Hi all,
I have another 1296 transverter project in progress which will reside at the top of my tower. I am getting to the stage of building up the PA section.
I bought an RA18H1213G PA module from the NZ distributor for Mitsubishi.
First thing I note was the pin spacings no longer match the batch of PCBs I made for these PAs, the bias pin is wrong along with the supply pin. RF In / Out look OK.
I can get over that problem but being aware of the heatsink flange issues, I got my stainless ruler out to chek the flatness. Wow! I was horrified to see that this one looks to have about 1.5mm gap in the middle with the overall profile banana shaped.
In the past, there has been discussion about sanding off the back so it is flat all over. No way I could do that on this one as I would sand off the mounting eyes..
I will go backto the supplier but anyone else seen this? I am wondering if the disty has been had up by some chinese rip off by chance. I've heard of similiar rip offs of PA transistors.
In the meantime, it's gone back in the static bag until I find out more.
Cheers,
Simon
ZL1SWW
RA18H1213 Heatsink Flange Banana Shaped.
Re: RA18H1213 Heatsink Flange Banana Shaped.
Hi Simon,
Wow that is bad.
I once ran the WIA Equipment Supplies (for 15 years!), in that time we sourced (directly from Mitsubishi Electric) and sold hundreds of the now obsolete M577** & some M677** series Mitsubishi modules. Nearly all would have a slight bend (0.1 - 0.3mm gap) caused by the heat process bonding the hybrid circuit & transistors to the base. A couple of types did have the "banana" problem from batch to batch but usually only 0.5mm most notably the M57716 70 cm 10W SSB was the worst. M57762 23cm generally ok. In commercial transceivers would often see a shim under the main area (where the heatsource is) as a production fix for the 0.1 - 0.3mm gap ranges. Yes a common fix was also to use Wet & Dry to take some material off the ears to flatten out (usually the M57716 as borderline to shim) .. tiresome process! The newer RA types do seem to have the same issue but I
Sounds like you could slide a ruler into the gap under this one! If the module isn't flat enough under the body to successfully shim I'd be contacting Mitsubishi.
Cheers
David VK5KK
Wow that is bad.
I once ran the WIA Equipment Supplies (for 15 years!), in that time we sourced (directly from Mitsubishi Electric) and sold hundreds of the now obsolete M577** & some M677** series Mitsubishi modules. Nearly all would have a slight bend (0.1 - 0.3mm gap) caused by the heat process bonding the hybrid circuit & transistors to the base. A couple of types did have the "banana" problem from batch to batch but usually only 0.5mm most notably the M57716 70 cm 10W SSB was the worst. M57762 23cm generally ok. In commercial transceivers would often see a shim under the main area (where the heatsource is) as a production fix for the 0.1 - 0.3mm gap ranges. Yes a common fix was also to use Wet & Dry to take some material off the ears to flatten out (usually the M57716 as borderline to shim) .. tiresome process! The newer RA types do seem to have the same issue but I
Sounds like you could slide a ruler into the gap under this one! If the module isn't flat enough under the body to successfully shim I'd be contacting Mitsubishi.
Cheers
David VK5KK
Re: RA18H1213 Heatsink Flange Banana Shaped.
Hi David,
Yup she's a bit excessive, I've just popped up a photo be it a bit fuzzy with me tryng to hold the ruler and the phone on macro as well, so it's a bit jiggly. Definitely the worst I have seen and I have used about 3 of these before from the same supplier over the years and those have been OK, and as you say, about 0.2 - 0.3mm gap. I remember the debarcle on a UK site about weather to flatten the back totally. Tried it on one device and that packed up after a while where I suspect broken substrate / tracks. Haven't looked in that one yet as I don't use that PA now.
I have sent an email off to the NZ supplier who are pretty good so we'll wait and see what they say. I did ask them to run a stright edge over the rest of the batch.
I also see that there are two types of package HS and some other variant re pin spacing.
Cheers,
Simon
ZL1SWW
Yup she's a bit excessive, I've just popped up a photo be it a bit fuzzy with me tryng to hold the ruler and the phone on macro as well, so it's a bit jiggly. Definitely the worst I have seen and I have used about 3 of these before from the same supplier over the years and those have been OK, and as you say, about 0.2 - 0.3mm gap. I remember the debarcle on a UK site about weather to flatten the back totally. Tried it on one device and that packed up after a while where I suspect broken substrate / tracks. Haven't looked in that one yet as I don't use that PA now.
I have sent an email off to the NZ supplier who are pretty good so we'll wait and see what they say. I did ask them to run a stright edge over the rest of the batch.
I also see that there are two types of package HS and some other variant re pin spacing.
Cheers,
Simon
ZL1SWW
- Attachments
-
Re: RA18H1213 Heatsink Flange Banana Shaped.
I can't imagine Mitsubishi making a mass produced device with the idea that each and every one of them needs to be sanded before installation. The links on the MiniKits website should help clear up the question of sanding
http://www.minikits.com.au/mitsubishi-pa.htm
http://www.minikits.com.au/mitsubishi-pa.htm
Lou - VK3ALB
Being right doesn't excuse bad behaviour
Being right doesn't excuse bad behaviour
Re: RA18H1213 Heatsink Flange Banana Shaped.
Simon,
That's not unusual for this series of PA modules. I was aghast to find the same "bow" in the module I bought for my 23cm transverter several years ago. I Subsequently bought two more which were the same. Acquaintences of mine who have built 23cm transverters using these modules have noted the same bowed mounting flange. I flattened mine by filing until it was, in fact, dead flat. But TAKE CARE if you do this. For one thing, you will note that the plastic cover is not hermetically sealed like the old bipolar power modules. You don't want copper filings finding their way into the delicate bits. I found this out the hard way.......... So before filing the flange, seal the cover with neutral cure silicone sealant. Another point is that the filing process generates a lot of heat. Have some provision for cooling the module as you go. I used a dampened sponge. Finally, your heat sink must be very flat. Be careful to apply a very even coating of heat sink compound, then carefully "work" the module onto the heat sink until it's properly seated before torquing down the mounting screws. Don't be tempted to overtighten them.
My present transverter has a 50W PA with two of these modules combined. I described the PA in an article in AR magazine some years ago. It's been very reliable, having been used both as my home rig and for field operation. I personally know of several examples that have been built by others. All keep delivering the goods.
Here are some photos. The first three are of an example constructed by David, VK3QM. The fourth is VK3BJM's implementation.
That's not unusual for this series of PA modules. I was aghast to find the same "bow" in the module I bought for my 23cm transverter several years ago. I Subsequently bought two more which were the same. Acquaintences of mine who have built 23cm transverters using these modules have noted the same bowed mounting flange. I flattened mine by filing until it was, in fact, dead flat. But TAKE CARE if you do this. For one thing, you will note that the plastic cover is not hermetically sealed like the old bipolar power modules. You don't want copper filings finding their way into the delicate bits. I found this out the hard way.......... So before filing the flange, seal the cover with neutral cure silicone sealant. Another point is that the filing process generates a lot of heat. Have some provision for cooling the module as you go. I used a dampened sponge. Finally, your heat sink must be very flat. Be careful to apply a very even coating of heat sink compound, then carefully "work" the module onto the heat sink until it's properly seated before torquing down the mounting screws. Don't be tempted to overtighten them.
My present transverter has a 50W PA with two of these modules combined. I described the PA in an article in AR magazine some years ago. It's been very reliable, having been used both as my home rig and for field operation. I personally know of several examples that have been built by others. All keep delivering the goods.
Here are some photos. The first three are of an example constructed by David, VK3QM. The fourth is VK3BJM's implementation.
- Attachments
-
-
-
-
Re: RA18H1213 Heatsink Flange Banana Shaped.
FWIW,
I have a 4 module 1296 amp with the four modules mounted on a fan assisted tunnel heat sink (Jaycar HH-8532 etc), which has a 0.35 degC/W thermal rating (recommended rating is 0.42 degC/W) which has been used successfully on a number of field day contests by a variety of amateur stations.
The modules all had a bow, estimated at 0.3 mm in the centre, but I noticed that when the modules were attached to the heatsink at the recommended torque (0.4 to 0.6 nM on M3 screws, the bowing was reduced by a small amount, taken up by the thermal compound. (I used a feeler gauge which was in thousandths, so not exactly 0.3 mm)
The Mitsubishi application note recommends a minimum of 120 cubic millimetres of compound over an area of 40mm x 6mm which gives a thickness of 0.5mm.
but on the other hand, it talks about a flatness on the heatsink of 50 micrometres!
The picture you have posted certainly suggests that you have out of tolerance packages - it is worth downloading the application notes from Mitsubishi and using them to argue your point for replacements.
Cheers
Peter VK3QI
I have a 4 module 1296 amp with the four modules mounted on a fan assisted tunnel heat sink (Jaycar HH-8532 etc), which has a 0.35 degC/W thermal rating (recommended rating is 0.42 degC/W) which has been used successfully on a number of field day contests by a variety of amateur stations.
The modules all had a bow, estimated at 0.3 mm in the centre, but I noticed that when the modules were attached to the heatsink at the recommended torque (0.4 to 0.6 nM on M3 screws, the bowing was reduced by a small amount, taken up by the thermal compound. (I used a feeler gauge which was in thousandths, so not exactly 0.3 mm)
The Mitsubishi application note recommends a minimum of 120 cubic millimetres of compound over an area of 40mm x 6mm which gives a thickness of 0.5mm.
but on the other hand, it talks about a flatness on the heatsink of 50 micrometres!
The picture you have posted certainly suggests that you have out of tolerance packages - it is worth downloading the application notes from Mitsubishi and using them to argue your point for replacements.
Cheers
Peter VK3QI
Re: RA18H1213 Heatsink Flange Banana Shaped.
Thanks guys for your input.
I like the dual PA AMP pictures, it looks really good! There has been sevral discussion s re these PAs in the past in regard to heatsinking etc.
I am in discussions with the supplier here in NZ who is very good, who in turn has been liaising with Japan re the issue. It seems they are saying that mine is out of tolerance but I will go back and measure more accurately tonight and pass on the details.
It will be interesting to see what the JA's say about it.
Been having email dialogues with Mark VK5EME over the matter as well as he has loads of experience with these units.
Cheers & thanks,
Simon
ZL1SWW.
I like the dual PA AMP pictures, it looks really good! There has been sevral discussion s re these PAs in the past in regard to heatsinking etc.
I am in discussions with the supplier here in NZ who is very good, who in turn has been liaising with Japan re the issue. It seems they are saying that mine is out of tolerance but I will go back and measure more accurately tonight and pass on the details.
It will be interesting to see what the JA's say about it.
Been having email dialogues with Mark VK5EME over the matter as well as he has loads of experience with these units.
Cheers & thanks,
Simon
ZL1SWW.
Re: RA18H1213 Heatsink Flange Banana Shaped.
Simon,
If you don't have some, borrow a set of feeler gauges which would be very useful in quantifying the degree of bowing.
Measure at the centre, and several points each side to confirm uniformity of the bow.
Mark should be able to give you an indication of the degree of bowing to expect, given that he has handled literally hundreds of these modules over the years.
Cheers and good luck!
Peter VK3QI
If you don't have some, borrow a set of feeler gauges which would be very useful in quantifying the degree of bowing.
Measure at the centre, and several points each side to confirm uniformity of the bow.
Mark should be able to give you an indication of the degree of bowing to expect, given that he has handled literally hundreds of these modules over the years.
Cheers and good luck!
Peter VK3QI
Re: RA18H1213 Heatsink Flange Banana Shaped.
Hi Peter,
Yes well I measured tonight with my digital verniers. As you say, I took a sample of measurements and they range from 0.89 mm to 1.12 mm. In the application notes and comments from the disty is that it typically should be 0.4 mm or so. Also have to be careful to measure with the thinnest contact point on the verniers as that affected the readings by averaging over a larger area. Also fingers holding the shaft of the verniers was varying the measurement.
It's pretty hot in Auckland at present.
I have sent off my findings to the disty who in turn is passing on to Mitsi. I initially mentioned 1.2 mm max that raised alarm bells with them so we'll see how 1.12 mm max pans out.
Meanwhile, nothing happening on my Xverter!! FYI, I started with a W1GHZ low cost converter board to see how it worked. It's a low budget one this one.. Interestingly enough, I found the LO supression to be only about 18 dB down from the wanted 1296. Not good. I have got it down to about 25 dB now, by fiddling with the filter element.
I had two boards and swept the filter on the sweep set and find it pretty much follows the published curve when the board is un populated but as soon as MMICs etc are added, the curve changes for the worst. It's now got a retrofit pre PA filter so hopefully should knock out the rest of the 1152. Sweeping that filter shows about a 35 odd dB rejection at 1152.
Re the PA, it's an interesting discussion point of which I hope to get some conclusive facts from the manufacturer. I do the VHF scene column for our national Ham Radio (Break-In) magazine so at least I'll have some material for the next one !!
Cheers,
Simon
ZL1SWW
Yes well I measured tonight with my digital verniers. As you say, I took a sample of measurements and they range from 0.89 mm to 1.12 mm. In the application notes and comments from the disty is that it typically should be 0.4 mm or so. Also have to be careful to measure with the thinnest contact point on the verniers as that affected the readings by averaging over a larger area. Also fingers holding the shaft of the verniers was varying the measurement.
It's pretty hot in Auckland at present.
I have sent off my findings to the disty who in turn is passing on to Mitsi. I initially mentioned 1.2 mm max that raised alarm bells with them so we'll see how 1.12 mm max pans out.
Meanwhile, nothing happening on my Xverter!! FYI, I started with a W1GHZ low cost converter board to see how it worked. It's a low budget one this one.. Interestingly enough, I found the LO supression to be only about 18 dB down from the wanted 1296. Not good. I have got it down to about 25 dB now, by fiddling with the filter element.
I had two boards and swept the filter on the sweep set and find it pretty much follows the published curve when the board is un populated but as soon as MMICs etc are added, the curve changes for the worst. It's now got a retrofit pre PA filter so hopefully should knock out the rest of the 1152. Sweeping that filter shows about a 35 odd dB rejection at 1152.
Re the PA, it's an interesting discussion point of which I hope to get some conclusive facts from the manufacturer. I do the VHF scene column for our national Ham Radio (Break-In) magazine so at least I'll have some material for the next one !!
Cheers,
Simon
ZL1SWW
Re: RA18H1213 Heatsink Flange Banana Shaped.
Hi All,
Been doing some tests, as advised by Japan / local disty.
Put said amount of heatsink goo on and tightened down in the staggered way as suggested.
Still see no change in the bed.
Interestingly, one of my knowledgable cohorts sees that heatsink plane has to have 0.5 micron deviaton as per JA reccomendations.
Measured 1.12 mm at the worst part of the concave.
Noted that pulling off the block off the test plane showed the thermal goo displacement does not even toch the heatsink in the middle.
Really good, I'll be pushing back on this one.
Cheers,
Simon
ZL1SWW
Been doing some tests, as advised by Japan / local disty.
Put said amount of heatsink goo on and tightened down in the staggered way as suggested.
Still see no change in the bed.
Interestingly, one of my knowledgable cohorts sees that heatsink plane has to have 0.5 micron deviaton as per JA reccomendations.
Measured 1.12 mm at the worst part of the concave.
Noted that pulling off the block off the test plane showed the thermal goo displacement does not even toch the heatsink in the middle.
Really good, I'll be pushing back on this one.
Cheers,
Simon
ZL1SWW
- Attachments
-
- It still doesn't pull down
-
- Adding the goo.
-
- It doesn't even touch the mating surface.
-
- Another view, you see it still doesn't pull down.
Re: RA18H1213 Heatsink Flange Banana Shaped.
Has anyone tried a copper shim under their Mitsubishi module?
I am in the process of building the Mini-Kits 23cm amp, and this banana thing on a flat heatsink is proving problematic, in terms of obtaining heat transfer.
My thinking is a copper shim cut to fit between the mounting lugs with thermal compound on each side of the shim, would provide enough 'packing' to allow the module to sit flat on the heatsink.
Your ideas and thoughts are most welcome.
Thanks.
Ron
VK4BVI
(I'm only new at this microwave caper.)
I am in the process of building the Mini-Kits 23cm amp, and this banana thing on a flat heatsink is proving problematic, in terms of obtaining heat transfer.
My thinking is a copper shim cut to fit between the mounting lugs with thermal compound on each side of the shim, would provide enough 'packing' to allow the module to sit flat on the heatsink.
Your ideas and thoughts are most welcome.
Thanks.
Ron
VK4BVI
(I'm only new at this microwave caper.)
Re: RA18H1213 Heatsink Flange Banana Shaped.
There is Mitsubishi data sheet covering this subject ( from a http://www.minikits.com link )ZL1SWW wrote:Hi all,
I have another 1296 transverter project in progress which will reside at the top of my tower. I am getting to the stage of building up the PA section.
I bought an RA18H1213G PA module from the NZ distributor for Mitsubishi.
First thing I note was the pin spacings no longer match the batch of PCBs I made for these PAs, the bias pin is wrong along with the supply pin. RF In / Out look OK.
I can get over that problem but being aware of the heatsink flange issues, I got my stainless ruler out to chek the flatness. Wow! I was horrified to see that this one looks to have about 1.5mm gap in the middle with the overall profile banana shaped.
In the past, there has been discussion about sanding off the back so it is flat all over. No way I could do that on this one as I would sand off the mounting eyes..
I will go backto the supplier but anyone else seen this? I am wondering if the disty has been had up by some chinese rip off by chance. I've heard of similiar rip offs of PA transistors.
In the meantime, it's gone back in the static bag until I find out more.
Cheers,
Simon
ZL1SWW
- Attachments
-
- ANGEN042D.pdf
- (484.58 KiB) Downloaded 232 times
Re: RA18H1213 Heatsink Flange Banana Shaped.
Has anyone thought of bending the heatsink to fit the module. The photo I saw of the heatsinh was just a 1 cm thick piece of aluminium. It wouldn't take to much force to bend that. In adittion you could put saw cuts a few cm apart to help both the bending process as well as increasing the area available for heat radiation.
QED
There is more than one way of skinning a cat.
David
QED
There is more than one way of skinning a cat.
David
Re: RA18H1213 Heatsink Flange Banana Shaped.
I wouldn't recommend that approach as any inaccuracy will result in bending the mounting flange which will bend the PCB substrate (which has earth connections to the flange) which will in turn stress the smd components enough to cause failure. I can recall seeing a aplication note which specifically said not to stress the flange during mounting for this reason.VK3AUU wrote:Has anyone thought of bending the heatsink to fit the module. The photo I saw of the heatsinh was just a 1 cm thick piece of aluminium. It wouldn't take to much force to bend that. In adittion you could put saw cuts a few cm apart to help both the bending process as well as increasing the area available for heat radiation.
QED
There is more than one way of skinning a cat.
David
Best approach, which I have successfully done several times, is to use a belt sander to flatten the back - best done with a fixed bench belt stander or lanisher. You need to take care to avoid overheating, and also be wary of static charge - I solder together all of the pins to reduce the chance of this.
73,
Wayne