Hello,

I'm assembling a classic preamplifier with the BF-981 dual-gate
mosfet. The datasheet says the best noise figure is obtained with
Vg2S= 4V, Vg1S= 0V and Id=10 mA.
So I ended with a Vdd of 9V (because I had a spare 7809) and obtained
4V for Vg2 with a 68k+56k divider, Vg1= 0V through the coil of the
tuned input filter, Rs first replaced with a multiturn 500 ohm trimmer
to find the correct 10 mA point. Rs and the 56k to ground bypassed
with 10 nF ceramic.
Easy, right?
The problem is I can't get the mosfet to drain more than 7 mA (at Rs=0
ohm). I also tried to increase slightly Vg2S by replacing the 68k with
another 56k but I end with about 7.5 mA max. Bringing Vd up to 12V
makes almost no difference.
Do I have a broken BF-981 (I don't have another one handy to test) or
it could be a normal value spread for this device?

Thanks

Francesco IZ8DWF

The datasheets generally gives you the average values. The typical
values for FETs and MOSFETs are all over the map - so I won't worry that
you're not seeing the 10mA drain current. Try out your circuit, you'll
probably be happy with it.

wrote:
Hello,

I'm assembling a classic preamplifier with the BF-981 dual-gate
mosfet. The datasheet says the best noise figure is obtained with
Vg2S= 4V, Vg1S= 0V and Id=10 mA.
So I ended with a Vdd of 9V (because I had a spare 7809) and obtained
4V for Vg2 with a 68k+56k divider, Vg1= 0V through the coil of the
tuned input filter, Rs first replaced with a multiturn 500 ohm trimmer
to find the correct 10 mA point. Rs and the 56k to ground bypassed
with 10 nF ceramic.
Easy, right?
The problem is I can't get the mosfet to drain more than 7 mA (at Rs=0
ohm). I also tried to increase slightly Vg2S by replacing the 68k with
another 56k but I end with about 7.5 mA max. Bringing Vd up to 12V
makes almost no difference.
Do I have a broken BF-981 (I don't have another one handy to test) or
it could be a normal value spread for this device?

Thanks

Francesco IZ8DWF

A couple of things come to mind...
The best noise figure also has another parameter: Vds=10V. It might
take that much to push 10mA through the channel.
One other thing to remember is that as Rs increases, Vg1s and Vg2s
decrease, which tends to further pinch the channel and reduce Id.
I'd try Vd=12V and Rs=200ohms, and use either a 10k resistor or
(better) an RF Choke between g1 and s...that will force Vg1s=0 no
matter what the bias condition. That will set Vs=2V, so then an equal-
resistor divider from Vd gives 6V which is Vg2s=4V.
If you really can't pull 10mA then it might be worth trying out anyway
(as previously suggested), but I'd suspect the device is no good.


On Nov 5, 8:19 am, wrote:
Hello,

I'm assembling a classic preamplifier with the BF-981 dual-gate
mosfet. The datasheet says the best noise figure is obtained with
Vg2S= 4V, Vg1S= 0V and Id=10 mA.
So I ended with a Vdd of 9V (because I had a spare 7809) and obtained
4V for Vg2 with a 68k+56k divider, Vg1= 0V through the coil of the
tuned input filter, Rs first replaced with a multiturn 500 ohm trimmer
to find the correct 10 mA point. Rs and the 56k to ground bypassed
with 10 nF ceramic.
Easy, right?
The problem is I can't get the mosfet to drain more than 7 mA (at Rs=0
ohm). I also tried to increase slightly Vg2S by replacing the 68k with
another 56k but I end with about 7.5 mA max. Bringing Vd up to 12V
makes almost no difference.
Do I have a broken BF-981 (I don't have another one handy to test) or
it could be a normal value spread for this device?

Thanks

Francesco IZ8DWF

On Nov 5, 7:19 am, wrote:
Hello,

I'm assembling a classic preamplifier with the BF-981 dual-gate
mosfet. The datasheet says the best noise figure is obtained with
Vg2S= 4V, Vg1S= 0V and Id=10 mA.
So I ended with a Vdd of 9V (because I had a spare 7809) and obtained
4V for Vg2 with a 68k+56k divider, Vg1= 0V through the coil of the
tuned input filter, Rs first replaced with a multiturn 500 ohm trimmer
to find the correct 10 mA point. Rs and the 56k to ground bypassed
with 10 nF ceramic.
Easy, right?
The problem is I can't get the mosfet to drain more than 7 mA (at Rs=0
ohm). I also tried to increase slightly Vg2S by replacing the 68k with
another 56k but I end with about 7.5 mA max. Bringing Vd up to 12V
makes almost no difference.
Do I have a broken BF-981 (I don't have another one handy to test) or
it could be a normal value spread for this device?

Thanks

Francesco IZ8DWF

Hi Francesco,

The data sheet I was able to quickly find suggests optimum noise
figure at a particular Vds, Vg2s, and Id, but does not mention the
Vg1s required to achieve that Id. I suppose that for your particular
part, you may need to run gate 1 slightly positive with respect to the
source to get the 10mA. However, I doubt that there will be a
significant advantage to doing so with respect to the noise figure.
It will almost certainly be more important to adjust the input
matching to achieve lowest noise figure.

The data sheet I found was old and only seems to have summary data. I
looked on the NXP web site, and found what seems to be a similar part,
the BF991. The data sheet for it is complete, but it seems still very
sparse. I am used to data sheets for low noise RF amplifier
transistors that include graphs of things like noise figure versus
input matching at some particular bias, and possibly for two or three
bias conditions. It was a disappointment to not see something like
that for the BF9x1 parts, though they may be good parts anyway.

Cheers,
Tom

Hello all and thanks for all the answers,

as I already said, I tried to raise Vd up to 12V and Id almost doesn't
change, as Vg2s remains constant.
g1 is always at 0V in this design since there's an inductor to ground
(part of a parallel resonant LC), so actually Vg1s is slightly
negative, less than -0.1V with a 15 ohm Rs. I doubt this really
affects the Id. I actually tried to keep Vg2s at 0V and change Vg1s
and what I found is a weaker effect on Id with respect to Vg2s, with
Vg2s=0V and Vg1s=4V the Id was less than 3mA. Clearly the two gates
aren't interchangeable.

On 6 Nov, 03:20, K7ITM wrote:


Hi Francesco,

The data sheet I was able to quickly find suggests optimum noise
figure at a particular Vds, Vg2s, and Id, but does not mention the
Vg1s required to achieve that Id. I suppose that for your particular
part, you may need to run gate 1 slightly positive with respect to the
source to get the 10mA. However, I doubt that there will be a
significant advantage to doing so with respect to the noise figure.
It will almost certainly be more important to adjust the input
matching to achieve lowest noise figure.


righ, but on the datasheet there's no data about the right source
impedance for best NF.

The data sheet I found was old and only seems to have summary data. I
looked on the NXP web site, and found what seems to be a similar part,
the BF991. The data sheet for it is complete, but it seems still very
sparse. I am used to data sheets for low noise RF amplifier
transistors that include graphs of things like noise figure versus
input matching at some particular bias, and possibly for two or three
bias conditions. It was a disappointment to not see something like
that for the BF9x1 parts, though they may be good parts anyway.

I'll have a look at the BF991 datasheet too, thanks for hint.
I have actually other spare mosfets like the BF960 and BF966, but they
seems worse than the BF981.

Thanks again
Francesco IZ8DWF

On Nov 6, 12:44 am, wrote:
Hello all and thanks for all the answers,
...

On 6 Nov, 03:20, K7ITM wrote:

....
It will almost certainly be more important to adjust the input
matching to achieve lowest noise figure.

righ, but on the datasheet there's no data about the right source
impedance for best NF.

Hello Francesco,

But I suppose that will be a problem no matter whether Id is 10mA or
7mA. ;-)


The data sheet I found was old and only seems to have summary data. I
looked on the NXP web site, and found what seems to be a similar part,
the BF991. The data sheet for it is complete, but it seems still very
sparse. I am used to data sheets for low noise RF amplifier
transistors that include graphs of things like noise figure versus
input matching at some particular bias, and possibly for two or three
bias conditions. It was a disappointment to not see something like
that for the BF9x1 parts, though they may be good parts anyway.

I'll have a look at the BF991 datasheet too, thanks for hint.
I have actually other spare mosfets like the BF960 and BF966, but they
seems worse than the BF981.

At least the BF991 data sheet lists a recommended source impedance for
optimal noise figure. I would guess that the BF981 will be quite
similar. Of course, it will always be best to adjust for lowest noise
figure.

I am not surprised that going between 9V and 10V made very little
difference in the drain current. You certainly must be operating
there in the "saturation" region of the FET (not to be confused with
saturation in a bipolar...), where Id is very nearly independent of
Vds, for constant gate voltages. Also, if you have a source resistor,
that provides negative DC feedback and will stabilize the drain
current--if the current decreases, the voltage drop across that
resistor decreases, and Vg1s (and Vg2s) become more positive, thus
tending to keep Id even more constant than it would otherwise be.

I would not do it myself, but if you feel a need to get Id up to 10mA,
you could remove the "cold" end of the inductor from ground, bypass it
to ground with a small capacitor, and apply a slightly positive
voltage from a voltage divider. I think it is much better to just go
with the 7mA or whatever it turns out to be for your particular part,
and concentrate on input matching adjustments.

You didn't mention what frequency this preamp is for, did you? I am
curious--is it for 144MHz or 440MHz or some other band?

Cheers,
Tom

Hello Tom,

On 6 Nov, 15:52, K7ITM wrote:

...
It will almost certainly be more important to adjust the input
matching to achieve lowest noise figure.

righ, but on the datasheet there's no data about the right source
impedance for best NF.

Hello Francesco,

But I suppose that will be a problem no matter whether Id is 10mA or
7mA. ;-)

yes indeed, unfortunately I don't have a NF test equipment to play
with, so I can't experiment too much.



I'll have a look at the BF991 datasheet too, thanks for hint.
I have actually other spare mosfets like the BF960 and BF966, but they
seems worse than the BF981.

At least the BF991 data sheet lists a recommended source impedance for
optimal noise figure. I would guess that the BF981 will be quite
similar. Of course, it will always be best to adjust for lowest noise
figure.

having a way to measure it, of course.


I am not surprised that going between 9V and 10V made very little
difference in the drain current. You certainly must be operating
there in the "saturation" region of the FET (not to be confused with
saturation in a bipolar...), where Id is very nearly independent of
Vds, for constant gate voltages. Also, if you have a source resistor,
that provides negative DC feedback and will stabilize the drain
current--if the current decreases, the voltage drop across that
resistor decreases, and Vg1s (and Vg2s) become more positive, thus
tending to keep Id even more constant than it would otherwise be.

That's right, I just checked what happened rising the Vd because it
was easy to do, and checking theory always helps.


I would not do it myself, but if you feel a need to get Id up to 10mA,
you could remove the "cold" end of the inductor from ground, bypass it
to ground with a small capacitor, and apply a slightly positive
voltage from a voltage divider. I think it is much better to just go
with the 7mA or whatever it turns out to be for your particular part,
and concentrate on input matching adjustments.

I won't do it either, I just have the feeling that the particular
mosfet I have is maybe failing, but I would have to steal another 981
from a working preamp to be sure, which I'm reluctant to do, or use a
different dual gate mosfet like the 960 or 966 (I have plenty of them
from TV tuners).


You didn't mention what frequency this preamp is for, did you? I am
curious--is it for 144MHz or 440MHz or some other band?

70 Mhz, not a really quiet band actually, but I live in a quiet place
anyway. With a transverter tuned by ear (currently I don't even have a
spectrum analyzer) and a 6 element yagi home made I can hear very
frequent MS pings from beacons I normally don't hear. It's often
enough to tune a beacon frequency and turn to the right direction and
wait some minutes to get some pings and be sure everything is working.
If we get permission to operate on this band in the next years we
might be able to make the first TEP contact on this band. I don't
think a preamp would make a difference in my setup, but it surely
won't harm.

Thanks again and best 73
Francesco IZ8DWF

Francesco,
I have some better answers now, as I have a very complete BF981
datasheet in front of me.
1. For fixed Vg1s and Vg2s, changing Vds has no effect on Id. Just
like you observed.
2. The *typical* Id for [Vg1s=0, Vg2s=+4, Vds=10V] is 11mA. The
minimum value of Id for the same bias point is 4mA. So your observed
7mA is within the process variation.
3. No the gates are not interchangable.
4. The datasheet does not give any Vg2s/Id curves, and Vg1s cannot be
positive (max gate current is exceeded nearly immediately) so you're
stuck with 7mA for this part.
5. Id *does* depends very strongly on Vg1s...for the typical part,
Vg1s=0V gives 11mA, but Vg1s=-0.5V gives Id=4mA.
6. Noise data is given at Vds=10V, Vg1s=0V, Vg2s=+4V: Fmin=0.6dB,
Gs,opt=0.6mmho, Bs,opt=-1.25mmho @ 100MHz.

The BF991 does look to be the same die as the BF981, just in the SMT
package.
GL,

On Nov 6, 1:44 am, wrote:
Hello all and thanks for all the answers,

as I already said, I tried to raise Vd up to 12V and Id almost doesn't
change, as Vg2s remains constant.
g1 is always at 0V in this design since there's an inductor to ground
(part of a parallel resonant LC), so actually Vg1s is slightly
negative, less than -0.1V with a 15 ohm Rs. I doubt this really
affects the Id. I actually tried to keep Vg2s at 0V and change Vg1s
and what I found is a weaker effect on Id with respect to Vg2s, with
Vg2s=0V and Vg1s=4V the Id was less than 3mA. Clearly the two gates
aren't interchangeable.

On 6 Nov, 03:20, K7ITM wrote:



Hi Francesco,

The data sheet I was able to quickly find suggests optimum noise
figure at a particular Vds, Vg2s, and Id, but does not mention the
Vg1s required to achieve that Id. I suppose that for your particular
part, you may need to run gate 1 slightly positive with respect to the
source to get the 10mA. However, I doubt that there will be a
significant advantage to doing so with respect to the noise figure.
It will almost certainly be more important to adjust the input
matching to achieve lowest noise figure.

righ, but on the datasheet there's no data about the right source
impedance for best NF.

The data sheet I found was old and only seems to have summary data. I
looked on the NXP web site, and found what seems to be a similar part,
the BF991. The data sheet for it is complete, but it seems still very
sparse. I am used to data sheets for low noise RF amplifier
transistors that include graphs of things like noise figure versus
input matching at some particular bias, and possibly for two or three
bias conditions. It was a disappointment to not see something like
that for the BF9x1 parts, though they may be good parts anyway.

I'll have a look at the BF991 datasheet too, thanks for hint.
I have actually other spare mosfets like the BF960 and BF966, but they
seems worse than the BF981.

Thanks again
Francesco IZ8DWF

Hello,

On 7 Nov, 02:20, nx7u wrote:
Francesco,
I have some better answers now, as I have a very complete BF981
datasheet in front of me.

Is it in electronic form? If so I'd be interested in it.

1. For fixed Vg1s and Vg2s, changing Vds has no effect on Id. Just
like you observed.
2. The *typical* Id for [Vg1s=0, Vg2s=+4, Vds=10V] is 11mA. The
minimum value of Id for the same bias point is 4mA. So your observed
7mA is within the process variation.
3. No the gates are not interchangable.
4. The datasheet does not give any Vg2s/Id curves, and Vg1s cannot be
positive (max gate current is exceeded nearly immediately) so you're
stuck with 7mA for this part.

hmm then I surely damaged the mosfet when playing with gates
reversed.
I better find another one.

5. Id *does* depends very strongly on Vg1s...for the typical part,
Vg1s=0V gives 11mA, but Vg1s=-0.5V gives Id=4mA.
6. Noise data is given at Vds=10V, Vg1s=0V, Vg2s=+4V: Fmin=0.6dB,
Gs,opt=0.6mmho, Bs,opt=-1.25mmho @ 100MHz.

Thanks for the informations, If I had seen the datasheet before I
wouldn't ever have swapped the gates.
Now looking for another mosfet :-)

73
Francesco IZ8DWF

Hi Allison,

On 7 Nov, 15:21, wrote:
On Wed, 07 Nov 2007 00:16:18 -0800,
wrote:



Hello,

On 7 Nov, 02:20, nx7u wrote:

4. The datasheet does not give any Vg2s/Id curves, and Vg1s cannot be
positive (max gate current is exceeded nearly immediately) so you're
stuck with 7mA for this part.

hmm then I surely damaged the mosfet when playing with gates
reversed.
I better find another one.

Reversing the gates wou;d not damage them. The gain would be poor
and other effects but no damage.

Hmm ok, I haven't yet desoldered the original one, I assumed it would
be bad when I read about the maximum gate current exceeded when Vg1s
is positive, which happened when I reversed the gates.


Thanks

Francesco