eHam

eHam Forums => HomeBrew => Topic started by: AJ4SN on November 14, 2014, 06:08:29 PM



Title: rugged replacement for IRF24ZN
Post by: AJ4SN on November 14, 2014, 06:08:29 PM
I built the 50 watt amplifier designed by Donald Huff, W6JL, for the Homebrew Challenge. It was featured in the June 2010 QST. The amp has been a great addition to my shack serving to "QRO" my Flex 1500 as well as other homebrew low power rigs. The problem that I've had is that occasionally I blow an output transistor, the IRF24ZN. The transistors are very inexpensive, but it is, of course, an inconvenience to change them out. They blow as a result of operator error (wrong bandswitch, wrong antenna, etc.), and they are not forgiving. I've blown a transistor just by bumping the keyer when the antenna was not connected. In this months QST, I saw a review of the Hardrock-50 which mentioned that it used RH16HHF1 MOSFETs as finals. The article stated that the amp needed no fault protection because the transistors would withstand a short or open circuit on the output. It made me wonder if there might be a replacement for the IRF24ZN that might be more rugged. I'd be happy to pay a little more for a transistor if it would help idiot proof the amp. Anyone have suggestions beyond finding a smarter operator?


Title: RE: rugged replacement for IRF24ZN
Post by: AJ4SN on November 15, 2014, 05:28:01 AM
Sorry for any confusion, but the correct part number for the MOSFET is IRFZ24N.


Title: RE: rugged replacement for IRF24ZN
Post by: KB1GMX on November 15, 2014, 05:42:29 PM

The problem is that the amp is designed around the transistor/MOSFET.
You cannot drop something in and expect better.  Dropping RD16HHF
transistors into the Huff design is likely to be troublesome.

Any device that more "rugged" is likely to have very different characteristics and that
changes the whole game.

I've looked at that design and it mat the cheap criteria but there were better designs
for example the WA2EBY (K5OOR HF packer) is one I built and have found very robust.
I've done terrible things to it and its still on the first set.

There is also the hardrock50.

Then again if your willing to risk 10$ worth of transistors it might work.


Allison


Title: RE: rugged replacement for IRF24ZN
Post by: N7EKU on November 16, 2014, 07:39:38 PM
Hi,

I know in some QRP finals, they protect the transistor from overvoltage by using a zener doide across the output if the transistor.  I don't know if that is possible or advisable on higher power amps, but perhaps some experts could chime in here?

Another protective option may be some current limiting, either by fuse or active circuit, that would limit the maximum current that could be supplied to the finals?

Just some ideas...

73,


Mark.



Title: RE: rugged replacement for IRF24ZN
Post by: K8AI on November 17, 2014, 03:13:29 PM
Allison, other than the fact that the RD16HHF1 has a different pinout and will possibly give the amp higher gain and less power out, what trouble would be expected from the substitution? I wouldn't think to expect the input and output impedances to be so different as to warrant any changes to the circuit at all.

Curt, K8AI


Title: RE: rugged replacement for IRF24ZN
Post by: KB1GMX on November 17, 2014, 03:47:00 PM
Using Zener would add capacitance across the output devices and alter the behavior.
That only is effective if the failure mod is over voltage, if it over current then its meaningless.

Current limiting is good if you know the max current.  I'd assume a fuse is there to protect
the batter/power source and wiring.

There are three ways to kill that amp, over voltage (no load at all),  Usually a high enough
voltage device is sufficient to protect against that.  Fr a 13.8V supply a 50V device is usual
adequate.

Another is over current (terrible SWR or shorted feed) the key here is a device that can
stand the worse case current draw.  (IRFZ24 is a 17amp 45W device)

Transmitter over drive is generally deadly to power MOSFETS, they have a limited tolerance
to voltages applied to the gate, usual max is 20V and can be lower.  If you stay under 5W
in all cases this is not possible.

The usual killer is one of the above combined with instability,  If the amp is unstable due
to layout or component choice it can oscillate and "take off" destructively.    Many of the
Amps that use non-RF devices (IRF510, IRFZ24, and a few others) if not designed carefully
or grounded well tend to do this.   Small changes in layout, even long leads can be catastrophic
in some cases.

More robust transistors require different circuit.  For example the Hardrock uses 4 RD16HHF and
most of the circuit is different. To drop 4 RD16HHFs into the current circuit your using you will
spent at least 20$ on transistors and you will not get the same power out.  Using 4 devices will
require a different heatsink and mechanical layout as well.   Using only 2 will insure less than
50W or stressing them to get 50.

The basic hardrock design is the Veatech amp publish in the Aug 2010 QST.  

The there is the WA2EBY amp published in the March 1999 QST and the April issue.  It uses
the still widely available IRF510.  I've built this and it also exists as a kit (HFpacker).  Mine is
run off AC power using a 28V power supply and its plenty of power, good even on 10M and
I haven't fried it yet (though I've done all the bad things to it.).  The board for the original
version is available from FAR circuits.  Hint don't cheat on the heat sink Mnne is 4 x 8  inch
with  1.5" tall fins.

The IRFZ is a 55V 17A 45W device, where the RD16HHF is a 50V 5A max.  I suspect there is something
more going on that its is fragile.  Places to look, heatsinking, running at high temps (flange temp
exceeding 70C) make for higher failure rate.   Bad grounding or not enough ground points in the
assembled circuit can lead to feedback paths that are not predictable.  Lead lengths or other layout
variations from the authors unit that create feedback paths. An example of that is the source lead has to
as short as possible.  Cut the drain lead off and use a solder lug from the flange (under the the insulator)
 to hook to the transformer. The transformers used are critical and should be copied as closely as
possible.  Others would be voltages or bias that was not part of the design for example running
a 12V design at 20V to get more power.  


Building even a 10 or 20 watt amp using modern solidstate devices has to be treated as if it were
for VHF as most of the devices used will work at VHF and the high currents make ground something
to be insured using through wires and wrapping the edges (two sided board).

Allison




Title: RE: rugged replacement for IRF24ZN
Post by: KB1GMX on November 17, 2014, 04:09:36 PM
Kurt,

The devices are significantly different.  The output capacitance and input
capacitance are different.  most other parameters are very different.
Keep in mind the IRFZ24 was really designed for switching power supplies
and pulsed motor drivers.  One is HEXFET (IRFZ) and RD is A flavor of
TMOS or LDMOS, event eh construction is differnt.

Also the higher gain of the RD would likely contribute to instability if the
layout does not allow for it..  An example of this is the Hardrock has
negative feedback to keep the gain in check, the Huff design does not.   
I've used RD16HFs at 50Mhz and they will go higher. 

Look at the Hardrock design then the Huff.  The Huff uses two deivces
and the hardrock uses 4 to get to 50W.  Reason for this is the RF part has a
higher saturation resistance and lower peak current.   Even the output
transformer is different  Huff is 1:3 turn winding (9:1 impedance) and the
Hardrock is 1:4 turns (1:16 impedance)  Considering the load is 50 ohms
1/16th (3.125 ohm primary) is very  different than 1/9th (5.56 ohm primary).
That is a reflection of the IRFZ and the RD parts differing load lines.

Like I said for 10$ you can throw two in and see...


Allison


Title: RE: rugged replacement for IRF24ZN
Post by: N7EKU on November 17, 2014, 10:41:39 PM
Using Zener would add capacitance across the output devices and alter the behavior.
That only is effective if the failure mod is over voltage, if it over current then its meaningless.

Current limiting is good if you know the max current.  I'd assume a fuse is there to protect
the batter/power source and wiring.


Allison

Hi Allison,

Yes I was thinking about both the over-voltage and over-current modes of failure -- that's why I was thinking about those two different methods.

Just thinking "theoretically" I was wondering if current limiting could be done on the collector/drain supply like can be done with power supplies.  I'm not sure if in the end it would be all that protective if the max current needs to be set high enough to cover peak current needs, that maybe too high for protection over long drains that high?

Yeah the zener would add more to the total capacitance from drain to source for sure.  I see on QRP rigs they talk about taking this into account when choosing the collector/drain choke (making sure the combo resonates about 1/2 the operating frequency).  But then I read others saying the combo should resonate AT the operating frequency -- so I'm mixed up there that's for sure.

In the posters amp, I was wondering if something else was going on too.  Maybe the heat sinking is not good enough -- check that there is good transfer from the transistor to heatsink (excellent flat physical contact + good heatsink grease.)

Or the amp is oscillating.  It may be worth it to check with an oscilloscope or wavemeter to see if there are some other strong oscillations during normal/low/high loading conditions?

Just more ideas,


Mark/n7eku





Title: RE: rugged replacement for IRF24ZN
Post by: G3RZP on November 18, 2014, 12:03:57 AM
Another point to consider is skin resistance of the "ground plane". The usual 1 ounce/square foot printed circuit board material has copper 0.0014 inches thick: 2 ounces is 0.0028 inches thick.

Skin depth is the depth in the copper where the current has dropped to roughly 37% of the surface current, and for it to be negligible, a rule of thumb is that the material needs to be 5 skin depths thick.

The skin depth in copper is 2.6/Square root of frequency in MHz: this gives the answer in thousandths of an inch. So at 2 MHz, the skin depth is 1.84 thousandths of an inch and so you really need copper 10 thousandths of an inch (or 10 mil in usual US terms) thick. At 30 MHz, skin depth is 0.48 thousandths of an inch......

Thus the 'ground plane' in most cases isn't that good on PCB material, and if you have large currents, you can get problems because of this: a point 'grounded' to the 'ground plane' at the middle of the board and putting a large RF current into that point can actually be at  some potential well above the 'ground' at the edge of the board.

It always surprises me how well people get away with ground planes which are not even one skin depth thick.....


Title: RE: rugged replacement for IRF24ZN
Post by: KH6AQ on November 18, 2014, 05:13:48 PM
Assuming the FET is failing due to overvoltage the IRFB4019PBF might be a suitable replacement.

FET            42         4212
V(BR)DS     55V       100V
Ciss          370 pF    550 pF
Crss          65 pF       35 pF
Coss         140 pF    350 pF
g              4.5 S       11 S

The 4212 has higher input Ciss but the reduced Crss will partiallay offset that. The Coss of twice might require a change in the output circuit.


Title: RE: rugged replacement for IRF24ZN
Post by: N3QE on November 22, 2014, 05:21:28 AM
I too have played around with finals based around non-RF mosfets in WA7EBY/W6JL style amps using IRF510's and derivatives, and the IRF24ZN is pretty much the same class.

The raw volts and amps ratings of the transistors in on-off switching service, are only vaguely related to power ratings you get when you use these non-RF transistors in HF amplifiers.

The "no-antenna" condition is what drives you over a peak voltage limit, results in the transistor shorting out, and then the transistor lets its magic smoke out. Usually these beefy switching transistors have no problem with the "antenna is a dead short" condition.

Some designs use zeners near the transistors to prevent overvoltage, but these add unnecessary stray capacitance. Other designs will shut down final drive when high voltages are present on the final transistors but that's not possible in these simple designs because usually the final is a standalone stage (no way to signal to previous stages that they have to shut down) although they are pretty common in commercial integrated 100W or 200W rigs where the finals are not an afterthought but part of the whole design. Since these 100W or 200W rigs are not using the high-current switching transistors, they have to also consider the shorted antenna fault case, and they often will scale back drive when they see high peak currents in the finals too.

If you read the ARRL review of the Hardrock-50, they note that it is the losses in the low-pass filter that "save" the final transistors from blowing under antenna fault conditions. This is typical of a highly evolved design - the losses in some components in a fault condition, prevent other more expensive components from blowing out under fault conditions.

It's actually funny to read the amplifier section of this website. Folks identify "weak points" in commercial amplifiers, then set out to correct these weak points, but the resulting amplifier is then no longer so evolved and is subject to loss of expensive parts under common fault conditions!


Title: RE: rugged replacement for IRF24ZN
Post by: N7EKU on November 23, 2014, 12:38:55 AM
Hmm,

Over-voltage from "no antenna" or "wrong antenna" or "wrong filter":  Use an SWR type circuit to pick up RF output voltage and rectify it, use this voltage on an FET in-line with the 12V PTT line biased to break the line when this voltage exceeds a determined level.

Over current from "shorted antenna" or "wrong antenna" or "wrong filter":  Use a fuse in the supply line rated a bit above your maximum operating current.

Wouldn't cost to much to try...

73,


Mark.





Title: RE: rugged replacement for IRF24ZN
Post by: WB6BYU on November 23, 2014, 12:45:43 AM
Quote from: N7EKU

Over current from "shorted antenna" or "wrong antenna" or "wrong filter":  Use a fuse in the supply line rated a bit above your maximum operating current.




A ordinary fuse isn't fast enough.  They say a transistor is the fastest fuse on three legs...

That's why Ten-Tec used electronic over-current protection in some of their early rigs.  It was built into the
power supply, which shut down almost instantaneously once the current exceeded a given
threshold.  A standard fuse may take several seconds to blow, especially if you are <25% over
the rated value.

By building it into the power supply, Ten-Tec was able to insert it between the chassis ground
and the negative side of the transformer / rectifier / capacitor bank.  Check out the circuit in
their 251/252 series power supplies.


Title: RE: rugged replacement for IRF24ZN
Post by: KB1GMX on November 23, 2014, 07:55:19 AM
Having built multiple amps for VHF and HF I can say fuses are still good to have.

Over the years I've popped a few fuses due to sustained overload with no damage.
I've also had better than average success with IRF510s.  I know people that pop
them like flies but I've had success with them even on 6M. I do however try to run
them carefully that meaning large heatsinks, boards with many vias for ground
and making sure the amp is stable.

For example mY copy of the EBY amp can easily do 65W on 15M at 32V DC but
that's pushing the dissipation and cutting it fine on voltage,  with the DC running at
27V its only 40W on 15 but to date (now 5 years of use) I've not fried it on any
band with any filter or wrong antenna.  I think keeping the voltage reasonable
and insuring the amp is stable is the solution that works for me.    I also do the
same with my VHF amps as those devices aren't cheap yet I've not fried one
yet despite multiple opportunitites.

With all that said I'd suggest the the first post of the thread...  The amp if built
carefully should be stable and  robust but if it  is unstable try adding a feedback
network  on each FET from drain to gate of .1uf in series with 100 ohms.  The resistor
will need to be a 2W non inductive part.  If the power drops a little then its about right
if not keep lowering the resistor (could be as low as 24ohms) till you see some
 reduction in power.  What this does is reduce the gain of the amplifier and makes
it more stable.

Electronic fusing.  This is good for non mobile cases (can be done but harder).
If the amp is used at home from 120V power use or build a power supply that
can be set to limit on current 10% more than peak current for the amp.  That may
require a bit of testing into a dummy load to determine.  Also if the power source
used at home has poor voltage regulation that can be a source of problem
until corrected.  This is especially true for amps designed for nominal 13.8V.

Last item, Excess drive power.  Must QRP rigs are not a problem with that but
many QRO (100W) radios do spike on transmit some as high as 140W and
that kills fets.  The only solution there is a power attenuator or not using the amp
at all.  While  covering that I've see a few QRP designs that get unstable with
reactive loads and generate spurs or even oscillate and that kind of drive can
be problematic for an amp that is also marginally stable.

One last thing.  Make sure the bias is stable.  I've had bias set potentiometers
get flaky and cause problems.


Allison



Title: RE: rugged replacement for IRF24ZN
Post by: N7EKU on November 23, 2014, 01:26:48 PM
Thanks for the information Allison :-)

Yes, I think the original poster mentioned using the W6JL amp on a Flex -- so I am assuming for multiple bands.

The W6JL was not set up as a multiband amp (it was specifically designed for 40M) and that is maybe where the problem is coming from.  I suggest the poster read the full article on the WA2EBY amp, which was designed for multi-band use.

As such, it has both input and feedback frequency compensation to give it stability and even power output on the HF bands.  Allison's experiences with the durability of the WA2EBY amp I think shows how good these design features are to have.

Just adding the feedback network suggested by Allison would be really easy to do and cheap too.  If it saves time from changing transistors all the time, that would just be worth it in itself I think!

73,

Mark/n7eku


Title: RE: rugged replacement for IRF24ZN
Post by: KB1GMX on November 25, 2014, 11:22:31 AM
Keep in mind the amp was built as a design for price to qualify for the prize
as such it only had to meet the spec of 50W on 40M with a max drive of 5W.

He provided PI network values for other bands but it was not a multiband design
nor did the layout support it well.  The amp does put some power above 20M but
I'd bet the efficiency and general performance is poor because the output
transformer is less than optimum and the input circuit is not at all optimum
above 20M.  Its not the devices fault its a general compromise to meet the
initial goal.

I'd add feedback to the amp and see if it calms down some.

The Veatech (AKA hardrock) took all that into account but was the goal was to
compete for the multiband design goal with considerations regarding stability
and good behavior.

If I were to build either I'd go with the Veatech as a more complete design.
However I'd use a better heatsink and case.

FYI the 'EBY amp the board set is available from far circuits.

Allison


Title: RE: rugged replacement for IRF24ZN
Post by: AJ4SN on November 27, 2014, 05:07:07 PM
Thanks so much for all the valuable feedback. I appreciate the time that you take to help me. I'm always impressed with the homebrew community of ham radio and its willingness to help. I understand now that the amp was designed to minimize cost, and it's not going to withstand lots of abuse. I'm going to try to construct an SWR sensing circuit and see if that will provide the protection against "no-antenna" and "wrong-antenna" situations.