1 kW RF power LDMOS destruction test by bad SWR

Den onsdag den 17. september 2014 14.58.03 UTC+2 skrev Jan Panteltje:

On a sunny day (Wed, 17 Sep 2014 06:32:01 -0700 (PDT)) it happened Lasse Langwadt Christensen wrote in :

LOL I like the term 'singe use LED'. Will have to use thicker copper...

Very impressive. My experience is that overdriving the input can cause desrruction too. I would have liked to see how much saturation it can take and also how high the gate voltage can go.

Too bad they are only good for 110 volts. Like most all RF parts, the DC specs are minimal, not even a transfer curve.

My first job interview, I told the guy that I preferred tubes to transistors because tubes were harder to blow up. He said "that won't do" and dismissed me. I said the same thing to the next guy, he laughed, and hired me. I designed about $100e6 worth of stuff for him.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

What's the status of the company that dismissed you?

I was employee #4. When they fired me 12 years later, they had 200 emps and were selling $20M per year, mostly marine automation and supervisory controls for pipelines.

They of course went into slow decline, were bought and sold a few times. Last I heard, they had been acquired by L-3 but may be gone now. New Orleans has great food and cheap land but it's a rotten place for technology.

One of their sillier ventures:

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

That seems typical of companies who cannot recognize talent when it slaps them in the face. Especially when it's their own employees.

New Orleans has a peculiar slave-labor, rich-owner mentality. That's fine for some businesses but not for technology.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Impressive. Even more impressive is that they didn't use silicon grease on the xsistor and didn't torque down the mounting screws. I'm fairly sure the ceramic frame wasn't torqued too much as there are no washers or gouges in the ceramic from previous torquing of the cap screws. The water cooling (red and blue hoses) probably helps some but not with a loosely mounted device.

I was watching the R&S wattmeter display for partial failures. I didn't see any, except in the last test. At 4:09 it shows -24.46 dBm after the short circuit test. The video doesn't show the power before the short circuit test, but the previous test at 3:01 shows -22.33 dBm. That's a power loss of about 2.1dB (40%) which might be some of the parallel transistors inside the device going poof. I compared shots of the test setups and they seemed identical. The power drop might be from a loose connector or a change in the setup, so this is far from conclusive. It may also have gone into oscillation. Dunno.

At the beginning of the video at 0:10, they show a presumably ordinary device going up in smoke, complete with sparks and some flames. I've blown up plenty of RF devices and that's NOT what I've seen. Usually, it's the ceramic lid of the device getting blown off with a loud band and a puff of white smoke. However, I've never blown up an LDMOS device, so that might be different.

Still, it's a rather impressive demonstration for those that mistreat their kilowatt amplifiers.

Incidentally, according to the calibration sticker, the R&S power meter is 3 years out of calibration.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Wasn't the jump from -23 to -24.46 dBm when they turned the supply voltage up from 50 to 55 V? The power display went from 1.20 to 1.25 or so kW at the same time, I think.

Impressive. Even more impressive is that they didn't use silicon grease on the xsistor and didn't torque down the mounting screws. I'm fairly sure the ceramic frame wasn't torqued too much as there are no washers or gouges in the ceramic from previous torquing of the cap screws. The water cooling (red and blue hoses) probably helps some but not with a loosely mounted device.

I was watching the R&S wattmeter display for partial failures. I didn't see any, except in the last test. At 4:09 it shows -24.46 dBm after the short circuit test. The video doesn't show the power before the short circuit test, but the previous test at 3:01 shows -22.33 dBm. That's a power loss of about 2.1dB (40%) which might be some of the parallel transistors inside the device going poof. I compared shots of the test setups and they seemed identical. The power drop might be from a loose connector or a change in the setup, so this is far from conclusive. It may also have gone into oscillation. Dunno.

At the beginning of the video at 0:10, they show a presumably ordinary device going up in smoke, complete with sparks and some flames. I've blown up plenty of RF devices and that's NOT what I've seen. Usually, it's the ceramic lid of the device getting blown off with a loud band and a puff of white smoke. However, I've never blown up an LDMOS device, so that might be different.

Still, it's a rather impressive demonstration for those that mistreat their kilowatt amplifiers.

Incidentally, according to the calibration sticker, the R&S power meter is 3 years out of calibration.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Y'er right. Pg 6 looks like the test setup in the video. Fig 10 on Pg 9 shows the effects of different Vds voltages. There's no curve for 55V, but extrapolating from the 45V -> 50V curves: 10 log((1300-1100)/1100) = 1.8 dB which is sufficiently close to the 2.1dB from the wattmeter. Thanks.

Time: Wattmeter readings:

1:25 22.44dBm 2:09 22.33 3:03 22.33 3:17 Change 50 -> 55 Vds 4:09 24.46

While I was scraping the numbers, I notice that the power supply was showing 50 V at about 10 Amps at various points (2:28 thru 2:38) in the video. That's 500 watts in, which for Class AB, I might expect about 60% final stage efficiency or 350 watts output. That doesn't come close to 1200 watts. At 60% efficiency at 1200 watts out, that's

2000 watts in or 40 Amps current drain. Huh?

Also, the brass mismatch box doesn't look like it could handle 1200 watts dissipation for very long.

Looking (again) at the setup, I noticed that the big terminator is actually an attenuator, with another attenuator and cable connected to the output and going somewhere. My guess(tm) is that it's going to a spectrum analyzer to detect when it goes nuts and into oscillation. Of course, that's not shown on the video.

Color me very suspicious.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Makes me wonder about what kind of heat sink (cold plate) they have in the test fixture. For that matter the whole fixture, operating frequency and more.

?-)

Yeah, there's no way either: it's doing 1200W, or dissipating anywhere near what it should (it would quickly rise to >500C at those power levels).

Curious how the current draw varies with mismatch phase, too.

Well, with a VSWR over 50, it should only have to dissipate like 24W (pretty sure that's not the right ratio, but also not terribly far from it).

If nothing else, operating it at high gate bias into a power supply with no input (or output) excitation will dynamite anything. "Indestructible" is an inherently laughable claim; imagine if they wouldn't even need an Absolute Maximum Ratings section!

Tim

--
Seven Transistor Labs 
Electrical Engineering Consultation 
Website: http://seventransistorlabs.com

On a sunny day (Wed, 17 Sep 2014 08:49:17 -0700) it happened John Larkin wrote in :

I have done half a kilowatt RF with tubes, and it worked. heater current, tube lifetime, driven till the anodes were red. I have done 100W RF or so with transistors, I prefer transistors any time. This transistor seems interesting, not difficult to interface, lower voltage tuning capacitors!!! Les danger of electrocution! Lower voltage coupling capacitors, less insulation distance, less cooling, simpler construction. NO AGING.

For very high power transmitters these days they use many of these transistor RF modules in parallel, if one goes the transmitter keeps working.

For sure those very high power medium wave tube transmitters are a lot simpler....

On a sunny day (Wed, 17 Sep 2014 09:43:03 -0700) it happened John Larkin wrote in :

I have seen that thing somewhere:

Den torsdag den 18. september 2014 09.26.32 UTC+2 skrev Tim Williams:

seemed very clear to me that "indestructible" was in quotes, meaning much more rugged than you'd expect from an RF transistor

this one with a regular blf578 will do around 500W from 2-100MHz with

1Vpp input, power supply limited to ~1ms bursts and 10% duty cycle and can be biased for class-A

-Lasse

On a sunny day (Thu, 18 Sep 2014 08:39:30 -0700 (PDT)) it happened Lasse Langwadt Christensen wrote in :

1.5 kW dual MOSFET:

I was considering this, until i found thsoe big 1KW transistors: "

Mosfet Linear Rf Amplifier.htm" 800w Mosfet Linear Rf Amplifier.htm

My goodness, spaces in a URL, whats next

Den torsdag den 18. september 2014 17.59.46 UTC+2 skrev Jan Panteltje:

it gets easier at the lower frequencies, I think there is a kW design somewhere with cheap irf510 mosfets

-Lasse

Yep, especially with no silicon grease applied. Well, they could have polished the base but I don't see any evidence of that. The problem with such long and narrow device packages is that they warp in production. Getting them to lay flat by simply torquing the end screws usually doesn't work.

Offhand, it would seem that the full power was applied for maybe 15 seconds at a time. At 1200 watts out and 60% efficiency, that's 800 watts that needs to go somewhere. I would give it about 5 seconds to smoke, even with water cooling.

Incidentally, whenever I try water cooling, there's water everywhere and a very big radiator somewhere. I didn't see any puddles or radiators in the video.

Yep, and notice that the changes in current are NOT smooth with adjustment of the mismatch box. Kinda looks like it's oscillating at various points. Too bad they didn't show the output on a spectrum analyzer. It's impressive that the device can survive such oscillations and instabilities as the increased power dissipation at higher frequencies usually helps destroy the device.

Ummm... Infinite VSWR means that all the power that's suppose to be dissipated in the load, gets bounced back to the output device. So, if the output device is dissipating 800 watts due to the 60% efficiency, and now has to dissipate an additional 1200 watts that would normally be dissipated by the load, it's now burning the full

2000 watts in the device. Depending on line length, the voltage or current through or across the output devices could double.

That's a nifty gadget inside the brass box. On the Smith chart, it goes around the outside edge, in a constant VSWR circle. That means it's going from shorted to open with one adjustment. I wonder what's inside the box. I could do it with a transmission line but that's not adjustable with a single rotary knob.

Yeah, but that's not what NXT is selling. They're suggesting that if you design it right, and build it well, a properly designed amplifier will survive most forms of load abuse. Cooking the device by changing the operating bias voltages isn't quite fair.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558