MOSFET output impedance
Bojangle Chicken
I'm working with a self-designed rf amplifier using an irf530 power mosfet.
I've got the amplifier biased and working, sort of. My question is in
regards to measuring the output impedance so that I can match the load to
the amplifier.
Basically, I was sitting there at the bench thinking of how to model the
mosfet using ac analysis when it suddenly hit me that I was wasting my time
[I think...that's the root of my question-to-come]. What I did was to
replace the load with discrete resistors from about 500 ohms down to about
12 ohms, measure the rms voltage, and then compute the power (Vrms^2/R). I
got a peak power around 82 ohms with a dc-blocking capacitor of
1nF...therefore, I have the real part of the load. Correct? Also, I tried
several different sizes of capacitor, and 1nF produced the larges rms
voltage.
Am I correct in assuming that the [very approximate] output impedance of the
amplifier is Z = 82 - j10^9/w ohms? I'm essentially using basic concepts
for equal power transfer, but with imaginary components as well. Is there
any better way of getting a rough estimate for the output impedance?
Thanks for the help.
Pete
Bojangl...@yahoo.com
Cirip
It is not clear if you modelled or played with the actual part.
Also, I am not sure if you want to build a power amplifier, or just a small
signal amplifier. What class? What frequency?
Output impedance of the transistor is not of great importance in a power
amplifier. You may want to compensate for the reactive components, but the
real part has nothing to do with what the device sees from the load.
Whether you simulated the PA (it is interesting though if you have the
right model) or you just played with it on the bench, the so called optimum
load resistor, for peak output power, varies with the device gain, power
supply and drive level. The point is that complex conjugated impedances
theory does not apply in PA matching. The concept is different. You may
want to go to the Motorola (or ON Semiconductors) or Philips web site and
read some app notes about how to design a RF PA.
Good luck,
Cirip
Bojangle Chicken <bojangl...@yahoo.com> wrote in article
<9cg7ft$eet$1...@dipsy.missouri.edu>...
> Simple question...
Reg Edwards
Bojangle Chicken <bojangl...@yahoo.com> wrote in message
news:9cg7ft$eet$1...@dipsy.missouri.edu...
Why do you want to know the output (internal) impedance of the
transistor? What will you do with that information if you ever get
to know it ?
The external load on the transistor is simply calculated from the
peak voltage across the transistor and the peak current (supplied
by the transistor) flowing through it. R = V / I.
---
Reg, G4FGQ
Ian White, G3SEK
>
>Why do you want to know the output (internal) impedance of the
>transistor? What will you do with that information if you ever get to
>know it ?
>
designed according to the manufacturer's datasheet information about the
load that need to be PRESENTED TO the transistor - in other words, the
load impedance that the transistor "wants to see".
A few year ago, an applications engineer explained on this newsgroup how
the optimum load impedance is determined. It is a compromise between
several different factors, including: maximum voltage rating, maximum
current rating, power output, power dissipation, power gain and (if
relevant) linearity. The engineer uses a variable tuner to explore the
effects of load impedance on all of these characteristics, and comes up
with a recommendation that balances all the different factors involved.
You could do exactly the same, given the equipment, but you would almost
certainly come up with a different optimum load impedance, reflecting
your own judgement about your own set of priorities.
The internal impedance of the device is irrelevant to any of that, and
you don't need to know it.
One point to check is that some device manufacturers specify the actual
load impedance, while others specify its conjugate - but that still is
not the internal impedance of the device.
73 from Ian G3SEK Editor, 'The VHF/UHF DX Book'
'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.com/g3sek
Roy Lewallen
amplifiers is (Vcc^2)/(2*Po) where Po is the output power in watts. If
the transistor is driven to saturation and has appreciable Vce(sat),
then Vcc-Vce(sat) is often substituted for Vcc in the formula.
You'll find this formula used in Hayward & DeMaw's _Solid State Design
for the Radio Amateur_ (ARRL) -- which you definitely should get if you
don't already have a copy -- and other publications dealing with RF
power amplifiers, such as Motorola's app note AN721, "Impedance Matching
Networks Applied to RF Power Transistors".
Roy Lewallen, W7EL
"Ian White, G3SEK" wrote:
> Nothing at all! The output matching network for a power transistor is
> designed according to the manufacturer's datasheet information about the
> load that need to be PRESENTED TO the transistor - in other words, the
> load impedance that the transistor "wants to see".
>
> A few year ago, an applications engineer explained on this newsgroup how
> the optimum load impedance is determined. It is a compromise between
> several different factors, including: maximum voltage rating, maximum
> current rating, power output, power dissipation, power gain and (if
> relevant) linearity. The engineer uses a variable tuner to explore the
> effects of load impedance on all of these characteristics, and comes up
> with a recommendation that balances all the different factors involved.
> . . .
Wes Stewart
wrote:
>Let me add to Ian's summary that a long-used rule of thumb for HF
>amplifiers is (Vcc^2)/(2*Po) where Po is the output power in watts. If
>the transistor is driven to saturation and has appreciable Vce(sat),
>then Vcc-Vce(sat) is often substituted for Vcc in the formula.
The constant "2" in the 2*Po part of the above formula is usual for
class C operation. The ARRL Handbook, using "K" in place of the "2"
says:
"K = a constant that approximates the RMS current to dc current ratio
appropriate for each class (of operation). For the different classes
of operation:
Class A, K = 1.3
Class AB, K = 1.5 --- 1.7
Class B, K = 1.57 --- 1.8
Class C, K = 2 "
Wes Stewart, N7WS
Bojangle Chicken
> transistor? What will you do with that information if you ever get
> to know it ?
It is not the impedance of the transistor that I care about. It's the
impedance of the amplifier that matters to me. I'd like to be able to get a
match to it. If the amplifier impedance is 10 ohms and I go sticking a
1kohm resistor across the output, I'm definitely *not* going to get optimum
power out.
> The external load on the transistor is simply calculated from the
> peak voltage across the transistor and the peak current (supplied
> by the transistor) flowing through it. R = V / I.
This is pretty much what I did. However, I did not have any books with me
at the lab, nor do I have many at my present location (most of them are 2
hours away), so I was just looking at a quick-n-dirty method of finding the
approximate impedance that I need the load to be to get near-maximum power
out--i.e. I want to get as much of the rf current into the load as I can. I
will not using transformers to match the load.
Thanks for the approximations Roy, Ian and Wes.
J M Noeding
<bojangl...@yahoo.com> wrote:
>> The external load on the transistor is simply calculated from the
>> peak voltage across the transistor and the peak current (supplied
>> by the transistor) flowing through it. R = V / I.
>
>This is pretty much what I did. However, I did not have any books with me
>at the lab, nor do I have many at my present location (most of them are 2
>hours away), so I was just looking at a quick-n-dirty method of finding the
>approximate impedance that I need the load to be to get near-maximum power
>out--i.e. I want to get as much of the rf current into the load as I can. I
>will not using transformers to match the load.
>
>Thanks for the approximations Roy, Ian and Wes.
>
Thanks to Wes for a good formula explanation, but doesn't a device
have lower impedance on input and output when frequency approaches
maximum freqency, or is it just caused by the fact that voltage swing
and gain is lower?
73
Jan-Martin
LA8AK
Wes Stewart
The formula that Roy first presented and I expanded on computes an
optimum resistive, i.e. non-reactive value for the output load
resistance. In an r-f (or any AC) situation, it normally assumes that
the reactive parts are dealt with by some type of tuning so that the
net reactance is zero.
The formula does not address the input at all, nor does it address the
gain of the device. It simply calculates the value that delivers the
required power output.
In other words, if (sometimes a big if) you can drive the device hard
enough to cause the peak-peak voltage swing to occur at the drain,
collector or anode, then with the calculated load resistance you will
achieve design power output.
To clarify further, this value is the resistance that needs to be
presented to the device output terminal. It is not the resistance
that is presented by the terminal to the outside world.
Wes Stewart, N7WS
>
>73
>Jan-Martin
>LA8AK
Reg Edwards
external load is adjusted to equal the internal resistance you are
certain to blow it up by exceeding the peak voltage rating.
----
Reg
Roy Lewallen
where the transistor's ft is typically much higher than the operating
frequency. At VHF and above, the transistor's S parameters are usually
used for designing input and output networks.
Roy Lewallen, W7EL
Bojangle Chicken
very near future...I can almost 'see' it.
By the way, using the method that I originally described (the ki0eg method)
I obtained about 80 ohms resistive. According to the calculations, this is
about right. Also, measuring the voltage (rms) transfer, I found the output
to be related to the input by a factor of 5.1. For the entire measured
range, the Vout/Vin plot was a straight line (very straight). I made the
assumption (I know it's not realistic) that the voltage transfer worked at
any voltage, which allowed me to calculate an output power of 20W. I've got
a 10W 80ohm load that gets unbearably hot dissipating the output signal, so
I'm quite sure that I'm getting more than 10W, but probably less than 20W.
The point of all this is that when I decreased the load resistance by a
factor of 2 (i.e. Rload = 40ohms) I got an increase in output power that
roughly corresponds to somewhere between Class AB and Class B using the
Rload = Vdd^2/(kPo), where k is variable between 1 and 2. This means that,
although the formula seems to be off by a constant in my case, it's roughly
correct. It's nice to see the 'theory' work out in practice (although I'd
be willing to bet that this formula was found experimentally, sorta).
Thanks to those who responded. The amplifier that I came up with works just
fine [as tested] into any load between 40 and 80 ohms (producing between 10
and 20 watts over the range of loads). It's really comforting to see a
"design from scratch" work outright. I guess I chose the right major
afterall!
Thanks.
Pete
ki0eg
Ian White, G3SEK
>S-parameters are something that I'm going to become familiar with in the
>very near future...I can almost 'see' it.
>
used for small-signal design, where linear behavior can be assumed. In
other words, only for class A.
>Thanks to those who responded. The amplifier that I came up with works just
>fine [as tested] into any load between 40 and 80 ohms (producing between 10
>and 20 watts over the range of loads).
This is a slightly unusual application, where the load can vary so
widely. As previously mentioned, for an amplifier designed to work into
50 ohms you should use the device manufacturer's recommended load
impedance Z_load. They also state a design value for input impedance
Z_in.
Having said all that, the techniques that you'd use to design the input
and output networks are pretty much the same for both S-parameters and
Z_in/Z_load so you don't have to learn two different sets of skills.
Clifton T. Sharp Jr.
> I've got
> a 10W 80ohm load that gets unbearably hot dissipating the output signal, so
> I'm quite sure that I'm getting more than 10W, but probably less than 20W.
Considering your knowledge, it seems silly to ask, but... you're sure the
80 ohm load isn't inductive, right?
--
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Cliff Sharp | Hate spam? Take the Boulder Pledge! |
| WA9PDM | http://www.zdnet.com/yil/content/mag/9612/ebert9612.html |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Bojangle Chicken
> >very near future...I can almost 'see' it.
> >
> - though probably not for PA design. S-parameters are generally only
> used for small-signal design, where linear behavior can be assumed. In
> other words, only for class A.
I understand that part. Someone mentioned s parameters...that's just
something that I'd like to be able to use. Most of my designs up to this
point have demanded linearity, also, which means that s-parameters work. Do
you think that the s-parameter approach would give approximately correct
results, even for large-signal analysis?
> >Thanks to those who responded. The amplifier that I came up with works
just
> >fine [as tested] into any load between 40 and 80 ohms (producing between
10
> >and 20 watts over the range of loads).
>
> This is a slightly unusual application, where the load can vary so
> widely. As previously mentioned, for an amplifier designed to work into
> 50 ohms you should use the device manufacturer's recommended load
> impedance Z_load. They also state a design value for input impedance
> Z_in.
That's the beautiful thing...it was my design goal to have at least 10W
output into a "load." When I asked what the load had to be, I was simply
told "just make it output 10W" (up to 25W max). The amp is being used by a
graduate student at my school for [he hopes] his thesis project, which
involves ionizing a flash tube. The amp will basically be operating into an
inductor, coiled around the tube. I imagine he's going to need impedance
matching for proper transfer.
I will consider the thoughts presented more carefully when designing more
mainstream rf systems, but approximations and good-enoughs were fine for
this project. Thanks to those who replied once again.
Pete
ki0eg
Wes Stewart
>Bojangle Chicken wrote:
>>S-parameters are something that I'm going to become familiar with in the
>>very near future...I can almost 'see' it.
>>
> - though probably not for PA design. S-parameters are generally only
>used for small-signal design, where linear behavior can be assumed. In
>other words, only for class A.
Generally true. While I've left the industry, I still have a few trade
magazines dribbling in. I don't study them as intently as when I was working,
but I have noted that there are now available computer-controlled tuners used to
make high-power (relatively speaking) load pull measurements, that in
conjunction with a lot of calculations can deliver "high power" s-parameters.
Maury Microwave is one company that has info on this.
Wes Stewart N7WS
Ian White, G3SEK
>> >S-parameters are something that I'm going to become familiar with in the
>> >very near future...I can almost 'see' it.
>> >
>> - though probably not for PA design. S-parameters are generally only
>> used for small-signal design, where linear behavior can be assumed. In
>> other words, only for class A.
>
>I understand that part. Someone mentioned s parameters...that's just
>something that I'd like to be able to use. Most of my designs up to this
>point have demanded linearity, also, which means that s-parameters work. Do
>you think that the s-parameter approach would give approximately correct
>results, even for large-signal analysis?
>
and how far it departs from the small-signal regime.
>> >Thanks to those who responded. The amplifier that I came up with works
>just
>> >fine [as tested] into any load between 40 and 80 ohms (producing between
>10
>> >and 20 watts over the range of loads).
>>
>> This is a slightly unusual application, where the load can vary so
>> widely. As previously mentioned, for an amplifier designed to work into
>> 50 ohms you should use the device manufacturer's recommended load
>> impedance Z_load. They also state a design value for input impedance
>> Z_in.
>
>That's the beautiful thing...it was my design goal to have at least 10W
>output into a "load." When I asked what the load had to be, I was simply
>told "just make it output 10W" (up to 25W max). The amp is being used by a
>graduate student at my school for [he hopes] his thesis project, which
>involves ionizing a flash tube. The amp will basically be operating into an
>inductor, coiled around the tube. I imagine he's going to need impedance
>matching for proper transfer.
>
Been there, done that, but only with tubes. The impedance matching
problems for that application are horrible, because the load is
completely different before/after ionization. I hope the devices have
good VSWR withstanding capability.
>I will consider the thoughts presented more carefully when designing more
>mainstream rf systems, but approximations and good-enoughs were fine for
>this project. Thanks to those who replied once again.
>
>Pete
>ki0eg
>
>
73 from Ian G3SEK Editor, 'The VHF/UHF DX Book'
Ian White, G3SEK
The same kind of thing is available for low-level applications such as
exploring noise figure contours versus the load supplied to the input.
The problem for high-level applications is that the resulting S-
parameters are both frequency- and power-dependant. This presents
something of a network design problem...
Bojangle Chicken
for the load. They are all in parallel. I have absolutely no doubt in my
mind that the load is not purely resistive. I'd be more surprised to find
that it didn't have stray inductance (and especially capacitance) than to
find that it did have those qualities. However, the blocking capacitor that
I used should shift the phase enough to present a nearly resistive load the
output. I basically designed the load by measuring peak power to the load
(Vrms^2/R). Once I had the load, I chose the blocking capacitor for maximum
power transfer (again). In this way, I figure I got close enough.
Pete
Clifton T. Sharp Jr. <cli...@clifto.com> wrote in message
news:3AEF0A5D...@clifto.com...
J M Noeding
<G3...@ifwtech.com> wrote:
>Only as a starting-point. How approximately will depend on the device
>and how far it departs from the small-signal regime.
>
>
data books often shows a Smith Chart presentations of R+jX (G+jB) for
input and output, but I am curious if this is available for a mosfet
device like IRF630 on the HF bands? Suppose the device is designed for
power supplies, but may perform well at least on 40-160m bands. I have
some hundred devices and would like to try some experiments
73
Jan-Martin
Roy Lewallen
it's likely you can find a SPICE model. What I've done is build a SPICE
model that imitates an S parameter measurement setup and use it to get
the S parameters. In fact, for one microwave transistor, I found pretty
bad disagreement between the SPICE-derived S parameters and the
published S parameter values -- and using the SPICE-derived values gave
circuit modeling results which were closer to actual performance. I'd
think this would be unusual, though, since S parameters are measured
pretty directly, while SPICE parameters are inferred from other
measurements. I imagine that the manufacturer's S parameter test fixture
was defective, improperly calibrated, or improperly used.
Roy Lewallen, W7EL
Rex Allers
wrote:
>I don't know whether you can get S parameter data for these devices, but
>it's likely you can find a SPICE model. What I've done is build a SPICE
>model that imitates an S parameter measurement setup and use it to get
>the S parameters. In fact, for one microwave transistor, I found pretty
>bad disagreement between the SPICE-derived S parameters and the
>published S parameter values -- and using the SPICE-derived values gave
>circuit modeling results which were closer to actual performance. I'd
>think this would be unusual, though, since S parameters are measured
>pretty directly, while SPICE parameters are inferred from other
>measurements. I imagine that the manufacturer's S parameter test fixture
>was defective, improperly calibrated, or improperly used.
>
>Roy Lewallen, W7EL
Hmmm...
Just for future reference, can you tell us what that microwave
transistor was?
-Rex, KK6MK
Roy Lewallen
package parasitics furnished by NEC for their comparable package than
the ones furnished by Siemens. More recently, I discovered that Siemens
had made some pretty substantial changes in their package parasitic
model, and the new ones work much better. I don't know if they've redone
the S parameters.
The accuracy of the various models was based on how well the model
agreed with actual operation characteristics in a particular
application.
Roy Lewallen, W7EL