Measurements on IRFP32N50K Mosfets

Welcome and feel free to ask questions about class E, mosfet's, and AM.
You must be registered to post.

Measurements on IRFP32N50K Mosfets

Postby WA2WVL » Thu Dec 22, 2005 2:08 pm

I am building a Class E amplifier using 8 or 10 IRFP32N50K. This FET looks to be a good one for the application. 32A, 500V, 460W and very low reverse capacity. ( $6 ea)
I checked 15 parts and the following measurements were made:

Cin (static no voltages) 6150 pf +- 10
Rin @ 3.8 Mhz = .92 ohms
Rin @ 7.2 Mhz = 1.13 ohms
Rin @ 14.2 Mhz = 1.39 ohms

Rin will drop some at large signal due to C charge and discharge but these are good numbers to design with.
Rin was measured using a high Q "L" network matching to 50 ohms.
Rin does not include the capacity.(only bonding wire and skin effect R)
Capacity is tuned out by the "L" network.
Last edited by WA2WVL on Tue Jan 10, 2006 10:19 pm, edited 1 time in total.
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby steve_qix » Thu Dec 22, 2005 6:28 pm

It might work out.. What did you figure on a DC drain voltage? I figure about 22 to 25 (unmodulated carrier) volts to maintain the recommended 100% safety factor. 70 volts or thereabouts max positive modulation peak.

You should be able to run about 4 amperes max. with that FET and not exceed the current rating on modulation peaks.

So, probably about 100 watts input per device.... maybe 90 watts out.

I looked at MOSFETs with similar ratings when I was evaluating what would be the most bang for the buck MOSFET. The FQA11N90 was the choice :-) With the FQA, You can run 45 volts at 1.25 amperes - about 56 watts input, 50 watts output per device. So, a pair will run about 111 watts input - 100 watts OUTPUT, and the cost for the two of them is $5.60 (at the time $5.00 for two) The advantage is you're running at about 1/2 the impedance for slightly more power...lower losses in many areas, and simpler modulator [and filter] design. The disadvantage is you're using two devices instead of one, and the voltage is higher, so potentially larger components are needed.

I'm curious how you make out. Let us know your operating parameters.
All ears!! :wink:

Talk later and Regards,

Steve
Visit the class E web site at:

http://www.classeradio.com
User avatar
steve_qix
Class E Guru
 
Posts: 434
Joined: Fri Jan 31, 2003 11:03 am
Location: Townsend MA

Postby WA2WVL » Thu Dec 22, 2005 7:56 pm

Steve, your internet site is great. I am learning a lot.
I am not expecting to run AM. On CW or SSB peak drain voltage will be about 80 volts at 25 amps. I think the amp may run class D on 160 and class E on 40 and maybe 20. It is push/pull and will be powered with a full bridge switching modulator also using the IRFP32N50K device. This device has a reverse transfer capacity of only .8% of the input C. The FAQ11N90 is 1.1%. The first pair of IRFP32N50K I ran was well saturated with 3 watts and gave an easy 300 watts at 40 volts (with 50ma of bias/Fet) at 3.8Mhz. Then I ran out of power supply.

I have previous experience with EER. From one pair of MRF150's, 600 watts PEP at 90%, -35 db IMD.
I have built an efficiency metering circuit which will drive a front panel meter. A manuscript was submitted to QEX.

Keep up the good work and I will be reporting on my progress.
Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby WA2WVL » Thu Dec 22, 2005 8:31 pm

Steve:
There seems to be no full analysis available on the internet on Push/Pull Class E amplifiers. Do you or anyone elce know of any info on this?
I would not expect the turnoff voltages to be as high as in the classic single ended design.
Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby frank carcia » Thu Dec 22, 2005 10:13 pm

Floyd,
Yes you will see similar peak drain voltages on the drain with a push pull class e. Vrey cool on the MRF150s. I have a pile of them that I want to turn into a linear someday. The problem with the bigger power supply fets is lead inductance. I started with IRF250s in the 80s which have a smaller die but I think single gate bond wire. Good luck keep us posted.
So far the Fairchild fets Steve found provide the biggest bang for the buck.
It is about time for a better one to emerge after what 3 or 4 years now.
I found a cool linear for 6 meters send me an email address and I will send you a copy of the article. It is at work so can't send it till 1/3/06
gfz
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby WA2WVL » Thu Dec 22, 2005 11:11 pm

Frank:
Nice article in QEX. Five IRFP's in parallel measure 31,000 pf. I belive I have a solution for the bonding wire problem which would allow large devices to run in parallel at much higher frequencies. I am going to attempt 20 meters with this bandswitchable technique.
Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby frank carcia » Fri Dec 23, 2005 8:07 pm

The 6 meter rig uses IRF520s so a lot of smaller fets in parallel. It is two 500 watt blocks. I think each block has 8 FETs push pull.
I've seen a number of class e rigs have random failures and suspect gate inductance or drive related. I went through a stage of failing fets for a while in the 160 meter rig. Once the drive was made stiff enough all my problems went away. I used series 3 ohm resistors on each gate but thought that added too much delay for 75 meters.
It will be interesting to see how you cure this problem. Good Luck!

Thanks for the kind words on the article. Now to find the time to finish the modulator. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby WA2WVL » Fri Dec 23, 2005 10:00 pm

Frank:
I have been doing some calculations based on a loaded Q=3.1 and I get 5600pf/phase and 34A circulating thru Cout and C1. (at 1500Watts)(3ohms load/side. I went back and read your QEX article for the 10th time and my numbers seen close to what you ended up with.
The key to driveing large FETs in parallel at frequencies above 80 meters is to use a current xfmr at each FET gate to source with their secondarys in series. (current drive) This eliminates the parallel resonance formed by the bonding wires and parallel strap. I matched 4 IRFP460's at 14.2 Mhz up to 25 ohms using 1:2 turns including a blocking cap (.1mf) to allow me to apply bias.
I sent an inquiry to IR asking about the bonding inductance and their reply was that the IRFP32N50K might not work above 1 Mhz so I explained Class E to them. They are going to make the measurements.

The caps to use for C1 are ATC 100E. (up to 20a per cap) if you can afford them. (Q>10,000)
Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby frank carcia » Sat Dec 24, 2005 12:07 am

Floyd,
Interesting on the drive but how did you deal with the connections between transformer primaries. I suppose strip line. I can see how you use the transformers to reduce loop area around the FET.
I reduced the area with planes but that drove the input Z very low having all those gates in parallel. I tacked the input transformer in with short wires during initial testing. Later I soldered the two parts with a solid wide joint. The input swr went up meaning the transformer actually wanted a bit of series L. I was able to dial it back in by using the small inductor at the far end of the gate traces.
I found I need higher shunt cap values than most guys. The lower values didn't produce the efficiency. I tried changing in 500 pf steps and the values I arrived at work pretty well. I find the gate looks cleaner with the higher value. Heck the value is going down when it warms up and there is a lot of current in the shunt C. I also find the gate looks cleaner as the L/C tank ratio gets right. My Q is still a bit high. When we first tested the rig at Steve's house the extra caps were not installed. the rig wasn't very stable above 750 watts out. Now I can load it to 1200 at 46 volts. RDS on starts effecting efficiency at that power level but it is stable.
It will be cool to see your method, fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby WA2WVL » Sat Dec 24, 2005 1:43 pm

Frank:

The one turn winding that drives the FET is ferrite loaded brass tubing.
(such as used for input xformers in AB amps.) available from RF parts.
They are not connected togeather. The 2t secondarys is a continuous piece of #22 teflon wire wound thru all transformers. No pc board. The gate lead does not have to be short because it and the gate bond wire inductance is tuned out at the 50 ohm point.

Assuming the IRFP32N50K has an input impedance at 3.8Mhz of 1-j7then the 2 turn secondary looks like 4-j28. Putting 5 of these in series gives 20-j140. The leakage inductance of this transformer string tunes out the input capacitive reactance and the gate bonding lead. This frequency comes out in the 10-15Mhz range. (20+j0) To tune it higher a series C does the job. To go down to 40 meters a series L works. Since this is current drive to the FET if each FET has the same input impedance then they have the same voltage. Looks like any size FET could be driven this way.
I am orderind some BN-43-202 Balun beads from Amidon to build up a 5 FET transformer drive assembly. They are .55 in wide and the to-247 is about .625 in wide. 1/8 inch brass tubing will be used with the #22 teflon wire.
Is the design clear now?
Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby WA2WVL » Sat Dec 24, 2005 1:52 pm

Frank:

The IRFP32N50K has about 1/7th the on R of the Fairchild part and looks to be a little faster. I think it has a little brouther in the to-220 package.
Floyd
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby WA2WVL » Sat Dec 24, 2005 2:30 pm

Frank:

The little brother in a to-220 package is the IRFB20N50K.
20A
500V
.25 ohms
2870 pf in
320 pf out
34 pf fb

Might drop in for the Fairchild part. (no I think the FQA part is to-247)
Price unknown but available from Newark.

Floyd
Last edited by WA2WVL on Tue Jan 10, 2006 10:07 pm, edited 1 time in total.
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby WA2WVL » Sat Dec 24, 2005 5:53 pm

Frank:

When you had instability in your amp was it low frequency?
I tried a 100 uh choke in the output centap and I got a low freq oscillation.
(my single push/pull pair). I am estimating 20-26 db gain at 80 volts so oscillation is likely without swamping. I would be happy with 15db.

On the 6 meter amp was the efficiency mentioned?

Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby WA2WVL » Sat Dec 24, 2005 5:58 pm

Frank:

IR has a still larger part. 47A 500V but only 550 watts dissipation.
Many small parts would probably have better reliability.

Floyd
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby frank carcia » Sat Dec 24, 2005 8:14 pm

Floyd,
Yup I understand your transformer. I have the 6 meter thing at work so will send it after 1/3.
The instability was like a modulation riding on the gate and drain. I never checked the frequency because I was afraid of blowing it up. When a increased the series C to reduce power the problem went away. So by increasing the drain Z the amplifier tamed down. Adding more shunt C allowed me to run at a lower Z. Reducing the Q increased the efficiency.

The nice thing about the Fairchild devices is the voltage rating allowing a 1:1 isolation transformer as a power transformer. I used IRF840s in the 160 meter rig with a peak voltage of about 80 volts. This put me at about 300V peak on the drains. The higher voltage just made things easier.

I see guys having random failures and think the door knob shunt caps must get hot and reduce in value making finals go unstable being over loaded.
A number of people have asked me about the inductor in the center tap acting like a current source. My response is the transformer primary is doing the same thing. I have also seen articles with an inductor
maybe it matters but I have not seen it. The cores I'm using Chromerics
A637s are about 1 uh each so I have 8 uh on a side. the value drops with current but seem like there is enough. I would like to try the next bigger core though.
I worked with the guy who got the patent on the flat matrix transformer.
I argued with him that the RF guys have been using that method before he did it in power supplies. That came to mind when you talked about your method. I'm running the connections in my mind to reduce leakage z
in the secondary. I guess a hunk of PC board at each end of the core would support it and line up the holes for the primary wire.
My C.T. is 20 .005 UF caps in parallel with 4 orange drops. So a fair ground for 2 ohm input. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.


Return to Class E Discussion

Who is online

Users browsing this forum: No registered users and 1 guest

cron