Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Terry Demol]

[quote title=danlavry wrote on Wed, 20 October 2004 18:28]

Terry Demol wrote on Wed, 20 October 2004 16:22

danlavry wrote on Mon, 18 October 2004 01:49 Hi Dan, Interesting thread WRT low noise input design. If I may add some relevant info. You had mentioned the addition mechanism of current noise which manifests itself through the source resistor, this only applies to BJT (transistor) input devices. JFETs and tubes have virtually no input current noise. In absolute noise terms high gm JFETs are similar to bipolars WRT noise (0.7 to0.9nV/rt Hz) and there are actually some tubes which are not far behind at about 1.5nV/rtHz although these are all frame grid tubes and microphony can be a real problem. Examples of JFET input pres are the very well regarded Forsell Fetcode pre and Manly SLAM. Cheers, Terry As a rule, JFET technology is pretty noisy (voltage noise), compare to bipolar, and at low source impedance, bi-polars are much better overall. At high impedance FETS may be better, but we are talking about low impeadance application s (mic pre). Take a look at OPAMPs. It is the same problem! You can find a number of bipolar sub 1nVsqrtHz. Not a single FET input like that. If I recall, the pure JFET opamps are at best at 6nVsqrtHz and above, and the BIfets are a bit better. A 6nVsqrtHz compared to a 1nVsqrtHz is a huge difference (15.5dB loss in dynamic range), and at that point you would not even care about current and 250 Ohm source. Simply too noisy for a micpre... BR Dan Lavry

Sorry Dan but this is incorrect. Please re-read my post as I
was referring to high gm JFETs not generic jfets. These
have nothing in common with noisier JFET IP opamps and AFAIK are
not suitable for IC fabrication.

As stated in my post they are specced at between 0.7nV/rtHz
and 1nV/rtHz input noise depending on models.

High gm JFETS have been used for 20 odd years in many high
performance moving coil phono stages. These have to deal with
signal levels much smaller than even low output ribbon mics
although in these applications they are usually paralleled to
get somewhere around 0.3 to 0.5 nV/rtHz (super low noise).
However non paralleled are fine for mic pre IP's to get EIN specs
of around -128dBu or better.

Here are some examples of low noise apps for JFETs:

http://www.royerlabs.com/SF-24.html
http://www.forsselltech.com/fetcode.htm
http://www.audioresearch.com/PH5.html
http://www.stereophile.com/amplificationreviews/935/index2.h tml
http://www.audiophilia.com/hardware/phono1.htm

There are many more.

I have used them myself in all sorts of low level apps including
Ribbon mic pre's.
I am surprised you are not familiar with current low noise JFET
technology as they present some very interesting and specific
advantages for mic pre's but as always, some disadvantages.

They definately sound different to bipolar inputs, and as usual,
some people love 'em ... some don't

Best Regards,

Terry

[FredForssell]

Hey everyone,

Dan, it's good to see you here doing this forum.

Toshiba is still making the 2SK170 JFET with typical noise performance at about 0.95nV/rt Hz.  The actual performance is typically better than that.  They use to make a dual matched pair of this JFET called the 2SK389 but that has recently been discontinued.

As Terry correctly points out, a few of these devices in parallel can result in noise performance that matches that of bipolar devices.

The current noise in JFETs is extremely low and at audio frequencies can be pretty much regarded as zero and is largely caused gate leakage. There is some gate current flowing dynamically as a result of driving the gate capacitance, but that is true for bipolar transistors as well.

EIN figures of -128 to -130 dBu (22-22kHz) are obtainable with proper JFET selection and circuit design.  At least that has been my experience. YMMV.

JFETs vs bipolar transistors? Well, that's a subjective topic so I won't go there.


Cheers,

[danlavry]

"Sorry Dan but this is incorrect. Please re-read my post as I
was referring to high gm JFETs not generic jfets. These
have nothing in common with noisier JFET IP opamps and AFAIK are
not suitable for IC fabrication.

As stated in my post they are specced at between 0.7nV/rtHz
and 1nV/rtHz input noise depending on models.

High gm JFETS have been used for 20 odd years in many high
performance moving coil phono stages. These have to deal with
signal levels much smaller than even low output ribbon mics
although in these applications they are usually paralleled to
get somewhere around 0.3 to 0.5 nV/rtHz (super low noise).
However non paralleled are fine for mic pre IP's to get EIN specs
of around -128dBu or better.

Here are some examples of low noise apps for JFETs:

http://www.royerlabs.com/SF-24.html
http://www.forsselltech.com/fetcode.htm
http://www.audioresearch.com/PH5.html
  http://www.stereophile.com/amplificationreviews/935/index2.h tml
http://www.audiophilia.com/hardware/phono1.htm

There are many more.

I have used them myself in all sorts of low level apps including
Ribbon mic pre's.
I am surprised you are not familiar with current low noise JFET
technology as they present some very interesting and specific
advantages for mic pre's but as always, some disadvantages.

They definately sound different to bipolar inputs, and as usual,
some people love 'em ... some don't

Best Regards,

Terry"

Terry, I know many things about low noise. It is possible to be wrong and I always welcome learning more things. But if you want to tell me about some devices (dual matched pairs), please do so by pointing the:

1. Mame of a device.
2. Manufacturer.

It would be great for me and others to be able to look at some specific data sheets.

I am trying to keep the conversation here technical, so please do not use a product review in audiophile magazines to make a point.

BR
Dan Lavry

[Terry Demol]

danlavry wrote on Thu, 21 October 2004 18:38

Terry"[/i] Terry, I know many things about low noise. It is possible to be wrong and I always welcome learning more things. But if you want to tell me about some devices (dual matched pairs), please do so by pointing the: 1. Mame of a device. 2. Manufacturer. BR Dan Lavry

Hi Dan,

You've got mail... some data sheets for high gm jfets.

It seems as they say down here "you owe me a beer"...
but I'll buy the second round

Best Regards,

Terry

[danlavry]

Terry Demol wrote on Sat, 23 October 2004 02:38

danlavry wrote on Thu, 21 October 2004 18:38 Terry"[/i] Terry, I know many things about low noise. It is possible to be wrong and I always welcome learning more things. But if you want to tell me about some devices (dual matched pairs), please do so by pointing the: 1. Mame of a device. 2. Manufacturer. BR Dan Lavry Hi Dan, You've got mail... some data sheets for high gm jfets. It seems as they say down here "you owe me a beer"... but I'll buy the second round Best Regards, Terry

My comment earlier was:

"there was a Toshiba made JFET, matched pair, and it was operating at about 1nV/srtHz. That is in the neighborhood of a good transistor (noise voltage wise). But they discontinued it!"

So I looked at the data sheets you sent and found 5 devices, same manufacture I mentioned earlier - Toshiba:
3 single JFETS of the same old family with great specs.
2 duals that are speced to be much noisier

But all of them are long discontinued devices. The one I used in the past was 2sk46 (or 2sK146). I know of a very large company that tried to convince Toshiba to make it at some serious quantities. To my knowledge, Toshiba is out of the discrete semi business.

It is possible that one can find some limited stock. I have about 20-30 of them. But it seems to fall under discontinued devices category.    

I'll buy you beer, you can buy me wine

BR
Dan Lavry

[FredForssell]

Dan,  

You said...

"To my knowledge, Toshiba is out of the discrete semi business."

Not so. Toshiba still has well over 2000 discrete small signal transistors, jfets, and mosfet in production.  Here's a link...

http://www.semicon.toshiba.co.jp/openb2b/servlet/websearch/p ip2a9.ws.cGraphicalBrowse-loadPage?classID=C8DBBC6304&im agesOn=true#browse

They also still make really good performance JFET comp pairs like the 2SK170/2SJ74 and 2SK370/2SJ108.  Here's a link...

http://www.semicon.toshiba.co.jp/openb2b/servlet/websearch/p ip2a9.ws.cSearchResult-loadPartDetails?partKey=2SK170

Alas, Toshiba no longer makes the high performance dual JFETs like the 2SK389/2SJ109, only singles.


With regards to the 2SK146, a Texas company is still making a version of that part. They call it the IFN146. Here's a link....

http://www.interfet.com/pdf/DS_IFN146.pdf

Cheers,

[danlavry]

FredForssell wrote on Sat, 23 October 2004 22:04

Dan,   You said... "To my knowledge, Toshiba is out of the discrete semi business." Not so... Cheers,

Thank for the good links. I am definitely very interested especially in that last one, that IFN146 (because it is a dual device). I was sad to see such wonderful devices (2SK146)disappear, and never understood why it happened. Of course it was not the first good part to be discontinued...

When going to the main Toshiba web, one does not see discrete small signal semiconductors listed. I wonder why. I'll check the Toshiba link you mentioned, and certainly look for any of the duals I can find.

That dual IFN146 is of great interest to me. I really liked the 2SK146. I hope the IFN146 is as good. They say "equivalent to". I'll check it out.

If I recall, that old 146 was not a dual in the traditional sense of 2 devices on the same substrate. It was made3 out of 2 separate devices, each in a TO92 package, pressed to each other side by side and held together by some metal can. The device matching was very good. I recall that the sources where at one end, and the drains at the other end. I do not remember if source and drain were interchangeable, or if they made 2 versions of TO92 (one would be Source, gate, drain, the other drain, gate, source).

Thanks again for the good links.  

BR
Dan Lavry  

[Terry Demol]

danlavry wrote on Sun, 24 October 2004 00:16

FredForssell wrote on Sat, 23 October 2004 22:04 Dan,   You said... "To my knowledge, Toshiba is out of the discrete semi business." Not so... Cheers, Thank for the good links. I am definitely very interested especially in that last one, that IFN146 (because it is a dual device). I was sad to see such wonderful devices (2SK146)disappear, and never understood why it happened. Of course it was not the first good part to be discontinued... When going to the main Toshiba web, one does not see discrete small signal semiconductors listed. I wonder why. I'll check the Toshiba link you mentioned, and certainly look for any of the duals I can find. That dual IFN146 is of great interest to me. I really liked the 2SK146. I hope the IFN146 is as good. They say "equivalent to". I'll check it out.

Hi Dan,

They can, at a price, do a slightly better spec than original.  
The only area which these matched jfets fall down slightly is
matching of Vgs which is usually about 5mV.


Cheers,

Terry

[Terry Demol]

FredForssell wrote on Sat, 23 October 2004 22:04

Alas, Toshiba no longer makes the high performance dual JFETs like the 2SK389/2SJ109, only singles. Cheers,

Gedday Fred,

Hope you are well, thanks for taking time to supply all
that info. WRT 389/109 yes bad news, I am wondering where some
of the hi-end companies that use these devices are getting theirs
from, Passlabs comes to mind, they are used in most of their
products. Maybe Nelson has most of the worlds remaining
stock

Also on the discontinued product list are Black Gate capacitors.

Best,

Terry

[FredForssell]

Dan, sorry for using your forum for this but...

Terry, could you drop me an email?  The email address I have for you is bad.

Again, sorry Dan.

Cheers,
Fred Forssell

[Barish]

Brian Roth wrote on Fri, 08 October 2004 06:12

I need to figure out how to do "quotes" on this board... Bri

Hi Brian,

Click on the "quote" button at the bottom right corner of the message you'd like to quote from and the Reply page will open with the message to be quoted automatically inserted in with

[quote title=barish wrote on Fri, 08 October 2004 06:12]quoted message body[/quote]

on either side, which will look like this when submitted:

barish wrote on Fri, 08 October 2004 06:12

quoted message body

If you use the first part of the quote code simply like this:

quoted message body

..then it will look like this:

Quote:

quoted message body

instead of all that "barish wrote on Fri, 08 October 2004 06:12" at the beginning. That's useful if you are replying to the quoted message phrase by phrase.



For a hidden link under a wording:

Click here to go to Lavry Engineering Website

..then it will look like this:

Click here to go to Lavry Engineering Website


To make the wording with a hidden link bold :

Click here to go to Lavry Engineering Website

..then it will look like this:

Click here to go to Lavry Engineering Website



To insert images from another website without uploading them here:

..which will result in having this:





Hope this helps.


M.