Added Series "Dropping" Resistor on Drain Provides Temp. Compensation?

[[email protected]]

Hello,


I have 2-18GHz MMIC amplifiers that I buy off the shelf
from companies like Velocium and Hittite. Pretty small signal
stuff, like at max 18dBm output, although sometimes I'll compress
them, so it's not all Class A.

The data sheets ask for a nominal drain voltage, but since all the amps
have different values, what I end up doing is supplying them all
from one +7VDC regulator, and use various dropping resistors to
target the desired Vdd (usually around +3 to +6 volts),
based on the average operating current (around 50-150mA or so).
There are plenty of decoupling caps on the regulator side of the
resistors,
so a good AC short to ground is maintained.

My question is can I consider the action of adding a series resistor
to the drain supply (thereby making them more like current-sources
instead of voltage sources) a method of adding built-in temperature
compensation to the amplifier?

It seems to me that as the transconductance goes down with colder
temp, that there will be less current and less voltage drop across the
resistor, and so a higher drain voltage.


Thanks for your input.


Slick

 

[[email protected]]

so ya i have no idea what your talking about but i had to post one of
these comments for my class so dont pay any mind to it ok well bye

 

[[email protected]]

thats cool man nice but hey heres tyhe thing i am workin on my pc and
tryin to fig out if i can go from a 754 socket motherboard to a 949 one
you know if i can?

 

[Winfield Hill]

[email protected] wrote...

I have 2-18GHz MMIC amplifiers that I buy off the shelf from
companies like Velocium and Hittite. Pretty small signal
stuff, like at max 18dBm output, although sometimes I'll
compress them, so it's not all Class A.

The data sheets ask for a nominal drain voltage, but since all
the amps have different values, what I end up doing is supplying
them all from one +7VDC regulator, and use various dropping
resistors to target the desired Vdd (usually around +3 to +6 volts),
based on the average operating current (around 50-150mA or so).
There are plenty of decoupling caps on the regulator side of
the resistors, so a good AC short to ground is maintained.

My question is can I consider the action of adding a series
resistor to the drain supply (thereby making them more like
current-sources instead of voltage sources) a method of adding
built-in temperature compensation to the amplifier?

It seems to me that as the transconductance goes down with colder
temp, that there will be less current and less voltage drop across
the resistor, and so a higher drain voltage.

I don't see the value in that reasoning, because gm = Ic/VT,
where VT = kT/q, for a negative transconductance tempco with
constant current. [To eliminate gm tempco you need to use a
PTAT (proportional to absolute temperature) current source.
To get a PTAT CS, use a BJT with Vb = Vg0 = 1.23V (e.g. an
LM385-1.2 or LM4041 reference) and an emitter resistor.
Vg0 is the bandgap of silicon extrapolated to 0K, where 0K
means absolute zero... alphabet soup.]

You may find saving a bunch of regulators appealing for a
personal one-off, but I wouldn't do it for any serious use
or for production. A low-current part would experience
more voltage than intended, perhaps even damaging it, and
a high-current part would have inadequate voltage, and be
unable to reach the specified output power. Even if you
trim the resistors for your one-off, if you ever needed to
change a component, you'd be back to square one. Not good.

 

[Mark]

My question is can I consider the action of adding a series resistor
to the drain supply (thereby making them more like current-sources
instead of voltage sources) a method of adding built-in temperature
compensation to the amplifier?

Check the app notes, many MMIC RF amps REQUIRE a sereis resistor to
stabilize the current vs temperature...

what you reaaly want is to construct the V/I load line of the amp at
various temperatures and then see what the voltage and current will be
for various supply and resistor values.

Mark

 

[[email protected]]

It seems to me that as the transconductance goes down with colder
temp, that there will be less current and less voltage drop across
the resistor, and so a higher drain voltage.

I don't see the value in that reasoning, because gm = Ic/VT,
where VT = kT/q, for a negative transconductance tempco with
constant current. [To eliminate gm tempco you need to use a
PTAT (proportional to absolute temperature) current source.
To get a PTAT CS, use a BJT with Vb = Vg0 = 1.23V (e.g. an
LM385-1.2 or LM4041 reference) and an emitter resistor.
Vg0 is the bandgap of silicon extrapolated to 0K, where 0K
means absolute zero... alphabet soup.]

I suppose my question is really this:

Will the ouput power of a MMIC vary less over temp
with a series dropping resistor (more like a current source)
than with a straight voltage source.

The question is complicated by the fact that the
MMIC will have built in RF chokes on the die, so we are not
actually tied to the drain directly.

You may find saving a bunch of regulators appealing for a
personal one-off, but I wouldn't do it for any serious use
or for production. A low-current part would experience
more voltage than intended, perhaps even damaging it, and
a high-current part would have inadequate voltage, and be
unable to reach the specified output power. Even if you
trim the resistors for your one-off, if you ever needed to
change a component, you'd be back to square one. Not good.

That's not really a concern because even if
the device draws zero current, the voltage from the regulator
will still be less than the absolute maximum drain voltage rating on
the data sheets.

And as the other poster mentions, the MMIC supplier
recommends a dropping resistor in the data sheet for proper
operation. It would have to be extreme lot to lot and wafer variations
for us to need to adjust these during production (once we dial
them in during prototyping).

Thank you for the input, Mr. Hill.


Dr. Slick

 

[[email protected]]

Check the app notes, many MMIC RF amps REQUIRE a sereis resistor to
stabilize the current vs temperature...

That's the reason we are using dropping resistors in
the first place. It seems you agree with me then that one of the
reasons
for using the dropping resistor is for temp comp.

what you reaaly want is to construct the V/I load line of the amp at
various temperatures and then see what the voltage and current will be
for various supply and resistor values.

Well, if you look at an AC load line (assume
linear class A for simplicity) over your DC I/V curves, you
can see how lowering your drain voltage will begin to compress your
amplifier sooner, at a lower output power. So if we lower
the transconductance at high temp, the entire AC load line is shifted
down,
and the average drain current drops. But if we increase the
drain voltage, then we shift the load line to the right, and can
get the voltage swing back to the room temp values.

A bit simplistic of a model, i'll agree, but seems to
make sense.


Slick