J308 for really-low voltage oscillator

[John Larkin]

On Fri, 25 Mar 2016 13:00:59 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/24/2016 06:45 PM, John Larkin wrote:
On Thu, 24 Mar 2016 14:49:02 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/23/2016 07:34 PM, mrdarrett@gmail.com wrote:
On Wednesday, March 23, 2016 at 4:31:35 PM UTC-7, mrda...@gmail.com wrote:
On Wednesday, March 23, 2016 at 4:07:25 PM UTC-7, John Larkin wrote:

...

A phemt might work better than a jfet. Transconductance can be really
high, 10x a jfet maybe.

http://www.digikey.com/product-detail/en/cel/NE3508M04-A/NE3508M04-A-ND/949555


Ooh, Digikey is out (non-stock). What is a Phemt?


Oh, never mind, I found it! It really is a thing!

https://en.wikipedia.org/wiki/High-electron-mobility_transistor

Michael


pHEMTs are amazing parts for some things, but I'm pretty sure that
John's pulling your leg about using one in a Joule Thief. The slowest
pHEMTs are about 10 GHz iirc. They also have seriously crummy Early
voltages, so they might not work as well as a JFET when V_DS gets down
to the hundreds of millivolts.

I was serious. Rds-on is about 6 ohms for that NEC part at zero gate
voltage, and transconductance is outrageous compared to a jfet or a
depletion mosfet. Just what you want to power some oscillator thing
from a thermocouple or some such millivolt source.

Okay, my mistake. (That "lunatic fringe" thing again.)

The transconductance is nice and high at fixed bias, true. However, the
Avago parts have the wimpiest drain impedance I've ever seen--something
like 160 ohms. You can't even use them as source followers unless you
bootstrap the drain. The Skyworks parts (SKY65050) are quite a bit
better, but still not even as good as a JFET. I haven't tried the NEC
parts.

It would be entertaining to try using a 14-GHz part on a white plastic
breadboard. OTOH they're surprisingly stable for such fast and quiet
devices--a lot more stable than the SiGe parts!

Cheers

Phil Hobbs

Phemts have some potential as low-frequency amps and switches. I don't
know what is the record for making an oscillator powered by
millivolts, but a phemt is a candidate to do that mostly useless
thing. I've seen the claim of 5 mV, with paralleled jfets.

They are very stable as compared to bipolars. Drain-gate capacitances
are absurdly low, and the substrate is the source. A hacked copperclad
breadboard usually works fine.

I have some DC measurements of the NE3508 if anybody's interested.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[M Philbrook]

In article <nd4759$bpo$1@dont-email.me>, nojunk@knology.net says...

On 3/23/2016 6:07 PM, John Larkin wrote:
On Wed, 23 Mar 2016 14:36:49 -0700 (PDT), mrdarrett@gmail.com wrote:

Hey,

I am looking for a J308, but it looks like it's obsolete.

http://www.mouser.com/ProductDetail/InterFET/J308/?qs=OxRSArmBDfwrWzaWx947ZA%3D%3D

Any ideas?

I'm trying to build a low voltage oscillator that works down to about 0.15 V, basically these:

http://www.google.com/patents/US4734658

http://web.archive.org/web/20150316093734/http://neazoi.gr/energy-harvesting/index.htm

Thanks!

Michael

I've posted this before, author says the best iteration will start
oscillating at 7mV 1.1Ma. He is not drawing power from the circuit, so...
Oops, he got it down to 5.5Mv 0.41 Ma, he moved the scope to a lower
impedance point, it was loading his circuit.

http://www.dicks-website.eu/fetosc/enindex.htm

Mikek

You do mean
5.5mV 0.41mA ?
Us old guys get confused..

Jamie

 

On Saturday, March 26, 2016 at 10:11:33 AM UTC-7, John Larkin wrote:

On Fri, 25 Mar 2016 13:00:59 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/24/2016 06:45 PM, John Larkin wrote:
On Thu, 24 Mar 2016 14:49:02 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/23/2016 07:34 PM, mrdarrett@gmail.com wrote:
On Wednesday, March 23, 2016 at 4:31:35 PM UTC-7, mrda...@gmail.com wrote:
On Wednesday, March 23, 2016 at 4:07:25 PM UTC-7, John Larkin wrote:

...

A phemt might work better than a jfet. Transconductance can be really
high, 10x a jfet maybe.

http://www.digikey.com/product-detail/en/cel/NE3508M04-A/NE3508M04-A-ND/949555


Ooh, Digikey is out (non-stock). What is a Phemt?


Oh, never mind, I found it! It really is a thing!

https://en.wikipedia.org/wiki/High-electron-mobility_transistor

Michael


pHEMTs are amazing parts for some things, but I'm pretty sure that
John's pulling your leg about using one in a Joule Thief. The slowest
pHEMTs are about 10 GHz iirc. They also have seriously crummy Early
voltages, so they might not work as well as a JFET when V_DS gets down
to the hundreds of millivolts.

I was serious. Rds-on is about 6 ohms for that NEC part at zero gate
voltage, and transconductance is outrageous compared to a jfet or a
depletion mosfet. Just what you want to power some oscillator thing
from a thermocouple or some such millivolt source.

Okay, my mistake. (That "lunatic fringe" thing again.)

The transconductance is nice and high at fixed bias, true. However, the
Avago parts have the wimpiest drain impedance I've ever seen--something
like 160 ohms. You can't even use them as source followers unless you
bootstrap the drain. The Skyworks parts (SKY65050) are quite a bit
better, but still not even as good as a JFET. I haven't tried the NEC
parts.

It would be entertaining to try using a 14-GHz part on a white plastic
breadboard. OTOH they're surprisingly stable for such fast and quiet
devices--a lot more stable than the SiGe parts!

Cheers

Phil Hobbs

Phemts have some potential as low-frequency amps and switches. I don't
know what is the record for making an oscillator powered by
millivolts, but a phemt is a candidate to do that mostly useless
thing. I've seen the claim of 5 mV, with paralleled jfets.

Neato. Is that what boosts the mV from the thermocouple in a gas water heater, to control the temperature?

They are very stable as compared to bipolars. Drain-gate capacitances
are absurdly low, and the substrate is the source. A hacked copperclad
breadboard usually works fine.

"... very stable as compared to bipolars." That would make a funny mental health joke.

I have some DC measurements of the NE3508 if anybody's interested.

Sure!

--

John Larkin Highland Technology, Inc

lunatic fringe electronics

Michael

 

[amdx]

On 3/26/2016 12:37 PM, M Philbrook wrote:

In article <nd4759$bpo$1@dont-email.me>, nojunk@knology.net says...

On 3/23/2016 6:07 PM, John Larkin wrote:
On Wed, 23 Mar 2016 14:36:49 -0700 (PDT), mrdarrett@gmail.com wrote:

Hey,

I am looking for a J308, but it looks like it's obsolete.

http://www.mouser.com/ProductDetail/InterFET/J308/?qs=OxRSArmBDfwrWzaWx947ZA%3D%3D

Any ideas?

I'm trying to build a low voltage oscillator that works down to about 0.15 V, basically these:

http://www.google.com/patents/US4734658

http://web.archive.org/web/20150316093734/http://neazoi.gr/energy-harvesting/index.htm

Thanks!

Michael

I've posted this before, author says the best iteration will start
oscillating at 7mV 1.1Ma. He is not drawing power from the circuit, so...
Oops, he got it down to 5.5Mv 0.41 Ma, he moved the scope to a lower
impedance point, it was loading his circuit.

http://www.dicks-website.eu/fetosc/enindex.htm

Mikek

You do mean
5.5mV 0.41mA ?
Us old guys get confused..

Jamie

Sorry, yes that's what I meant. Looks like I got it right 25% of the
time. Do you grade on the curve?
Mikek

 

[John Larkin]

On Sat, 26 Mar 2016 16:03:29 -0700 (PDT), mrdarrett@gmail.com wrote:

On Saturday, March 26, 2016 at 10:11:33 AM UTC-7, John Larkin wrote:
On Fri, 25 Mar 2016 13:00:59 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/24/2016 06:45 PM, John Larkin wrote:
On Thu, 24 Mar 2016 14:49:02 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/23/2016 07:34 PM, mrdarrett@gmail.com wrote:
On Wednesday, March 23, 2016 at 4:31:35 PM UTC-7, mrda...@gmail.com wrote:
On Wednesday, March 23, 2016 at 4:07:25 PM UTC-7, John Larkin wrote:

...

A phemt might work better than a jfet. Transconductance can be really
high, 10x a jfet maybe.

http://www.digikey.com/product-detail/en/cel/NE3508M04-A/NE3508M04-A-ND/949555


Ooh, Digikey is out (non-stock). What is a Phemt?


Oh, never mind, I found it! It really is a thing!

https://en.wikipedia.org/wiki/High-electron-mobility_transistor

Michael


pHEMTs are amazing parts for some things, but I'm pretty sure that
John's pulling your leg about using one in a Joule Thief. The slowest
pHEMTs are about 10 GHz iirc. They also have seriously crummy Early
voltages, so they might not work as well as a JFET when V_DS gets down
to the hundreds of millivolts.

I was serious. Rds-on is about 6 ohms for that NEC part at zero gate
voltage, and transconductance is outrageous compared to a jfet or a
depletion mosfet. Just what you want to power some oscillator thing
from a thermocouple or some such millivolt source.

Okay, my mistake. (That "lunatic fringe" thing again.)

The transconductance is nice and high at fixed bias, true. However, the
Avago parts have the wimpiest drain impedance I've ever seen--something
like 160 ohms. You can't even use them as source followers unless you
bootstrap the drain. The Skyworks parts (SKY65050) are quite a bit
better, but still not even as good as a JFET. I haven't tried the NEC
parts.

It would be entertaining to try using a 14-GHz part on a white plastic
breadboard. OTOH they're surprisingly stable for such fast and quiet
devices--a lot more stable than the SiGe parts!

Cheers

Phil Hobbs

Phemts have some potential as low-frequency amps and switches. I don't
know what is the record for making an oscillator powered by
millivolts, but a phemt is a candidate to do that mostly useless
thing. I've seen the claim of 5 mV, with paralleled jfets.


Neato. Is that what boosts the mV from the thermocouple in a gas water heater, to control the temperature?

They usually use a multi-junction thermocouple, 30 millivolts maybe,
and a low-resistance solenoid valve. This was done way before
transistors were invented.

They are very stable as compared to bipolars. Drain-gate capacitances
are absurdly low, and the substrate is the source. A hacked copperclad
breadboard usually works fine.


"... very stable as compared to bipolars." That would make a funny mental health joke.

https://dl.dropboxusercontent.com/u/53724080/bipolar.JPG

I have some DC measurements of the NE3508 if anybody's interested.


Sure!

https://dl.dropboxusercontent.com/u/53724080/Parts/NE350x/NE350x.zip


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Michael A. Terrell]

amdx wrote:

Do you grade on the curve?

No, on a trapezoid! ;-)

 

On Saturday, March 26, 2016 at 9:45:12 PM UTC-7, John Larkin wrote:
> On Sat, 26 Mar 2016 16:03:29 -0700 (PDT), mrdarrett@gmail.com wrote:

....

They are very stable as compared to bipolars. Drain-gate capacitances
are absurdly low, and the substrate is the source. A hacked copperclad
breadboard usually works fine.


"... very stable as compared to bipolars." That would make a funny mental health joke.



https://dl.dropboxusercontent.com/u/53724080/bipolar.JPG

And they're the same transistor?

I have some DC measurements of the NE3508 if anybody's interested.


Sure!



https://dl.dropboxusercontent.com/u/53724080/Parts/NE350x/NE350x.zip

Wow, a lot of info there!

Thanks!

Michael

 

[Phil Hobbs]

On 03/27/2016 11:42 PM, mrdarrett@gmail.com wrote:

On Saturday, March 26, 2016 at 9:45:12 PM UTC-7, John Larkin wrote:
On Sat, 26 Mar 2016 16:03:29 -0700 (PDT), mrdarrett@gmail.com wrote:

...

They are very stable as compared to bipolars. Drain-gate capacitances
are absurdly low, and the substrate is the source. A hacked copperclad
breadboard usually works fine.


"... very stable as compared to bipolars." That would make a funny mental health joke.



https://dl.dropboxusercontent.com/u/53724080/bipolar.JPG


And they're the same transistor?




I have some DC measurements of the NE3508 if anybody's interested.


Sure!



https://dl.dropboxusercontent.com/u/53724080/Parts/NE350x/NE350x.zip


Wow, a lot of info there!

Thanks!

Michael

Thanks for the info, John.

That family of drain curves (DSC01311.JPG) shows the Early voltage
problem very well. The -0.3V curve in the 3508 datasheet (P 3) passes
through (3.5V, 23 mA) and (1.2V, 10 mA) for an Early voltage of

VA = - (10 mA/13 mA)*(3.5 V - 1.2 V) -1.2 V = 0.6 V.

OTOH the saturation plot (DSC01309.JPG) shows the nice high
transconductance, like 300 mA/V at 1V D-S.


The Murata BLM18BB-series beads do a nice job of stabilizing a BFP640
cascode device, though the bandwidth sacrifice is considerable. One
fine day I'll have to get down Carson and Hollister and figure out a
broader-bandwidth solution, but so far it's been fast enough just with
the bead.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

 

[John Larkin]

On Mon, 28 Mar 2016 10:13:20 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/27/2016 11:42 PM, mrdarrett@gmail.com wrote:
On Saturday, March 26, 2016 at 9:45:12 PM UTC-7, John Larkin wrote:
On Sat, 26 Mar 2016 16:03:29 -0700 (PDT), mrdarrett@gmail.com wrote:

...

They are very stable as compared to bipolars. Drain-gate capacitances
are absurdly low, and the substrate is the source. A hacked copperclad
breadboard usually works fine.


"... very stable as compared to bipolars." That would make a funny mental health joke.



https://dl.dropboxusercontent.com/u/53724080/bipolar.JPG


And they're the same transistor?




I have some DC measurements of the NE3508 if anybody's interested.


Sure!



https://dl.dropboxusercontent.com/u/53724080/Parts/NE350x/NE350x.zip


Wow, a lot of info there!

Thanks!

Michael


Thanks for the info, John.

That family of drain curves (DSC01311.JPG) shows the Early voltage
problem very well. The -0.3V curve in the 3508 datasheet (P 3) passes
through (3.5V, 23 mA) and (1.2V, 10 mA) for an Early voltage of

VA = - (10 mA/13 mA)*(3.5 V - 1.2 V) -1.2 V = 0.6 V.

Since I generally slam them on or off, Early voltage isn't a problem.

They do enhance nicely, to around 2x Idss.

OTOH the saturation plot (DSC01309.JPG) shows the nice high
transconductance, like 300 mA/V at 1V D-S.


The Murata BLM18BB-series beads do a nice job of stabilizing a BFP640
cascode device, though the bandwidth sacrifice is considerable. One
fine day I'll have to get down Carson and Hollister and figure out a
broader-bandwidth solution, but so far it's been fast enough just with
the bead.

Oscillation isn't a problem either!


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

On Thursday, March 24, 2016 at 3:01:41 PM UTC-7, Sjouke Burry wrote:

On 24.03.16 15:36, John Larkin wrote:
On Wed, 23 Mar 2016 16:31:27 -0700 (PDT), mrdarrett@gmail.com wrote:

On Wednesday, March 23, 2016 at 4:07:25 PM UTC-7, John Larkin wrote:
On Wed, 23 Mar 2016 14:36:49 -0700 (PDT), mrdarrett@gmail.com wrote:

Hey,

I am looking for a J308, but it looks like it's obsolete.

http://www.mouser.com/ProductDetail/InterFET/J308/?qs=OxRSArmBDfwrWzaWx947ZA%3D%3D

Any ideas?

I'm trying to build a low voltage oscillator that works down to about 0.15 V, basically these:

http://www.google.com/patents/US4734658

http://web.archive.org/web/20150316093734/http://neazoi.gr/energy-harvesting/index.htm

Thanks!

Michael


BF862, a jfet, might work.


Ok thanks! About $0.54 each at Mouser.

http://www.mouser.com/Search/Refine.aspx?Keyword=bf862



A phemt might work better than a jfet. Transconductance can be really
high, 10x a jfet maybe.

http://www.digikey.com/product-detail/en/cel/NE3508M04-A/NE3508M04-A-ND/949555


Ooh, Digikey is out (non-stock). What is a Phemt?

And, if I'm trying to run at really low voltages (basically, wringing out all the energy out of dead AA batteries) what kind of characteristics am I looking for in a JFET?


High transconductance and fairly high Idss, mostly.

You want to modulate a bunch of drain current with a small gate swing,
but how much gate swing is available depends on the circuit.

A nearly dead (low voltage) battery won't have much energy left.


Putting a 1-10 farad supercap across the battery might be
the best way to get most power out of the battery.
The cap gets a trickle charge from the battery, and can then
for some minutes supply low loss power.

There's an idea too!

Still, I'd likely need to increase the voltage to something more than 1 volt.

Also, strange things can happen when I string lots of weak batteries together... sometimes I noticed a battery reversing polarity when stringing together 8 NiMH cells in series

Thanks,

Michael

 

[Phil Hobbs]

On 03/29/2016 12:07 AM, John Larkin wrote:

On Mon, 28 Mar 2016 10:13:20 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 03/27/2016 11:42 PM, mrdarrett@gmail.com wrote:
On Saturday, March 26, 2016 at 9:45:12 PM UTC-7, John Larkin wrote:
On Sat, 26 Mar 2016 16:03:29 -0700 (PDT), mrdarrett@gmail.com wrote:

...

They are very stable as compared to bipolars. Drain-gate capacitances
are absurdly low, and the substrate is the source. A hacked copperclad
breadboard usually works fine.


"... very stable as compared to bipolars." That would make a funny mental health joke.



https://dl.dropboxusercontent.com/u/53724080/bipolar.JPG


And they're the same transistor?




I have some DC measurements of the NE3508 if anybody's interested.


Sure!



https://dl.dropboxusercontent.com/u/53724080/Parts/NE350x/NE350x.zip


Wow, a lot of info there!

Thanks!

Michael


Thanks for the info, John.

That family of drain curves (DSC01311.JPG) shows the Early voltage
problem very well. The -0.3V curve in the 3508 datasheet (P 3) passes
through (3.5V, 23 mA) and (1.2V, 10 mA) for an Early voltage of

VA = - (10 mA/13 mA)*(3.5 V - 1.2 V) -1.2 V = 0.6 V.

Since I generally slam them on or off, Early voltage isn't a problem.

They do enhance nicely, to around 2x Idss.


OTOH the saturation plot (DSC01309.JPG) shows the nice high
transconductance, like 300 mA/V at 1V D-S.


The Murata BLM18BB-series beads do a nice job of stabilizing a BFP640
cascode device, though the bandwidth sacrifice is considerable. One
fine day I'll have to get down Carson and Hollister and figure out a
broader-bandwidth solution, but so far it's been fast enough just with
the bead.

Oscillation isn't a problem either!

I bet a BFP640 could get in a cycle or two even with your edge rates.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

 

[whit3rd]

On Wednesday, March 30, 2016 at 9:31:44 AM UTC-7, mrda...@gmail.com wrote:

> Also, strange things can happen when I string lots of weak batteries together... sometimes I noticed a battery reversing polarity when stringing together 8 NiMH cells in series

Not at all unexpected. The big battery users (diesel/electric submarines, for example) always
had to do cell-by-cell testing, and shunting a dead or weak cell was a routine maintenance
chore (unless it related to battle damage...).