .pdf document that explains the entire circuit and has the
schematic. Oh Yeah, they have dozens
of other great info and circuits there too. Best it's free!!! Just remember
where to buy
your transformers ;-)
CP
meej...@my-deja.com
place.
Back in the late 70s ('78 I believe), Valley People (see thread)
designed the TransAmp that was alleged to replace the 2520/transformer
combinations and other similar arrangements. In fact, at API, we
designed a console for Val Garay - spec'd by Steve Waldman - that used
transamps in the mic pres and summing busses. If this became popular it
would seriously impact Deane's sales as API consoles used many Jensen
transformers. He gave the schematic to the world in hopes that it would
be able to compete with the transparency of the transamp and still
provide the benefits of transforming coupling. It was a mother of a
circuit the way Deane had spec'd it (remember the ferrite inductors?)
and virtually impossible to "tape" (we made PC boards with black tape
back then!) and be in compliance with Saul Walker's taping specs. As it
turned out, the TransAmp wasn't the threat it might have been and Paul
Buff moved on to the photo flash business.
Scott Dorsey
>Not to detract from the MM opamp (or the Jensen 990 design), both of which
>are excellent sounding units. But I'm curious as to why the package should
>dictate the model number as opposed to the underlying circuitry. That would
>be like calling every opamp in a certain DIP case a "741". Its hard for me
>to imagine the "990" referring to anything but Deane Jensen's circuit
>design, as published in the original AES paper.
That's a good analogy, since for a while there were all kinds of different
op-amps, all with about the same bandwidth but differing in other
characteristics, all called the 741. Then there was the 1741, which wasn't
much better than a 741.
For an even BETTER example, look at the 2N2222. Whatever that is.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Bill Whitlock
(in part):
> Back in the late 70s ('78 I believe), Valley People (see thread)
> designed the TransAmp that was alleged to replace the 2520/transformer
> combinations and other similar arrangements. In fact, at API, we
> designed a console for Val Garay - spec'd by Steve Waldman - that used
> transamps in the mic pres and summing busses. If this became popular it
> would seriously impact Deane's sales as API consoles used many Jensen
> transformers. He gave the schematic to the world in hopes that it would
> be able to compete with the transparency of the transamp and still
> provide the benefits of transforming coupling.
I first met Deane in 1972 and he was working on the 990 circuit design then.
I was an analog circuit designer and he was interviewing me (for hours!) as
his replacement at Quad-Eight. I got the job - he left to start Jensen
Transformers and I became chief engineer at Quad-Eight. He was bubbling with
excitement over discovering *the* patented feature of the 990 (absent in
most of the "knock-offs") - the small inductors in the emitters of the input
pair (I'll spare the engineering details here). Knowing Deane as well as I
did, I don't business considerations were his biggest motivations to give
away the patent, which he did with a magazine article. I think the major
motivation was simply to get a patent to prove he could do it to his father
(a Ph.D. scientist) - winning his father's acceptance was part of Deane's
emotional baggage.
> It was a mother of a
> circuit the way Deane had spec'd it (remember the ferrite inductors?)
But they are essential to the 990.
> As it turned out, the TransAmp wasn't the threat it might have been and
> Paul Buff moved on to the photo flash business.
Ahh ... Allison Research and Paul's ill-fated automation system, too.
Memories ...
meej...@my-deja.com
uncharacteristically bold business move on his part. I think he knew the
990 wasn't yet ready for prime time but released it anyway just to buy
some time or throw out some interference. He was an amazing individual.
I had the priveledge of attending a dinner with another API person at
Trader Vics in L.A. in '79. Very brilliant but very lonely guy then.
I didn't realise he was working on the 990 as early as '72. Didn't that
pre-date the LM394 supermatch by quite a bit or was that not essential
to the design?
As for the baggage, he had considerably more than just needing his
father's approval. His mother (an MD) blamed him for the death of his
twin brother when they were kids. (at least that's what he told me).
High-Mu
theirs?
Tim
In article <9380j8$491$1...@newsmonger.rutgers.edu>,
Kevina25
We still get at least one call a year about memories little helper. However
there are still a number of 65k systems out there working every day. I still
have one of the two 24 chan automated" 550 " clone eqs [apple2 cpu ;-]] we
built along with the endless belt optical faders.
Allison/Valley was afun place just a couple decades ahead of the biz.
kla
Bill Whitlock
<meej...@my-deja.com> wrote in message news:9389s5$d47$1...@nnrp1.deja.com...
(in part):
> I didn't realise he was working on the 990 as early as '72. Didn't that
> pre-date the LM394 supermatch by quite a bit or was that not essential
> to the design?
Not essential.
> As for the baggage, he had considerably more than just needing his
> father's approval. His mother (an MD) blamed him for the death of his
> twin brother when they were kids. (at least that's what he told me).
Oh, yes. His twin, Alan, was mom's favorite because he wasn't as rebellious
as Deane. Alan was killed by a drunk driver when, at age 14, they were both
crossing the street. Deane was a truly wonderful, if a bit eccentric,
person. Many, many fond memories ...
David Satz
> designed the TransAmp that was alleged to replace the 2520/transformer
> combinations and other similar arrangements. In fact, at API, we
> designed a console for Val Garay - spec'd by Steve Waldman - that used
> transamps in the mic pres and summing busses. If this became popular it
> would seriously impact Deane's sales as API consoles used many Jensen
> transformers.
I well remember the TransAmp "LZ" kits, which a lot of classical
engineers I knew tried to use as mike preamp circuits. But they
turned out to have very narrow overload margins for some reason--
I remember having to watch the clipping LEDs very carefully at all
times--and their best noise performance wasn't available with very
low-impedance condenser microphones such as Schoeps. I never
found out why that was or whether anything could be done about
it, and soon the company disappeared.
Scott Dorsey
The front end on these was run damn near into starvation to get the
low noise floor.... I think the collector resistor was something like
1M or so. I have the schematics for these things around, and Hall
Electronics in Charlottesville has a huge pile of the things sitting
around in their closet (and may have the retrofit kits for the MCI
consoles too).
ScotFraser
engineers I knew tried to use as mike preamp circuits. But they
turned out to have very narrow overload margins for some reason--
I remember having to watch the clipping LEDs very carefully at all
times->>
They also tended to be right on the edge of stability & would oscillate at high
gain.
Scott Fraser
John La Grou
<ka...@rci.rutgers.edu> wrote:
>Not to detract from the MM opamp (or the Jensen 990 design), both of which
>are excellent sounding units. But I'm curious as to why the package should
>dictate the model number as opposed to the underlying circuitry. That would
>be like calling every opamp in a certain DIP case a "741". Its hard for me
>to imagine the "990" referring to anything but Deane Jensen's circuit
>design, as published in the original AES paper.
Luke,
Good point. As I explained in my earlier post, I believe the term
"990" has become synonymous with the 1" x 1" package -- largely in
part to John Hardy's fine reputation. People call us looking for a
replacement 1" x 1" opamp that sounds different. They more often than
not ask about a "990-like amplifier" or, less frequently, for a
"2520-like amplifier." They're talking about the "package," not the
circuit. I would think that John and Bill would take this as a
compliment. Looks like I was dead wrong. I apologize to Bill and John
and will immediately change the name if that is their wish.
John La Grou, President
Bob Olhsson
>I didn't realise he was working on the 990 as early as '72.
I always understood the 990 AND the Jensen transformers were conceived
as high-end upgrades for the explosively popular Automated Processes
consoles which many people considered a sonic step backwards from
previous high-end consoles, especially in the mike preamp department.
Many of us felt that the API consoles had the first built-in eq. that
wasn't a joke compared to any outboard unit. I bought a pair of API
550s instead of a pair of EQP-1as for the same price in 1972 because I
couldn't even imagine a studio without Pultecs!
--
Bob Olhsson Audio Mastery Recording Project Design and Consulting
Box 555, Novato CA 94948 Tracking, Mixing and Mastering
415.457.2620 FAX 415.456.1496 Mix Evaluation and Quality Control
38 years of making people sound better than they thought possible!
Jonathan Little
Brown data book (which I turned my lab over looking for) was an op-amp
with the same pin out as the 2520/990. This would of pre dated the
2520. Also much has been said of the 990s small inductors in the
emitters of the input differential as to what makes the 990 what it is.
When ever I have eliminated the lil buggers I have never had a stability
problem or sonic difference.. Not to say theoretically they don't make
any difference, just didn't in my applications. Any one have any
comments on the sonic quality of Millenias opamp verse the, I believe,
newer independent forsell design? Oh yeah, sonic holy grail searching
nerd folks watch out... rumor has it design god (and great guy) Steve
Hogan is working on an updated 990 2001...
jwl
Little Labs
Professional Audio Design, Manufacturing and Consulting
http://www.littlelabs.com/
vox/fax :323.851.6860
Mike Rivers
> I always understood the 990 AND the Jensen transformers were conceived
> as high-end upgrades for the explosively popular Automated Processes
> consoles which many people considered a sonic step backwards from
> previous high-end consoles, especially in the mike preamp department.
Hmmm . . . the very same API mic preamps that people prize so highly
today?
--
I'm really Mike Rivers (mri...@d-and-d.com)
Mark Plancke
They sure sound God like on drums or anything else where you're
looking to add some wack and (good) edginess to the sound.
Mark Plancke
SOUNDTECH RECORDING STUDIOS
Windsor, Ontario, Canada
http://SoundTechRecording.com
I don't know the secret of success, but the secret
of failure is to try to please everybody. --Bill Cosby
fred_f...@my-deja.com
Jona...@littlelabs.com wrote:
> Old timers correct me if I am wrong, but I thought in a very old Burr
> Brown data book (which I turned my lab over looking for) was an op-
amp
> with the same pin out as the 2520/990. This would of pre dated the
> 2520.
The footprint used by John Hardy's 990 and the API products is a
basically the same as that used by many companies discrete opamps
dating back to the very early 1960's and perhaps late, late 1950's.
The first time I encountered it was with some Philbrick discrete opamps
that my friend and mentor Bascom H King had in his parts stash. Also a
company called Modular Audio Products (MAP) made a series of opamps
with this footprint about the mid-1970's. Most of the older opamps had
an additional pin for DC offset. This pin was on the output/power side
of the footprint making that side a 5 pin configuration unlike the 990
(opps!), er, the other type of footprint used by Hardy and API (which
has 4 pins set to one side due to the missing DC offset pin). (Egad! I
must have become an oldtimer when I wasn't looking!)
> Also much has been said of the 990s small inductors in the
> emitters of the input differential as to what makes the 990 what it
is.
> When ever I have eliminated the lil buggers I have never had a
stability
> problem or sonic difference.. Not to say theoretically they don't
make
> any difference, just didn't in my applications.
Those inductors didn't make any difference in the experiements I did
back in the late 1970's with Deane's design either. But I thought they
were a cool idea back then. Actually, I think Deane's idea of an
output isolator (R with L in parallel) was/is a WAY cool idea,
especially when an opamp has to drive an output transformer.
Fred Forssell
Forssell Technologies
www.forsselltech.com
Scott Dorsey
>
>The footprint used by John Hardy's 990 and the API products is a
>basically the same as that used by many companies discrete opamps
>dating back to the very early 1960's and perhaps late, late 1950's.
>The first time I encountered it was with some Philbrick discrete opamps
>that my friend and mentor Bascom H King had in his parts stash. Also a
>company called Modular Audio Products (MAP) made a series of opamps
>with this footprint about the mid-1970's. Most of the older opamps had
>an additional pin for DC offset. This pin was on the output/power side
>of the footprint making that side a 5 pin configuration unlike the 990
>(opps!), er, the other type of footprint used by Hardy and API (which
>has 4 pins set to one side due to the missing DC offset pin). (Egad! I
>must have become an oldtimer when I wasn't looking!)
MAP is still around and still has some of those things in stock. They
are now mostly in the business of making power supplies, but they still
have some of the old audio gear in the closet.
Then again, I'm still buying Opamp Labs stuff now and then. It sounds
good, in spite of a really lousy Ft.
>Those inductors didn't make any difference in the experiements I did
>back in the late 1970's with Deane's design either. But I thought they
>were a cool idea back then. Actually, I think Deane's idea of an
>output isolator (R with L in parallel) was/is a WAY cool idea,
>especially when an opamp has to drive an output transformer.
This is a common trick among the RF guys! The ARRL Handbook used to
talk about it.
Monte P McGuire
>engineers I knew tried to use as mike preamp circuits. But they
>turned out to have very narrow overload margins for some reason--
Probably because the app note circuit uses a single op amp to do the
common mode rejection stage that follows the TA-LZ, which is also the
output stage. Running this stage off of +-15V, you will barely get
+20dBV before it clips.
Of course, one could use two such stages in parallel to make a push
pull output and/or use higher supply rails, but that wasn't spelled
out in the app notes.
>and their best noise performance wasn't available with very
>low-impedance condenser microphones such as Schoeps.
That is somewhat odd, but it is possible that the circuit only has a
low equivalent input noise spec when it's run at very high gains.
After all, it's basically just a diff pair with a TL072 doing feedback
around it, and if you run the diff pair at low gains, the TL07x stage
dominates the noise performance.
Personally, I think the topology is a good one, but it can be
implemented poorly. The Neotek / Sytek design is almost identical to
a classic TransAmp, and people seem to like those. The most clever
part about the TransAmp design is that with a single gain pot, both
the first stage and second stage gains vary so that the output
bandwidth remains constant. This makes it easy to adjust gain over a
wide range without stability problems.
BTW, to the other poster who complained about instability at high
gains, this circuit is least stable at _low_ gains! Again, this
circuit is not easy to implement, but when it's done right, it does
sound good.
Regards,
Monte McGuire
mcg...@world.std.com
Monte P McGuire
>output isolator (R with L in parallel) was/is a WAY cool idea,
>especially when an opamp has to drive an output transformer.
I never quite liked that solution, and usually prefer a simple
resistor of 50 or 100 ohms. My theory is that it's better to take the
rolloff caused by large cable capacitance than to use the inductor to
provide a low Z at audio frequencies and ask the amp to provide more
and more drive current at HF.
If you have to drive a transformer, put the transformer before the
resistors so the transformer sees the low impedance of the op amp but
the whole shebang is isolated from the cable. I suppose the
transformer's stray capacitance is not isolated from the amp, but the
cable capacitance still is and that's the only unknown quantity.
Again, I have no real reason why I prefer a simple resistor, only a
theory, but it's always sounded better to me than an LR isolator.
Regards,
Monte McGuire
mcg...@world.std.com
Scott Dorsey
>In article <93cqcu$pf9$1...@nnrp1.deja.com>, <fred_f...@my-deja.com> wrote:
>>Actually, I think Deane's idea of an
>>output isolator (R with L in parallel) was/is a WAY cool idea,
>>especially when an opamp has to drive an output transformer.
>
>I never quite liked that solution, and usually prefer a simple
>resistor of 50 or 100 ohms. My theory is that it's better to take the
>rolloff caused by large cable capacitance than to use the inductor to
>provide a low Z at audio frequencies and ask the amp to provide more
>and more drive current at HF.
On the whole, I'd rather provide the transformer with as low a source
impedance as possible across the spectrum. And I don't mind adding a
little push-pull stage with a couple transistors on the output of the
op-amp (but inside the feedback loop) to improve the drive current.
It's not a lot of board space.
fred_f...@my-deja.com
mcg...@world.std.com (Monte P McGuire) wrote:
> I never quite liked that solution, and usually prefer a simple
> resistor of 50 or 100 ohms. My theory is that it's better to take the
> rolloff caused by large cable capacitance than to use the inductor to
> provide a low Z at audio frequencies and ask the amp to provide more
> and more drive current at HF.
>
Right. I agree that is the best approach when you are not driving into
an output transformer. But if you use that approach to drive into a
transformer the THD goes way up. I know that you know that.
> If you have to drive a transformer, put the transformer before the
> resistors so the transformer sees the low impedance of the op amp but
> the whole shebang is isolated from the cable. I suppose the
> transformer's stray capacitance is not isolated from the amp, but the
> cable capacitance still is and that's the only unknown quantity.
>
You often can't do that with a lot of opamps without some stability
issues. That is why I think that the approach that Deane and others
used/use is a good one... when you are driving a transformer.
> Again, I have no real reason why I prefer a simple resistor, only a
> theory, but it's always sounded better to me than an LR isolator.
>
I'd probably agree with on the sonic point, but I don't usually use
output transformers in my designs and therefore have not done the
listening tests that you have. If I get around to it, I'll give it a
try myself and let you know what I think about the sonic impact, not
that my opinion matters to anyone but me.
Fred Forssell
Techmeister
Doesn't the FET input work AGAINST you in low-Z situations like a
summing buss? Higher noise, more offset and drift??
The TLM series were always noisier as summers, mainly from impedance
issues as I recall.
db
PS: ditch the "my-deja" address, LaGrou, they never work...
In article <93796a$kgq$1...@nnrp1.deja.com>, johnl...@my-deja.com wrote:
> Hi John,
>
> The 1" x 1" op-amp package and pinout has for decades been used by a
> number of manufacturers under a variety of names. Some may disagree, but
> in recent years I believe the term "990" has become synonymous with this
> package -- largely in part to your work and fine repuation. We thought
> about calling it an MM-2520 (per API), but went with the "MM-990"
> designator because we felt more people were familiar with Deane's
> original moniker. Please take this as an unqualified compliment.
>
> John La Grou,
> Millennia Music & Media Systems
> http://www.mil-media.com
>
>
> Sent via Deja.com
> http://www.deja.com/
--
David 'db' Butler, Consultant
Acoustics by db
"...all the rest are just brokers"
now on the web at http://www.db-engineering.com
Boston, Mass Phone 617 969-0585 Fax 617 964-1590
Monte P McGuire
>Monte McGuire <mcg...@world.std.com> wrote:
>> If you have to drive a transformer, put the transformer before the
>> resistors so the transformer sees the low impedance of the op amp but
>> the whole shebang is isolated from the cable. I suppose the
>> transformer's stray capacitance is not isolated from the amp, but the
>> cable capacitance still is and that's the only unknown quantity.
>
>You often can't do that with a lot of opamps without some stability
>issues. That is why I think that the approach that Deane and others
>used/use is a good one... when you are driving a transformer.
You know, I don't have a whole lot of experience driving transformers
with op amps directly, without any isolation, so I guess I have to
play with this some and see how much of a problem it really is!
The only box that I have that will get an O/P transformer is a hacked
up 4 transistor McCurdy mike amp that has relatively little overall
feedback, so I don't think it's going to complain much about a
transformer load.
OTOH, I have been putting API 2503 transformers in series with some
otherwise clean op amp output mike amps to get more dirt, and in that
case, the extra distortion you get from the series resistors is
actually part of the desired effect! Heresy, I suppose, but iron
distortion can be fun sometimes, in small doses. The thing that
isolators seem to do for me is add some sort of harshness, and I don't
get that with a resistor, whether there's a transformer following it
or not.
>> Again, I have no real reason why I prefer a simple resistor, only a
>> theory, but it's always sounded better to me than an LR isolator.
>>
>I'd probably agree with on the sonic point, but I don't usually use
>output transformers in my designs and therefore have not done the
>listening tests that you have. If I get around to it, I'll give it a
>try myself and let you know what I think about the sonic impact, not
>that my opinion matters to anyone but me.
Perhaps, but it's so much fun to compare opinions! Sometimes, it
points to some good stuff or to some avenues that haven't been
explored.
Thanks for the reply,
Monte McGuire
mcg...@world.std.com
Monte P McGuire
Techmeister <dbt...@icnt.net> wrote:
>Doesn't the FET input work AGAINST you in low-Z situations like a
>summing buss? Higher noise, more offset and drift??
PMFJIH...
Some JFETs have very large geometries and high transconductance, and
their voltage noise can be well under what's commonly required for low
impedance (50-150 ohm) sources. There are some Toshiba JFETs such as
the 2SK170 and 2SK370 that have voltage noises around 1nV/root Hz, and
that's very low, about as noisy as 50 ohms or so. What's more, JFETs
have a lot less current noise, so they keep their low noise
performance with high source impedances, whereas large geometry
bipolars get very noisy under the same conditions.
The big problem is that almost none of these JFETs were ever made by
US semiconductor companies, so all you can generally get over here are
the standard, noisy, small geometry parts from Temic (Siliconix) and
Motorola.
But, they do exist and have some good advantages noise wise.
Regards,
Monte McGuire
mcg...@world.std.com
John La Grou
wrote:
>John,
>
>Doesn't the FET input work AGAINST you in low-Z situations like a
>summing buss? Higher noise, more offset and drift??
>
>The TLM series were always noisier as summers, mainly from impedance
>issues as I recall.
>
>db
>
>PS: ditch the "my-deja" address, LaGrou, they never work...
First, I need to make an announcement. We have changed the name of the
MM-990. With profound apologies to Bill and John, it henceforth will
be known as the MM-99. While I still believe that "990" has become
somewhat of a standard designator for the 1" x 1" class of
pin-compatible amplifiers, I fully understand their concerns.
Anyway, mixamp requirements are (generally) not that much different
from any high performance audio requirement. Specifically look for a
device with high gain-bandwidth and low input current noise. After
that, selecting proper source and feedback resistance values is the
real key to optimizing summing noise performance. The MM-990 (now
MM-99) has always worked very well in summing amp applications.
Note that the summing amp inside our Mixing Suite is a slight
modification on the HV-3 front-end -- a discrete-hybrid bipolar
amplifier based on the Philips Microelectronics "Double Balanced"
circuit.
As for the "deja.com" e-mail address, I know the hassle.
Unfortunately, my IP's newsfeeder has recently been unreliable, so
I've been feeding at deja.com. Those wishing to contact me via e-mail
can use:
Thanks,
John La Grou
Millennia Media, Inc.
http://www.mil-media.com
Bill Whitlock
It is faithful to Deane Jensen's paper and patent.
Bill Whitlock
Jensen Transformers, Inc.
"High-Mu" <hig...@my-deja.com> wrote in message
news:938apv$dst$1...@nnrp1.deja.com...
Bill Whitlock
<fred_f...@my-deja.com> wrote in message
news:93cqcu$pf9$1...@nnrp1.deja.com... (in part):
> In article <3A5983DC...@littlelabs.com>,
> Jona...@littlelabs.com wrote:
> > Also much has been said of the 990s small inductors in the
> > emitters of the input differential as to what makes the 990 what it
> is.
> > When ever I have eliminated the lil buggers I have never had a
> stability
> > problem or sonic difference.. Not to say theoretically they don't
> make
> > any difference, just didn't in my applications.
>
> Those inductors didn't make any difference in the experiements I did
> back in the late 1970's with Deane's design either. But I thought they
> were a cool idea back then.
Deane's patent 4,287,479 (viewable at www.delphion.com) relates specifically
to the "reactive networks" in the emitters of the input differential pair.
This novel idea significantly eases the complex tradeoff issues in achieving
unconditional stability in operational amplifier designs. The parameters
which must be traded are stability (slope of open-loop gain rolloff), slew
rate, small-signal bandwidth, and noise. The fact that anecdotal evidence
sometimes indicates that they might be eliminated and not "make any
difference" certainly does not mean they serve no useful purpose. I suggest
a reading of one of the excellent books in the Burr-Brown series for those
who want to get some understanding of the complexities of proper
(unconditionally stable) op-amp design techniques. Source impedance (into
the + input) and load capacitance are just two potential de-stabilizing
factors a good, general purpose op-amp must be designed to cope with.
> Actually, I think Deane's idea of an
> output isolator (R with L in parallel) was/is a WAY cool idea,
> especially when an opamp has to drive an output transformer.
I agree it's a good idea but, to be fair, it did not originate with Deane.
The technique has been known (and used in power amplifiers, for example) for
many, many years. However, Deane probably did popularize the idea for
transformer and line driver applications.
Bill Whitlock
"Monte P McGuire" <mcg...@world.std.com> wrote in message
news:G6ux4...@world.std.com... (in part):
> I never quite liked that solution, and usually prefer a simple
> resistor of 50 or 100 ohms. My theory is that it's better to take the
> rolloff caused by large cable capacitance than to use the inductor to
> provide a low Z at audio frequencies and ask the amp to provide more
> and more drive current at HF.
The current required to drive cable capacitance will be higher for the L/R
isolator only at frequencies where the R only isolator would be causing
magnitude response to drop anyway. But this is *much* less of an issue with
the 990 because of its large drive current capabilities - some 200 mA!
Agreed, wimpy op-amps who can barely output 25 mA will complain (by current
limiting and distorting cymbal crashes, for example) when used with R/L
isolators. But they'll probably complain even if used with just low-value
resistors. Best advice is just don't use ordinary op-amps for cables over
about 100 to 200 feet!
> If you have to drive a transformer, put the transformer before the
> resistors so the transformer sees the low impedance of the op amp but
> the whole shebang is isolated from the cable. I suppose the
> transformer's stray capacitance is not isolated from the amp, but the
> cable capacitance still is and that's the only unknown quantity.
Yes. Always a good idea to drive an output transformer with a source
impedance significantly less than its primary DCR (typically 40 to 50 ohms
for Jensen types) to keep distortion low and maximize headroom. The parallel
capacitance is fairly low (hundreds of pF) but remember *all* the load
capacitance is reflected back to the primary (driver electronics) by a 1:1
output transformer.
Bill Whitlock
> OTOH, I have been putting API 2503 transformers in series with some
> otherwise clean op amp output mike amps to get more dirt, and in that
> case, the extra distortion you get from the series resistors is
> actually part of the desired effect! Heresy, I suppose, but iron
> distortion can be fun sometimes, in small doses.
Just to nit-pick: even the cheapest audio transformers are made with silicon
steel cores --- the better ones use 50%, or better yet, 80% nickel cores.
The steel ones will have the highest distortion and be most sensitive to
source impedance effects. If you really want to capture the "old-time"
transformer sound, use a steel core, drive from a high source impedance, and
put some DC in the winding, too. This will produce lots of even order
(mostly 2nd and 4th harmonic) distortion. We try to avoid all of these
things at Jensen, but if effects are what you're after, this may help.
> The thing that
> isolators seem to do for me is add some sort of harshness, and I don't
> get that with a resistor, whether there's a transformer following it
> or not.
Actually isolators must be tailored to specific op-amps to prevent
instability. For example, our JT-OLI-2 (specs at
www.jensen-transformers.com) is specifically designed for the 990. It likely
will not isolate capacitive loads sufficiently for other op-amps. What you
may be hearing is "harshness" caused by bursts of MHz oscillation caused by
instability in the op-amp.
Techmeister
mystery out.
Techmeister
mcg...@world.std.com (Monte P
McGuire) wrote:
--
fred_f...@my-deja.com
mcg...@world.std.com (Monte P McGuire) wrote:
>
> Some JFETs have very large geometries and high transconductance, and
> their voltage noise can be well under what's commonly required for low
> impedance (50-150 ohm) sources. There are some Toshiba JFETs such as
> the 2SK170 and 2SK370 that have voltage noises around 1nV/root Hz, and
> that's very low, about as noisy as 50 ohms or so. What's more, JFETs
> have a lot less current noise, so they keep their low noise
> performance with high source impedances, whereas large geometry
> bipolars get very noisy under the same conditions.
>
You are right on Monte. In fact if I rememeber correctly, at higher
source Z (about 10k) the 2SK170 has a lower noise figure than the LM394
does at any source Z. But it isn't often that audio circuit designers
find this type of operating condition, so that fact is only of passing
interest. The truth is that the LM394 will be quiter than the K170
type JFETs in most of the commonly encountered audio operating
conditions, but not by that much. IMO there is such a huge sonic
difference between the 2SK170 type JFETs and the LM394, that I'll take
the 1 dB or so greater noise figure of the 170 every time. It sounds
great.
Also the source Z on the mix buss of many large format mixers is quite
high compared to other points in the console. Source impedances of 400-
800 ohms (looking back into the mix buss) is not unusual. At that
impedance the noise difference between bipolar and good JFET devices is
often starting to become very small.
BTW, all of the older JFET opamps that I have designed use the
2SK170/2SJ74 JFETs. You can see a schematic of one of my JFET opamps on
my website (if you are interested in that kind of thing) at
www.forsselltech.com. My newer design still uses them on the front-end
and as current regulators, but it also uses a different JFET family as
well.
> The big problem is that almost none of these JFETs were ever made by
> US semiconductor companies, so all you can generally get over here are
> the standard, noisy, small geometry parts from Temic (Siliconix) and
> Motorola.
>
Again, right on Monte. Toshiba is the only company that I know of that
is making really, really good large geometry JFETs. I hope that they
come out with some of these great JFET dies in a SMT package one of
these days.
> But, they do exist and have some good advantages noise wise.
>
And they sound GREAT, IMO!
Fred Forssell
Forssell Technologies
fr...@forsselltech.com
Scott Dorsey
Monte says:
>
>> The big problem is that almost none of these JFETs were ever made by
>> US semiconductor companies, so all you can generally get over here are
>> the standard, noisy, small geometry parts from Temic (Siliconix) and
>> Motorola.
>
>Again, right on Monte. Toshiba is the only company that I know of that
>is making really, really good large geometry JFETs. I hope that they
>come out with some of these great JFET dies in a SMT package one of
>these days.
>
>> But, they do exist and have some good advantages noise wise.
This is not a problem, because B&D Electronics has pretty much all of the
standard Japanese semiconductors available at very good prices. And the
Japanese aren't discontinuing discretes right and left like the major
American manufacturers are. I just found Motorola discontinued my favorite
high voltage bipolars this week too.
>And they sound GREAT, IMO!
They sound great, they don't cost much, and they are easy to get.
Bill Whitlock
<fred_f...@my-deja.com> wrote in message
news:93i3nu$1f2$1...@nnrp1.deja.com... (in part):
> You are right on Monte. In fact if I rememeber correctly, at higher
> source Z (about 10k) the 2SK170 has a lower noise figure than the LM394
> does at any source Z. But it isn't often that audio circuit designers
> find this type of operating condition, so that fact is only of passing
> interest. The truth is that the LM394 will be quiter than the K170
> type JFETs in most of the commonly encountered audio operating
> conditions, but not by that much.
The fact that bipolar transistors, in large parallel arrays (i.e., LM394) or
operated at high collector currents are well-suited (have a good noise
figure or NF) for low-impedance sources been well-known among engineers for
40 years. It's one of the first decisions designers of low-noise amplifiers
have to make - specify a device whose noise characteristics are optimum for
the impedance of the signal source (see Motchenbacher & Fitchen's "Low Noise
Electronic Design" for example).
> IMO there is such a huge sonic
> difference between the 2SK170 type JFETs and the LM394, that I'll take
> the 1 dB or so greater noise figure of the 170 every time. It sounds
> great.
IMHO, this generalization is FAR too sweeping! There are other circuit
topology considerations that affect "sonic differences" much, much more the
choice of a bi-polar transistor (BJT) or field-effect transistor (FET) as
amplification device.
meej...@my-deja.com
wide threads! Took me back over 20 years to when I was into serious
audio design engineering. Great stuff..ah, nostalgia
>>sniff<<
Talk amongst
yourselves, I'm getting
verclempt...
jrg
fred_f...@my-deja.com
In article <F_276.1941$eT6.1...@paloalto-snr1.gtei.net>,
My response to Monte was directed specifically at the comparison of NF
figures between the 2SK170 and the LM394 in typical and non-typical
operating condiditions, not at the more general discussion of paralled
devices, which is well known and well documented, and which I didn't
address at all. Paralleling 2SK170 type devices works well for low
noise designs just as it does with bipolars designs. On the negative
side of the ledger, you'll get higher input capacitance with paralleled
JFETs. On the positive side, JFET devices don't make good RF detectors
like bipolar devices, something that may be of considerable sonic
importance in today's working enviroments. (Here is where you get to
insert your "transformers will take care of RFI" comments.)
>
> > IMO there is such a huge sonic
> > difference between the 2SK170 type JFETs and the LM394, that I'll
take
> > the 1 dB or so greater noise figure of the 170 every time. It sounds
> > great.
>
> IMHO, this generalization is FAR too sweeping! There are other
circuit
> topology considerations that affect "sonic differences" much, much
more the
> choice of a bi-polar transistor (BJT) or field-effect transistor
(FET) as
> amplification device.
>
My own experiments with JFET vs Bipolar device dates back to the early
1970's. We subjectively compared (listened to) designs with static
topologies while trying different types of devices in various
positions. We tried several different JFET and bipolar devices back
then, and I am still doing this today with newer devices as I become
aware of them and have the time. Devices like the old Crystalonics C413
(later the 2N6550), the 2SK117, 2SK147, and now 2SK170/K389 tended to
sound more neutral, natural, and relaxed when compared to the same
topology using bipolars in a particular position. We even used some
early V-Channel MOS devices as output devices, but gradually went back
to bipolars because matched P and N V-Channel MOS devices didn't exist
back then (or now?), and that limited the possible design topologies.
So generally, the result of my personal listening (and therefore
subjective)evaluation tests was (and still is) that large geometry JFET
devices sound better (to me) than any of the bipolar devices that I
have tried. I was NEVER happy with the sound of the LM-394 device in
any design in which I tested it, even though it usually measured well.
I preferred the MAT02 to the LM394 sonically. That was my experience
back then and is still my experience today. YMMV.
Monte P McGuire
>2SK170/2SJ74 JFETs. You can see a schematic of one of my JFET opamps on
>my website (if you are interested in that kind of thing) at
>www.forsselltech.com. My newer design still uses them on the front-end
>and as current regulators, but it also uses a different JFET family as
>well.
Thanks for all the interesting information... now if I can only find
the time to try out some of these neat things!!
I poked around on the net and www.bdent.com is the site for B&D
Enterprises, the people who carry Toshiba JFETs and other Japanese
semiconductors. They're out of stock for some devices, but they do
have others in stock and the prices aren't too scary.
Overall, I've been really impressed by modern Japanese discretes. You
can get some pretty amazing transistors, such as some high current
super beta transistors that are quite linear and fast at low cost. It
seems like the Japanese kept advancing their discrete devices well
after the late 60s, unlike us over here. You know what else is scary?
I saw an old Cramer catalog from 1963 and the epitaxial planar
Motorola 2N2222 was listed there, albeit for $6.75 each, but it was
available. It's amazing how old some of these devices are, and
unfortunately, how 'current' they are at the same time.
Have fun,
Monte McGuire
mcg...@world.std.com