Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

donate to eham
   Home   Help Search  
Pages: Prev 1 [2] 3 Next   Go Down
  Print  
Author Topic: Transistor specs at different frequencies  (Read 9348 times)
KM1H
Member

Posts: 3508




Ignore
« Reply #15 on: January 12, 2018, 03:41:01 PM »

Many if not most 100-200W 160-10M discrete transistor rigs make less power on 10m be they 12V bipolar or 28V FET; yet they are rated as 2-30 mhz.

Carl

Logged
G3RZP
Member

Posts: 165




Ignore
« Reply #16 on: January 13, 2018, 08:43:35 AM »

Quote
Many if not most 100-200W 160-10M discrete transistor rigs make less power on 10m be they 12V bipolar or 28V FET; yet they are rated as 2-30 mhz

Very true, Carl, but is this caused by the active device or the matching transformers or other parts of the PA circuit?
Logged
KD0REQ
Member

Posts: 2087




Ignore
« Reply #17 on: January 13, 2018, 09:55:01 AM »

well, in hollow-state days, the negative feedback to prevent parasitics had a lot to do with it.

on semis, if you get a spec that shows effective gain against frequency (most often in op amps,) you will see either a straight line down, or a flat line to some frequency and then a downward slant. not being a semiconductor applications engineer, and barely able to spell it, on a discrete device this would probably be due to internal factors like interdie capacitance and skin effect across the junctions running into less and less skin to effect. probably compensation issues in a complex circuit like a packaged op amp. I don't think you'd see anything like quantum leakage at any ham frequency in common use.

if ZRP, being one of those engineers, doesn't have the answer at hand, it's a cosmic dark secret.

// afterthought // our comp sci class in the early 80s had a lecturer from Univac, who specialized in failure analysis, show us photomicrographs of failed semis, where they dissolved the epoxy off with methylene chloride. he said sustained overdissipation (too dang hot for too dang long) would melt the junctions, where peak use failure (100 watts from a 10 watt device spikes) would go open. so if the "typical" graph shows 10 watts, and this means some devices will not hit 10 and some might provide more for a short time, don't plan on goosing the device past recommendations.
« Last Edit: January 13, 2018, 10:17:14 AM by KD0REQ » Logged
KM1H
Member

Posts: 3508




Ignore
« Reply #18 on: January 13, 2018, 02:03:39 PM »

Quote
Very true, Carl, but is this caused by the active device or the matching transformers or other parts of the PA circuit?

Good question Peter that I dont have an answer to.

OTOH the rigs Ive tested using an amp module do not appear to have that problem and are equal power to 6M. Granted that is a small sample and includes SDR.

Carl
Logged
WD4HXG
Member

Posts: 325




Ignore
« Reply #19 on: January 27, 2018, 09:09:55 AM »

 Motorola years back sold the
MRF-750, MRF-752 and MRF-754 UHF transistors which were
optimized for the 400 - 470 MHz range. Working on the assumption
that silicon in general  provides added gain as you go down in
frequency, the lead engineer in the group pressed the devices into
service at 175 MHz. They worked - sort of! When tuning the
transmitter chain one had to be careful to not set L & C to
values that the device did not like. When we met with the Motorola
rep later and inquired about the increased sensitivity to LC
component value changes at lower frequencies we were provided
the revelation that the devices actually had internal matching
that was already optimized for the limited UHF range.

If a vendor does not characterize a device at an alternate frequency
it is usually for a good reason. Think about it, why would I market
a transistor as being for 430 to 470 MHz if it would also work at
other frequencies without being a squirrel? Marketing guys would be
apoplectic.

On another thought process, building and smoking a SO-223 case transistor
that costs $0.15 each is not to much of a budget buster if you have the
time anyway. However when buying a $200.00 MRF1K50xx series device
I will find a way to characterize the device before blindly trying trial and
error design. Scattering Parameters are a good tool to tell you if you are
about to build an oscillator.

73

Chuck
Logged
WB6BYU
Member

Posts: 17482




Ignore
« Reply #20 on: January 27, 2018, 01:59:15 PM »

Quote from: WD4HXG

... the devices actually had internal matching...




The required matching coils and capacitors get smaller at higher frequencies,
to the point where they can be included on the transistor die.  As you approach
1 GHz you might have several tuned circuits in a multi-stage amplifier (especially
in lower power chips) included on the die, with specs stated for gain reduction
provided in other nearby bands.  I've seen GaAs die less than 1mm square that
contained just coils and capacitors for a duplexor, as well as larger chips that
had a 4-way combiner, including quarter wave phasing lines, all on the chip.

Generally, chips designed for use up to 180 MHz or so will probably work well
at HF, but higher frequency parts are more likely to have built-in tuned circuits.
Logged
KM1H
Member

Posts: 3508




Ignore
« Reply #21 on: January 27, 2018, 03:21:32 PM »

Quote
Scattering Parameters are a good tool to tell you if you are
about to build an oscillator.

73

An amplifier is simply a failed oscillator.

Carl
Logged
WD4HXG
Member

Posts: 325




Ignore
« Reply #22 on: January 29, 2018, 02:54:54 AM »



The required matching coils and capacitors get smaller at higher frequencies,
to the point where they can be included on the transistor die.  As you approach
1 GHz you might have several tuned circuits in a multi-stage amplifier (especially
in lower power chips) included on the die, with specs stated for gain reduction
provided in other nearby bands.  I've seen GaAs die less than 1mm square that
contained just coils and capacitors for a duplexor, as well as larger chips that
had a 4-way combiner, including quarter wave phasing lines, all on the chip.

Generally, chips designed for use up to 180 MHz or so will probably work well
at HF, but higher frequency parts are more likely to have built-in tuned circuits.

These devices were recent release (1987) silicon bipolar which as you pointed out
used internal tuning. I was low man on the totem pole and my misgivings
about using the devices at 174 MHz were pushed aside based on prior
success of other engineers. Motorola only provided data at 400 to 470 MHz.
While we had metrology to characterize the device at 174 MHz the time crunch
per management did not allow the luxury of "baby step designing." The two
extra weeks on the back end to just get the chain to a point one could find
a stable operating point and the required addition of a circulator to prevent
it seeing an impedance that would allow it to take off and oscillate were swept
under the rug.  That was the last time I let someone back me into a corner
without the front end work.
Logged
AF6LJ
Member

Posts: 469




Ignore
« Reply #23 on: January 29, 2018, 05:04:49 AM »



The required matching coils and capacitors get smaller at higher frequencies,
to the point where they can be included on the transistor die.  As you approach
1 GHz you might have several tuned circuits in a multi-stage amplifier (especially
in lower power chips) included on the die, with specs stated for gain reduction
provided in other nearby bands.  I've seen GaAs die less than 1mm square that
contained just coils and capacitors for a duplexor, as well as larger chips that
had a 4-way combiner, including quarter wave phasing lines, all on the chip.

Generally, chips designed for use up to 180 MHz or so will probably work well
at HF, but higher frequency parts are more likely to have built-in tuned circuits.

These devices were recent release (1987) silicon bipolar which as you pointed out
used internal tuning. I was low man on the totem pole and my misgivings
about using the devices at 174 MHz were pushed aside based on prior
success of other engineers. Motorola only provided data at 400 to 470 MHz.
While we had metrology to characterize the device at 174 MHz the time crunch
per management did not allow the luxury of "baby step designing." The two
extra weeks on the back end to just get the chain to a point one could find
a stable operating point and the required addition of a circulator to prevent
it seeing an impedance that would allow it to take off and oscillate were swept
under the rug.  That was the last time I let someone back me into a corner
without the front end work.

TRW back in the early eighties did the same thing. We were building L and S-band transmitters.
We had a predriver we could use on both bands. The DPA and PAs used hand tuned transistors that ran between $300.00 and $400.00 a copy. The base and collector circuits were series tuned, tuning was painstaking and if you were sloppy one or more of those devices would let go that would get you a lot of attention.

I remember reading a Motorola application note that described their "Controlled Q" technology.

We had gone through the trials and tribulations of CTC being bought up by TRW, then TRW being bought up by Motorola. ..It seems the people who were responsable for making the devices we used would get laid off or something....
There was always a huge struggle for our suppliers to re-learn how to make the transistors we were buying up at the rate of 200 or more a month.
Logged

Take Care
Sue,
AF6LJ
G0HZU
Member

Posts: 130




Ignore
« Reply #24 on: January 30, 2018, 02:04:36 PM »

Quote
Scattering Parameters are a good tool to tell you if you are about to build an oscillator.
Agreed. I measured the s parameters of a little jellybean BC547B on my VNA a while back and the data suggests it can still be configured to go unstable (and oscillate) up above 600MHz if operated at a reasonably high Vce. I don't think many people would expect that result.

I also used the VNA to model classic little JFETs like the 2N4416A or the 2N3819 and these can go unstable and oscillate  up towards 1GHz.
« Last Edit: January 30, 2018, 02:11:00 PM by G0HZU » Logged
AA4HA
Member

Posts: 2571




Ignore
« Reply #25 on: January 30, 2018, 05:14:13 PM »

While we had metrology to characterize the device at 174 MHz the time crunch
per management did not allow the luxury of "baby step designing." The two
extra weeks on the back end to just get the chain to a point one could find
a stable operating point and the required addition of a circulator to prevent
it seeing an impedance that would allow it to take off and oscillate were swept
under the rug.  That was the last time I let someone back me into a corner
without the front end work.

That seems to be a rite-of-passage for anyone who has done circuit design work while trying to meet budgetary or time constraints. I too was burned by the management-engineers who had forgotten the pointy end of a soldering iron.
Logged

Ms. Tisha Hayes, AA4HA
Lookout Mountain, Alabama
Free space loss (dB) = 32.4 + 20 × log10d + 20 × log10 f
N3QE
Member

Posts: 5214




Ignore
« Reply #26 on: January 31, 2018, 06:53:30 AM »

An amplifier is simply a failed oscillator.

And here I was, I thought it was the other way around! HI HI.
Logged
WD4HXG
Member

Posts: 325




Ignore
« Reply #27 on: February 05, 2018, 07:03:00 PM »


I also used the VNA to model classic little JFETs like the 2N4416A or the 2N3819 and these can go unstable and oscillate  up towards 1GHz.

During my intro years on the bench I learned that 22 ohm resistors and
ferrite beads were my best friends. The 22 ohm resistor in the drain of
a FET amplifier often calmed the savage beast's oscillations. I was also
known to crush ferrite beads, mix the shards with polyurethane and
apply to the metal case of devices which seemed intent on shipping
with a technician's finger. More than one engineer accused me of
practicing voodoo.

Then there was the design dumped in your lap were the sales team
at a show sold a device which did not exist. It was only a cut sheet
of a device envisioned by someone. There is no NRE budget and it is
slated for delivery in 30 days. Never mind the parts needed have an
eight week lead time.

After six months I gave up on antacids.   Cry


Logged
G3RZP
Member

Posts: 165




Ignore
« Reply #28 on: February 06, 2018, 01:49:47 AM »

Seems that we all have similar 'war stories' and none of them reflect very well on management, sales or marketing!
Logged
KX4OM
Member

Posts: 260




Ignore
« Reply #29 on: February 07, 2018, 09:00:13 AM »

As to the "rule of thumb" I've seen several statements that said said the RF frequency should be 1/10 the fT frequency from the data sheet. In other words, with an fT of 400 MHz, the transistor could be used at 40 MHz. I've also seen 1/6 and 1/3 rules of thumb, and that implies that the supposed "rule" is not linear.

Some data sheets simply list the use frequency, such as 27 MHz for CB power amplifier BJT transistors like the 2SC799. That data sheet says it is "intended for use in citizens Band communications equipment up to 30 MHz." The rating in the summary is f=27MHz, Vcc=12v, output power 3.5W, power gain=9.4dB typical. No fT is specified in the sheet, and that is typical of several CB power transistors, including the 2SC2078 and 2SC1678. Dave Benson, K1SWL gave me those as recommended replacements after I blew the PA during initial testing of my 1st SW-20+ rig several years ago (I bought several just in case.)

One datasheet for the audio power transistor BD139 list the fT as ">50" and I've seen it specified in homebrew transmitter articles for 80m and up to 20m. Several manufacturers do not list the fT, but Philips does list the fT as 190 MHz typical for their version.

I was doing some recent reading on the frequency affects on transistors, and one of the most important factors is the device's internal capacitance. The increasing frequency and the capacitance affect is a roll off of gain of the device. My interpretation; comments from others will be very welcome on that.

Ted, KX4OM
Logged
Pages: Prev 1 [2] 3 Next   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!