Brick amps - a learning experience....

Ideas, technical topics, help and discussion for ATV enthusiasts
Forum rules
This forum is run by the BATC (British Amateur Television Club), it is service made freely available to all interested parties, please do not abuse this privilege.

Thank you
Post Reply
radiogareth
Posts: 831
Joined: Wed Jan 06, 2016 9:46 am

Brick amps - a learning experience....

Post by radiogareth » Tue Jul 06, 2021 10:37 am

A while back I built a brick amp for 4m (&6m) as it would provide plenty of RF for our experimental DATV extension. Like a lot of the newer MOSFET modules it had the weird stepped hold down lugs and needed sanding flat for proper heat transfer to our usually flat (as perhaps custom die-cast commercial) heatsinks. Not wanting to spend hours over a piece of emery cloth I enlisted the help of my Lidl belt-sander. It even came with fixings to mount it upside down for use as a linisher. With care the back of the module was soon reduced to a continuous copper surface so I progressed the amp. It worked fine, as expected, for months and I think was used in the 2020 contest on 4m. Fast forward to more recently, I once again set up for tests with my local station, Martin G4FKK. Strange, no RF amplification present? What the driver was putting in is what was coming out. Somewhat puzzled (as no magic smoke had appeared) I put it aside for future investigations. I always thought the plastic case was irretrievable bonded on, but finding that they can actually be easily removed (and with nothing to lose) I took the cover off, gentle leverage being all that was required. Still no 'smoke egress' evidence, but there were lots of tiny little copper-coloured specs - a bit like what you see inside a big RF mosfet that has expired. Magnifier out and it transpired that they were actually copper dust/filings.....so much for a 'sealed module'.
I suspect that at some point, one little granule was jogged into a position to cause terminal harm to the internal circuitry and that was the end of the module.
Not sure what the best solution might be for future modules. Even the emery cloth way will raise particles, perhaps with less vigour and penetrating ability than a powered method.
Builders beware.....
Gareth, £45 the poorer if I replace the module....
Last edited by radiogareth on Thu Nov 18, 2021 8:29 am, edited 2 times in total.

G3GJA
Posts: 179
Joined: Sun Nov 20, 2011 3:44 pm

Re: Brick amps - a learning experience....

Post by G3GJA » Tue Jul 06, 2021 12:19 pm

My recommendation is that the back of the module is NOT sanded flat.

The assumption that it needs to be flat is based on the earlier Bipolar modules that were a flat slab of copper. This extract from the Mitsubishi data sheet for the 23cm RA18H1213G module is definitive:

Mounting:
Heat sink flatness must be less than 50 µm (a heat sink that is not flat or particles between module and heat sink may
cause the ceramic substrate in the module to crack by bending forces, either immediately when driving screws or later
when thermal expansion forces are added).
A thermal compound between module and heat sink is recommended for low thermal contact resistance and to reduce
the bending stress on the ceramic substrate caused by the temperature difference to the heat sink.
The module must first be screwed to the heat sink, then the leads can be soldered to the printed circuit board.
M3 screws are recommended with a tightening torque of 4.0 to 6.0 kgf-cm.


Clearly, the heatsink in proprietary applications is not raised to match the module. My interpretation is that the indent is on the base to prevent uneven mechanical stress being transferred from the copper heatsink to the alumina substrate by allowing the heatsink compound to migrate.

I wouldn't be surprised if the sanding flat of the heatsink has caused the copper to flex, which in turn has caused a hairline fracture in the alumina breaking the tracks or the bonding to the active devices. It could well have been worsened by thermal expansion.

The datasheet recommends keeping the module below 60C.

Clive

G4FRE
Posts: 161
Joined: Fri May 26, 2017 6:57 am

Re: Brick amps - a learning experience....

Post by G4FRE » Tue Jul 06, 2021 12:48 pm

I have always sanded the bottom of any modules with a step in them after reading the downeast microwave writeup at http://01895fa.netsolhost.com/PDF/MOSFET%20PA_pdf.PDF and havent lost a single module yet

Dave
G4FRE

G3GJA
Posts: 179
Joined: Sun Nov 20, 2011 3:44 pm

Re: Brick amps - a learning experience....

Post by G3GJA » Tue Jul 06, 2021 6:22 pm

I've seen that N2CEI article before, Dave. Steve's main issue was stability and it's a common problem that I've seen in many homebrew PAs using the Mitsubishi modules.

The MOSFET modules blow up with instability faster than you can switch them off. Bipolars are much more rugged. The best example was a dual M57762 design that would light a neon if you touched its leads on the amplifier's ground plane!

The stability can be fixed without grinding the base by using a pair of large solder tags (yellow sleeved circular terminal crimps with the sleeve removed are good) fitted under the mounting bolts. The crimp end should then be soldered directly to the ground plane of the PCB.

With a good CPU heatsink paste and mounting bolts that are tapped into the heatsink there is more than adequate cooling. The gap is only 9 hundredths of a millimetre after all.

Clive

M0YDH
Posts: 239
Joined: Sat Dec 31, 2016 5:59 pm

Re: Brick amps - a learning experience....

Post by M0YDH » Mon Jul 19, 2021 4:59 pm

Dear Clive
That method of construction agrees with all the Mini-kits amplifier kits that I've followed and built. I suspect Mark at Mini-kits sells a lot of Mitsubishi brick amps which always come with detailed notes.
I suspect a cardboard aperture mask of the right size will get the prescribed amount of heat sink paste in place.
I hope that Mini-kits gets a UK agent so we can resume getting them.

73
David M0YDH

User avatar
g7kpm
Posts: 37
Joined: Sat Feb 23, 2008 9:00 pm

Re: Brick amps - a learning experience....

Post by g7kpm » Mon Jul 19, 2021 11:07 pm

Just to add my twopenn'orth...

I fully agree with Clive. I have built quite a few PAs using this module over the years with kits from various suppliers such as Minikts, Steve G6ALU and homebrew PCBs. The early advice advocated by some was to sand the back flat, however, most of the ones I have done this to have failed prematurely. The ones which have not been sanded have never failed and the last one built is in use 24/7 on GB3GG, so my advice is with Clive and don't sand it. The construction details outlined by Clive are also good advice and best practice construction wise.

73s, Jason G7KPM

G3GJA
Posts: 179
Joined: Sun Nov 20, 2011 3:44 pm

Re: Brick amps - a learning experience....

Post by G3GJA » Tue Nov 16, 2021 11:42 pm

Help! RA18H1213G module urgently needed.

I've now got a problem on GB3EY that has an RA18H1213G in the Tx chain where it used as a driver with circa 400mW RF output. The amplifier has fluctuating gain when first switched on and then permanently loses 3dB until it cools down again.

The amplifier was bought as a finished housed amplifier from Bert PE1RKI. On removing the module I find that the flange has been machined flat and there's a patch of the original plating in the middle that suggests it's not as flat as it could be. There's also the copper dust that Gareth mentioned down in the recess between the flange and the plastic case.

It looks as though the issues Jason has seen with sanded flat flanges have been repeated in EY's Tx. Any help to locate a module would be welcome. Enigma and Anglia have no stock.

TIA

Clive G3GJA

SkyVision
Posts: 7
Joined: Fri Jan 24, 2020 1:51 pm

Re: Brick amps - a learning experience....

Post by SkyVision » Thu Nov 18, 2021 2:40 am

Hi Clive

Personally I'm a sander

Working on two-way radios it's common to pull off a brick, the module isn't flat, poor quality compound is used and air bubbles underneath
Clearly the manufactors don't care, but they don't use extra bias for linearity, not usually continual TX either
If you need one try eBay item number:132779456026
I've always found Ebay seller xihu888 to be fairly reliable and his goods have generally been quality, maybe surplus or seconds at a guess

You can expect reduced output at high temperature so the only solution it good cooling
Sanding, Cooper Heat Spreader, good quality compound and large heatsink with fan might help
Old broken solar inverters are a great source of large heatsinks, try your local solar installer or scrap recycler
Go for the Plain Aluminium ones and not the black painted ones

Good luck Roger VK5YYY

G3GJA
Posts: 179
Joined: Sun Nov 20, 2011 3:44 pm

Re: Brick amps - a learning experience....

Post by G3GJA » Thu Nov 18, 2021 10:01 am

Hi Roger

The datasheet for that module says it requires a flat heatsink to within 50 microns. There no mention of needing a specially formed surface matching the indent in the flange. It would have added significant tooling costs to incorporate the indent into the flange so clearly it was done for a purpose. It would have increased costs of the radio manufacturer too, if they had to make a heatsink with a raised pad or alternatively sanded it flat.

From that it seems to me that it is a deliberate attempt to reduce stress on the substrate by allowing the heatsink compound to distribute evenly within the 0.23mm gap. That gap is small enough for a good compound to provide adequate thermal transfer.

It is significant that of the two I have in use that the one that has failed has been sanded and the one not sanded continues to work. Jason G7KPM has also seen two examples of failures of sanded units.

The amplifier in question was made be Bert PE1RKI. It is housed in case milled from a solid billet of aluminium which is attached to a 200 x 150 x 50mm heatsink with a fan attached. See pictures is CQ-TV 265. The amplifier used as a driver at only 400mW RF out although at a fairly high bias level to keep in linear, so the module is not stressed and the whole amplifier is sufficiently engineered for 24/7 repeater use in my view.

Thanks for the info on the eBay seller. I'll take a look.

73 Clive

Post Reply

Return to “General ATV Discussion”