Topics

Alternate MOSFET finals for BitX40 and others

Frank Cathell
 

I'm in the early stages of testing alternate MOSFET finals for my BitX40, and since the basic circuit is similar (although push-pull) in the uBitx, I'm sure the parts are applicable here, at least for 80 and 40M. I have the transceiver operating satisfactorily using an older ON Semi NTP18N06L, which is a 15A, 60V logic level MOSFET.  The RDSon is rated at 100 milliohms for a 7.5A drain current.  This is significantly less than the IRF510. The typical gate charge is 7.3 nC which is slightly less than the IRF510's Qt rating. Being a logic level device, I had to adjust the gate bias to about 2.00 volts, which resulted in about 90 - 100 mA idle current in the MOSFET when keyed for SSB. With 13.0 volts input to the transceiver, the IRF-510 gave me about 7 watts PEP on sideband. The NTP18N06L is giving me a good 10+ watts PEP with the same input voltage. I am using a larger heatsink than was provided with the original kit. I would estimate a surface area of a little over 2.5 times the original heatsink. On extended "ragchews" the heatsink reaches about 60C with no air movement. Down the road I'm planning to test an IRLZ14 which is another 60V logic level MOSFET with an 8 amp rating and half the IRF510's RDSon. I would not recommend trying newer generation logic level MOSFETs because their operation is more non-linear since they are intended as a switching device rather than an amplifying device. The older devices of this logic level type seem to be OK.

Here is a link to the NTP18N06L which Mouser Electronics still carries: https://www.mouser.com/datasheet/2/308/ntp18n06l-d-304903.pdf

Will provide updates on this testing as they evolve.

Frank
W7YAZ
Tucson

ajparent1/kb1gmx <kb1gmx@...>
 

Frank,

You replaced a 100V device with a 60V.  Lowers swr margin. also lower margin for
higher than normal operating voltage.

The input capacitance and output capacitances are much higher than IRF510.
That will make 20M and up power harder to get with existing circuit.

Turn on and turn off delay time is slower than IRF510.

The Rds-on is unimportant unless your running switch mode, for linear you
never approach that value and if you do your far from linear. 

Gate charge is not a factor in liner use.  For switching use its important as the
driver has to supply current to drive that typical drivers push 6-12V int the
gate and then slam it to ground for hard on and hard off.  Far from linear.

Getting air movement on the heatsink is always a good thing.  With the
thermal resistance of the package if the heatsink is 60C the die (the silicon)
will be far hotter in the 120C or higher range.

The thermal die to case flange resistance is higher than for IRF510.

The IRLZ24 has been used in other amps for below 20mhz use.

Despite all that it might work.

Allison
---------------------------------
No direct email, it goes to bit bucket due address harvesting in groups.IO

Frank Cathell
 

The MOSFET tests have been for strictly 40M operation.  Operation on 20M and higher will need experimentation with their respective rigs. The tests are for 13.8 Vdc maximum input with assumed SWRs less than 2:1. 60 volts for this criteria should be adequate. I must disagree with your statements on RDSon and gate charge. Minimizing the former for any type of operation is helpful for heat reduction although it is certainly more important in the switching mode.  Gate charge will definitely be an issue as the operating frequency goes up independent of whether it is in pure switching mode or driven with an analog signal. High gate charge will distort the drive signal to the gate if the driver stage has a higher than necessary source impedance for driving the gate. Either way, the gate capacitance must be over come in a manner that doesn't significantly distort the gate drive waveform. Due to their diffusion properties, the older generation logic level MOSFETs tend to have lower gate charge than their equivalent rated "normal" MOSFETs with a gate threshold of around 4V. Newer technology "Trench FETS" and similar constructions will not typically have linear transfer characteristics when used in an analog mode.

Frank

On Thu, May 7, 2020 at 9:51 AM ajparent1/kb1gmx <kb1gmx@...> wrote:
Frank,

You replaced a 100V device with a 60V.  Lowers swr margin. also lower margin for
higher than normal operating voltage.

The input capacitance and output capacitances are much higher than IRF510.
That will make 20M and up power harder to get with existing circuit.

Turn on and turn off delay time is slower than IRF510.

The Rds-on is unimportant unless your running switch mode, for linear you
never approach that value and if you do your far from linear. 

Gate charge is not a factor in liner use.  For switching use its important as the
driver has to supply current to drive that typical drivers push 6-12V int the
gate and then slam it to ground for hard on and hard off.  Far from linear.

Getting air movement on the heatsink is always a good thing.  With the
thermal resistance of the package if the heatsink is 60C the die (the silicon)
will be far hotter in the 120C or higher range.

The thermal die to case flange resistance is higher than for IRF510.

The IRLZ24 has been used in other amps for below 20mhz use.

Despite all that it might work.

Allison
---------------------------------
No direct email, it goes to bit bucket due address harvesting in groups.IO



--
--------------------------------------------------------------------------------------------
"Thanks for the expensive 7 weeks FREE TRIAL of communism. I'd like to have my LIBERTY back now!" 

Frustrated California Resident

Conservatives want to go back to work. Liberals want to stay home.
Nothing new here!





 

Hi,

Except you quoted the typical gate charge for one device and the maximum gate charge for the other.  Maximum on the one you are testing is 20 something vs 8 something on the IRF510.

But experimenting is fun and it's good you have nice output on this test on 40m :-)

73.

ajparent1/kb1gmx <kb1gmx@...>
 

Repost due to buffer error.

At 40m charge and capacitance is less an issue but real.

RDs-on only occur with far more drive than is normal for a linear application.
If you doing class C for CW it may help.   For linear mode its not even in
the realm of relevant as you never full saturate the device and if you do
its called clipping.

The actual voltage swing on the gate for linear use is in the 1-2V range
centered at Vbias. more than that and things get nonlinear and IMD suffers
badly.

Gate charge for a line application if the driving impedance is suitable (low)
then the waveform is unimpaired again linear not saturate.  

Another factor not considered and is important is forward transconductance
as that is a strong indicator of how much gain is possible.  Too much gain
and instability is also a risk and often a problem.  Power Fets even switch
mode types have gain that has a more/higher frequency vs gain curve compared
to bipolar devices.

The nasty thing to watch for is Drain to Gate capacitance as its both negative
feedback and if the circuit has a phase shift from inductive feed it can go to positive
feedback and oscillation.

Allison
---------------------------------
No direct email, it goes to bit bucket due address harvesting in groups.IO

Frank Cathell
 

I just tested the IRLZ14 MOSFET which is another logic level device. It appears by far to be a better choice than the NTP18N06L device previously mentioned. It was much easier to set the bias of 100 mA with the bias pot (about 1.9 volts) as the setting was MUCH less touchy than it was with the NTP device. This part produced about 11 watts PEP SSB into a 50 ohm dummy load with 13.5 Vdc input to the rig. The heating was also noticeably less under a 3 watt, 30 second CW key down than the NTP device using the same heat sink. Again, this device was tested on 40M only.  The IRLZ14 is $0.87 at Mouser Electronics. Another interesting MOSFET is the FQP13N06L from Mouser.  Will probably order a couple of these at $0.89 apiece.

Frank
W7YAZ

Mick
 

Has anyone considered replacing the irf510 with an irf520? I’ve seen a case of an Irf530 being replaced with an Irf540.
--
 

73
Mick VA3EPM 

MVS Sarma
 

Better we select by low Cgs so that for multibanders it serves at even 10m babd. 


On Mon, 1 Jun 2020, 5:33 pm Mick, <Mgsebele@...> wrote:
Has anyone considered replacing the irf510 with an irf520? I’ve seen a case of an Irf530 being replaced with an Irf540.
--
 

73
Mick VA3EPM 

ajparent1/kb1gmx <kb1gmx@...>
 

the parameter Cgs is not that much a factor for example the BLF278 a very nice 300W
VHF device the input CCgs is 480pf!

Its capacitance from gate to drain is a mere 14pf.  That is important as its negative
feedback, and possible gain limiting factor.

In all cases feeding the gate require power and if you can deliver it.
More Cgs means more power required to impress the needed votlage
across the gate to source can do the job.  However long leads on the
mosfet really hurts as  series inductance in the gate to source path
that limits the available power to the gate to source junction.

So if the driver design is weak not enough power or poor impedance
match the power out will be limited and that pretty much regardless
of the device picked.

Allison
---------------------------------
No direct email, it goes to bit bucket due address harvesting in groups.IO

MVS Sarma
 

Thanks Ajparent. A new point learnt today.
Regards
Sarma vu3zmv

On Wed, 3 Jun 2020, 1:03 am ajparent1/kb1gmx, <kb1gmx@...> wrote:
the parameter Cgs is not that much a factor for example the BLF278 a very nice 300W
VHF device the input CCgs is 480pf!

Its capacitance from gate to drain is a mere 14pf.  That is important as its negative
feedback, and possible gain limiting factor.

In all cases feeding the gate require power and if you can deliver it.
More Cgs means more power required to impress the needed votlage
across the gate to source can do the job.  However long leads on the
mosfet really hurts as  series inductance in the gate to source path
that limits the available power to the gate to source junction.

So if the driver design is weak not enough power or poor impedance
match the power out will be limited and that pretty much regardless
of the device picked.

Allison
---------------------------------
No direct email, it goes to bit bucket due address harvesting in groups.IO