KH6HME 144 beacon

[n3...@aol.com]

For those of you not keeping up with the situation on the big island.

Local complaints about the 144 "CW" beacon caused them to turn it off. Chip has built a new one that uses WSPR that will TX every sequence but on the beacon frequency. It was mailed last week and if all goes well be on the air this week.

If you look at the map you will see me with 2 instance of 144 WSPR, one on the standard frequency and one on the non standard.

The 432 beacon had been removed and was held up in the mail. After investigation it was determined that the PA module had given up. So hopefully that is going to be repaired and returned to service.

As for me I've been off VHF but am slowly getting back into it. My Kenwood gave it up, so I added some goodies to my old faithful FT736 and have been stirring it up on 6. If all goes well I'll be causing QRM on 144 and 222 soon. Have 144 monitoring WSPR and FT8 right now, but that has always been a noisy band for me.

I have plans for a 144 TX on Mt. Otay but things are moving slowly.

73

Chris

[Bo, OZ2M]

Hi Chris

Interesting story with the 70 cm power module. Do you have any idea why the module died? At OZ7IGY we are using Mitsubishi power module from 23 cm to 6 m. The one for 70 cm has been running continuously, PI4 + CW FSK and carrier i.e. RF with 100% duty cycle, outputting 25 W since April 2013. Of course temperatures here are a notch or two lower than in Hawaii I can imagine.

Do you know when the 70 cm power module was manufactured? From the pre-RoHS period I know for a fact that the 4 m module had problems with a cracking substrate due to uneven bimetallic bending of the flange and substrate when the temperature changed. This caused the supply track to the first MOSFET to break. See more here http://rudius.net/oz2m/70mhz/pmpa_kitdocs.htm at the bottom of the page.

Another theory, but not mine, is that the bonding wires are subjected to more thermal stress when running CW OOK than CW FSK.

I am looking forward to the operation of the new beacon :-)

Bo

[Glenn Elmore]

Bo,

This is Glenn, n6gn, rather than Chris replying. I may not be much closer to understanding the failure as I have only spoken with Fred, K6IJ, about what he found when he brought the beacon down from the mountain.

He discovered that the PA was putting out only .75W and that the drive was 20 mW. That is the correct drive level - about what I had set it to for ~25 watts output.  The module which failed was the 30 watt Mitsubishi RA30H4047M which I had used to replace the previous 60W module that is similar.  I bought it from RFParts about 3 years ago but have no way of knowing when it was actually fabricated.  I changed to the 30W module  because the beacon runs unattended and also because KH6HME beacons are a guest at the site and someone else is paying the energy bill!  I wanted it to be reliable. This also had the benefit of cooling everything down a bit. I had wanted to avoid relying on a fan to keep the thing cool and dropping the input power helped that. As I sent it, even with no fan which Fred *did* add, under normal operation it temperature rise was no more than 15-20C. I used a big heat sink and maximum air temperature at 2500m is not over 30C so I wasn't stressing the part thermally. Also it was not over-voltaged, as far as can be determined since it ran from a common, shack-wide regulated supply which powered other beacons.

The beacon runs WSPR, JT965, JT9 and CW so is in intermittent service. The heat sink is large so there wasn't a lot of thermal cycling over the 10 minute frame in which all those modes transmit.  Antenna match was good as far as can be determined.

In short, I can't think of anything about the application that would explain failure after 18 months. I was hoping and expecting the knd of lifetime you have experienced.  We simply didn't get it and from the symptom it sounds like the output devices stopped operating and the drive power that made it through to the antenna is what Chris was spotting last fall. Not bad to copy less than 1W at 4100 km on 70cm though.

Though design caution causes me to call "one failure a pattern" I really don't know how to explain this.

I hope the replacement will last much longer.

Glenn n6gn

--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/2-meter-wspr/56682623-6b53-42fa-8a24-0ab80e1c4233%40googlegroups.com.

[Bo, OZ2M]

Hi Glenn

Well, two tings

- Coffin Manson thermal stress on package / bondwire

- Substrate-flange bimetallic effect

If this was the cycle used

there are periods with full current in the bonding wires. Then off. Then ~45% duty cycle during the CW, ... vs a constant current. It saves DC power but puts Coffin Manson thermal stress on the bond wires. So perhaps a constant output level would have saved the module, i.e. CW FSK instead of CW OOK and an offset carrier in the other pauses. Mitsubishi uses a transparent and liquid resin to protect the bond wires so perhaps it is possible to see if they are still intact.

If you still got the module it is worth checking if the supply track to the first stage has a crack in it. If so it is easy to fix. I would, however, not put it back into operation on such a remote location.

Bo

[Glenn Elmore]

Interesting thoughts. Although the part was, as I remember, biased part way towards class A so 'off' wasn't  zero current, it was certainly hard keyed by the U3S source which had no shaping and did generate noticeable key clicks. That stress continued 6 times every 10 minutes so that conjecture makes sense. It would likely have been more stressful than, say, SSB. 

It would have been simple, and still is, to change it from CW to FSK but conditions have changed for me since I built it. About 2 1/2 years ago I moved 1500 km to the East to the state of Colorado and am now very far from California and any QTH that will be likely to hear the beacon.

 All of the beacons at the KH6HME site are maintained by Fred, K6IJ, as   a project of love in memory of KH6HME who began it many years ago.  The trip up and down to the mountain top is a couple of hours.  Fred has indicated his intention to replace the part but I don't think he feels comfortable reprogramming it or doing much more than getting the PA working again in the manner it was and there seems to be nobody else on the Island of Hawaii ready to take on more technical support.

So if RFParts does in fact supply a replacement, perhaps we'll find out whether this was a one-off 'random' event or  evidence of one of the failure mechanisms you suggest.  In any case, as worthwhile and interesting as it might be, I doubt that the part will get carefully examined to better understand what happened.

I had also built a 2m version of that beacon that Chris, N3IZN, is interested in putting on from Southern California, near the border with Mexico. Presently it has only a 10W PA but it is running an almost identical program so maybe in a year or two we'll learn more. Hopefully I'll get around to shipping it out to him for deployment there. My WSPR activities have turned away from VHF and above in the last few years so I'll not likely be involved more.

Best,

Glenn n6gn

--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/2-meter-wspr/6a65bda5-ad24-4989-a910-0179d6d9ece1%40googlegroups.com.

[Bo, OZ2M]

OK, the high bias might have helped to reduce the thermal differences.

One more thing I forgot to mention yesterday. There is a small difference, 90 um, in the height of the flange on the Mitsubishi power modules. At the beginning we didn't do anything about this in the 4 m power module kits. But after the substrate crack we decided to include a piece of thin copper sheet to place in the "groove" which is actually about 80% of the contact area between the flange and the heat sink. I have seen other solutions where the flange is milled flat. But not everybody has the ability to do this.

Bo

[Glenn Elmore]

I can't remember the detail of the flange when I built it but I know I
did my best to make good thermal contact, even with the 30W device and
don't recall any concern about the bulk of the flange. I think I would
have noticed and at least been concerned about over-tightening the
mounting if I thought it would stress the substrate but it's now been
too long for me to remember any details. I would have used a thin layer
of compound and done my best to ensure good mating with the heat sink.

I'll CC Fred to consider this when he replaces the part.

Thanks,

Glenn

[Bo, OZ2M]

The power module flange "issue":

Instead of milling sanding may work out like here by Lars, OZ1BXM: http://www.oz1bxm.dk/1296/PA-Mitsubishi-RA18H1213G.html fix. 7 and fig 8. 5000 grit is easy to find here and perhaps for the final touch toothpaste, ~10 000 grit, on a glass plate.

Bo

[Glenn Elmore]

Fred has indicated that

"the flange has a even surface, and when I removed the chip from the heat sink it was super tight against the heat sink"

and that he expects to be installing the replacement soon.

Glenn

--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/2-meter-wspr/dd44230f-41ae-4ed2-8e38-2855e0d9bb17%40googlegroups.com.