Phil: SAV541 model

[John Larkin]

Hey, Phil,

Your SAV-551 Spice model is

.MODEL SAV551 NMF(vto=0.08, Beta=0.3,
+ Lambda=0.07, Alpha=4 B=0.8, Pb=0.7,
+ Cgs=0.997E-12, Cgd=0.176E-12, Rd=0.084,
+ Rs=0.054, Kf=5e-11, Af=2)

I was looking at the SAV-541. It conducts harder than 551 but looks
about the same otherwise. I measured gate and drain capacitances and
they look about the same. The s-params suggest a bit more capacitance
for the 541 but they are at different currents.

Rds-on is about the same, about 2.5 ohms, but the data sheet drain
curve of the 541 is a lot beefier. (Who measures Rds-on of a phemt?)

With the same capacitances but more drain current capability, it
should switch faster.

What might I change in your model for a SAV-541? I could simulate and
see how it aligns with the data sheet.

I wonder why the 551 is rated for 0.45 to 6 GHz and the 541 is rated
for 0.045 to 6 GHz. That's weird.

[Phil Hobbs]

John Larkin wrote:
> Hey, Phil,
>
> Your SAV-551 Spice model is
>

> ..MODEL SAV551 NMF(vto=0.08, Beta=0.3,

> + Lambda=0.07, Alpha=4 B=0.8, Pb=0.7,
> + Cgs=0.997E-12, Cgd=0.176E-12, Rd=0.084,
> + Rs=0.054, Kf=5e-11, Af=2)
>
> I was looking at the SAV-541. It conducts harder than 551 but looks
> about the same otherwise. I measured gate and drain capacitances and
> they look about the same. The s-params suggest a bit more capacitance
> for the 541 but they are at different currents.
>
> Rds-on is about the same, about 2.5 ohms, but the data sheet drain
> curve of the 541 is a lot beefier. (Who measures Rds-on of a phemt?)
>
> With the same capacitances but more drain current capability, it
> should switch faster.

Plus with more transconductance, other things being equal it should be
quieter too.

>
> What might I change in your model for a SAV-541? I could simulate and
> see how it aligns with the data sheet.

Dunno. I didn't dive into the guts of the mesfet model--my very
systematic procedure was to take Avago's model for the ATF38143 (which
has similar drain curves and capacitances), dump the parameters that
LTspice didn't understand, and dork the threshold voltage to make it an
enhancement device. A bit later I hacked the 1/f noise parameters
because the original was like 17 orders of magnitude too noisy. So I
don't know what the other params do.

> I wonder why the 551 is rated for 0.45 to 6 GHz and the 541 is rated
> for 0.045 to 6 GHz. That's weird.

Might be their testing, or maybe the 541 has lower 1/f noise. I've
measured the 1/f corner of the ATF38143 at ~10 MHz and the SKY65050 at
30 to 50 MHz. (!)

Of course they're ~10 dB quieter than a CPH3910 in the flatband, so with
a 10 MHz corner frequency they're still quieter down to about 1 MHz,
besides having about 10x the transconductance. Magic.

Cheers

Phil Hobbs


--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

[Gerhard Hoffmann]

Am 04.05.21 um 04:39 schrieb Phil Hobbs:

>> Rds-on is about the same, about 2.5 ohms, but the data sheet drain
>> curve of the 541 is a lot beefier. (Who measures Rds-on of a phemt?)
>>
>> With the same capacitances but more drain current capability, it
>> should switch faster.
>
> Plus with more transconductance, other things being equal it should be
> quieter too.
>
>>
>> What might I change in your model for a SAV-541? I could simulate and
>> see how it aligns with the data sheet.
>
> Dunno.  I didn't dive into the guts of the mesfet model--my very
> systematic procedure was to take Avago's model for the ATF38143 (which
> has similar drain curves and capacitances), dump the parameters that
> LTspice didn't understand, and dork the threshold voltage to make it an
> enhancement device.  A bit later I hacked the 1/f noise parameters
> because the original was like 17 orders of magnitude too noisy.  So I
> don't know what the other params do.
>
>> I wonder why the 551 is rated for 0.45 to 6 GHz and the 541 is rated
>> for 0.045 to 6 GHz. That's weird.
>
> Might be their testing, or maybe the 541 has lower 1/f noise.  I've
> measured the 1/f corner of the ATF38143 at ~10 MHz and the SKY65050 at
> 30 to 50 MHz. (!)

Many years ago I simulated the use of one of these Avago thingies
in a low noise / high IP3 preamp for the 144 MHz ham radio band.
It was impossible to stabilize without killing all the good properties.
Operation on the 432 MHz band was much easier.
That was done by s-parameter simulation with the Ansys simulator,
student version.

I see that these SKY and Qorvo amplifiers also need special
stabilization measures for, say, 100 MHz operation.
The SKY 67150 seems OK on 432 MHz, although I could not duplicate
the 0.3 dB noise figure on my board, mine measured measured 0.5 -0.6 dB.
I even swapped parts between the eval board and mine. The eval board
has a Rogers 4003 top layer, that must be it.

There is a new Qorvo chip that looks like the SKY at first glance, even
slightly better. But it has BIAS on Pin 1, not on Pin 3. :-(


> Cheers

Gerhard

[Phil Hobbs]

Weird. I use them routinely at baseband, specifically for bootstraps.
I did one last year that ran the bootstrap over a 2-inch FFC cable. It
needed a couple of BLM15BA050SN1 beads, one in the gate and one in the
source. Usually you don't need the beads.

I find that the 6-GHz class pHEMTs are actually pretty resistant to
oscillations--the 40-GHz SiGe:C BJTs are much much twitchier.

>
> I see that these SKY and Qorvo amplifiers also need special
> stabilization measures for, say, 100 MHz operation.
> The SKY 67150 seems OK on 432 MHz, although I could not duplicate
> the 0.3 dB noise figure on my board, mine measured measured 0.5 -0.6 dB.
> I even swapped parts between the eval board and mine. The eval board
> has a Rogers 4003 top layer, that must be it.
>
> There is a new Qorvo chip that looks like the SKY at first glance, even
> slightly better. But it has BIAS on Pin 1, not on Pin 3.  :-(

Cheers

[jla...@highlandsniptechnology.com]

I've tried the SAV551 as a long-tail pair, driven from differential
ECL, and it works fine. I was sort of surprised.

https://www.dropbox.com/s/83rai8vrmzuz6dq/SAV551_Diff_Pair.jpg?raw=1

The optional gate resistors weren't necessary.

I haven't manage to make one oscillate, unless I wanted it too.

As far as the SAV541 model goes, just doubling beta looks pretty good.
I do need to fine tune it to get a good general model. Nuisance.

Agree, these are much more stable than fast bipolars.



--

John Larkin Highland Technology, Inc

The best designs are necessarily accidental.

[Phil Hobbs]

What sort of voltage gain do you get? I tried building a
source-feedback Schmitt out of ATF38143s some years back, but it didn't
work because the drain impedance was so low. SKY65050s worked fine.

>
> I haven't manage to make one oscillate, unless I wanted it too.
>
> As far as the SAV541 model goes, just doubling beta looks pretty good.
> I do need to fine tune it to get a good general model. Nuisance.
>
> Agree, these are much more stable than fast bipolars.

[jla...@highlandsniptechnology.com]

On Wed, 5 May 2021 10:10:55 -0400, Phil Hobbs

The ECL swings about 0.8 volts on each pin, which turns one fet on
fully and the other off, so it just current-steer switches. So gain
doesn't mean much here.

I have found that back-biasing the gates a little speeds these things
up in switchmode, and this circuit accidentally does that.

Next problem is, the EP08 gate is too slow! Not that I have a use for
this circuit; I just threw it onto a multi-prototype 4-layer board to
see what might happen.

I might hack another board this weekend, or clean the garage. Tough
call.

>
>>
>> I haven't manage to make one oscillate, unless I wanted it too.
>>
>> As far as the SAV541 model goes, just doubling beta looks pretty good.
>> I do need to fine tune it to get a good general model. Nuisance.
>>
>> Agree, these are much more stable than fast bipolars.

The EPC GaN fets are similar, in that they have almost zero drain-gate
capacitance.


>
>Cheers
>
>Phil Hobbs