Replacing the T4 Tuner regulators with TO-220 devices

[stevebow]

Around a year ago, I modded my T4 regulators/heatsinks per the T4 - Tuners A/B have died thread due to my A/B tuners no longer functional. This involved replacing both 3.3V (LM1117) regulators with TO-252 parts, along with an improved heatsink.

Annoyingly, around 6 months ago, the tuner drop-outs (again A/B) started reappearing, becoming worse and worse over time. I found that by applying pressure to the heatsink, the problem went away, pointing to a physical problem somewhere. Upon close examination of the regulators, I found the solder joint around the 1.25V (LM1085) regulator tab to be microscopically cracked. It's likely the regulator had good enough contact with the PCB when cold to provide power to the tuners, but the joint failed when sufficiently heated by the regulator itself. This was a little frustrating as the initial tuner problem was due to one regulator, now it was the other! (I am not suggesting any other T4 regulator problems are due to a cracked joint, this is just what I have observed in my own unit. If you deep-sleep your T4 as I do, there is considerable PCB thermal stress around the regulars whenever the unit is power cycled.)

While simply re-flowing the solder around the regulator would have fixed this problem, I wondered if my original heatsink solution was adequate. While a better solution than the BW original, there's always the problem that heat transfer from the regulator die to the heatsink is via the regulator's plastic casing - much of the heat was still being sinked to the PCB - had this exacerbated a possible manufacturing fault perhaps?

Dang it, I decided I wanted to do away with the surface mount regulators and go with TO-220 parts instead - and eliminate regulator heat from the PCB entirely. My T4's PCB is quite discoloured around the regulators due to long term thermal stress, and as I deep-sleep my T4 when not in use, who knows what other problems may lurk in the future with continuous thermal cycling of the PCB each time I power it up and down.

I was able to source TO-220 variants of both the original regulators at Mouser. The next step was choice of heatsink - they needed to have an adequate thermal dissipation rating yet fit into the available space next to the tuners. Voltage/current measurements of both regulators showed that the 1.25V (LM1085) regulators were dissipating around 2.1W and the 3.3V (LM1117) regulator around 1.1W. After a little thermal math and (this was the tricky part) finding heatsinks physically small enough, I settled on heatsinks from from Altronics which covered the math with plenty of spare.

As I was going to such trouble, I wanted to add tantalum capacitors to the regulators. The datasheets of both regulators specify usage of 10uF caps on the "In" side and 10uF-22uF on the "Out" side, depending on whose manufacturer's datasheet you looked at. The existing regulator capacitors on the PCB are not marked with any polarity so I suspect they are not tantalum. They look a little on the wee side to be tantalums anyway.

After removing the existing SMD regulators, I piggy-backed 10uF caps on the "In" side (L-R:C738,C762,C708,C715) and 33uF caps on the "Out" side (L-R:C736,C760,C710,C706). I used 33uF caps as I already had these on hand - the usage of 33uF caps instead of 10-22uF is not a problem. The leads needed to be bent a little to ensure the caps themselves did not obstruct the heatsinks or regulators later on. Fortunately, the pin-out of the TO-220 regulators match the existing SMD devices, but the leads needed to be bent a little so that the LM1085s were a little forward of centre and the LM1117s a little rear of centre. This was to ensure that there would be adequate clearance around the mounted heatsinks.

I mounted the heatsinks using silicone thermal pads to electrically isolate the heatsinks from the regulators as the regulator tabs are live. While technically not required, I am happier when heatsinks are not live. Besides, thermal pads have little thermal resistance and the heatsinks used were more than adequate for the job.

Finally, after all done, now to test the efficiency of the heatsinking. This is the worrying part that would hopefully(!) confirm your thermal calculations... Reference (ambient) temperature was taken exterior of the T4 with its case on, around 5cm from the LHS case vent slots. For the A/B regulators, regulator case (tab) temperature was Ambient+41.3° for the LM1085 and Ambient+44.7° for the LM1117. For the C/D regulators, Ambient+37.7° and Ambient+38.8° for for the LM1085 and LM1117 respectively. The maximum operating temperature for both regulators is 125° so the heatsinks are much more than adequate. And I'm rather optimistic that ambient (living room) will remain below 80° in the future. In the end I could have gotten away with smaller heatsinks, but happy with my choice nevertheless.

I have had these regulators/heatsinks running for about 3 months now, with not a single issue noted. Touch wood [/me looks around in panic]. As you would expect, the heat of the T4's case is somewhat more spread out now compared to originally.

Parts used:

2 x TO-220 LM1085 Regulators (Mouser 926-LM1085IT-ADJNOPB)
2 x TO-220 LM1117 Regulators (Mouser 926-LM1117T-3.3/NOPB )
2 x Heatsinks for LM1085 (Altronics H0625)
2 x Heatsinks for LM1117 (Altronics H0635)
4 x 10uF/16V Tantulum Capacitors (Jaycar RZ6648)
4 x 33uF/16V Tantulum Capacitors (Jaycar RZ6665)
1 x TO-220 Silicon Rubber Insulating Kit - Pk.4 (Jaycar HP1176)
4 x M3 Screws/Nuts/Washers

Front View

Heatsinks Front View.jpg (102.22 KiB) Viewed 2230 times

Side View

Heatsinks Side View.jpg (102.04 KiB) Viewed 2230 times

Top View

Heatsinks Top View.jpg (104.35 KiB) Viewed 2230 times

Isometric View

Heatsinks Isometric View.jpg (116.21 KiB) Viewed 2230 times

[prl]

Now that's proper heat sinks

Stevebow's post really highlights that the problem with the thermal cement on these devices breaking down can cause other issues than just the failure of the regulators (and it's possible that the problems may not even always include regulator failure).

I'm not sure when I had this problem whether the regulators under these heat sinks failed, but the problem was more than just a pair of tuners dropping out - my T4 failed to boot at all.

[alfons]

Wow Stevebow,
What a great explanation of what you found and how you went about repairing the heat problem on your T4 voltage regulators. I have the same problem with my T4. i.e. I get frequent pixellation when viewing live TV on either Tuner A or B. I also get the same issue on Tuners C and D if the signal is passing through A and B to get to C and D. If I move the aerial cable to C, both C and D work flawlessly. So at the moment I have disabled Tuners A and B in the Tuner configuration and connected the aerial cable to C > D.

Before pulling the heat sinks off the regulators, I applied downward pressure to the A and B tuner regulator heat sink, and low and behold the picture stopped pixellating. As soon as I removed the downward pressure, pixellation commenced immediately. I repeated this a number of times to satisfy myself that it was either a dry joint or hairline crack causing the issue.

I have a solution I want to try without the need of replacing the regulators and would like anyone interested to comment please (Good or bad).

I am planning on removing the heat sink on the for Tuners A and B, resoldering the regulators back into their existing location and replacing with a bigger heat sink. I also have 2 miniature turbine centrifugal fans (Very quiet in operation) which I plan on mounting in front of each of the 2 heat sinks to provide active cooling. The fans are cheap as chips and should provide ample airflow around the heat sinks to keep the voltage regulating system relatively cooler than no airflow at all. I plan to connect the fans to the hard drives 12 volt power supply rail with an appropriate fuse.

Any comments?

[prl]

The problem with the regulators is chemical as well as thermal. The breakdown products from the thermal cement on the heat sinks can erode tracks on the PCB.

That may be causing what you're seeong when you apply pressure to the tuners.

Just replacing the regulators may not fix all the issues you're seeing.

[alfons]

Hi Peter,
Thanks for your comment. I agree with your comment regarding chemical erosion. Once I remove the heat sinks and the regulators off the PCB, I should have a better appreciation of how bad the problem is. For the record, I am applying pressure on the voltage regulator heat sink to get the clear picture, and not on the tuners.

[Paul_oz53]

Just returning to your point about active cooling, anything that lowers operating temperatures should be good for extending the life of the unit.

If noise is not an issue then fans are fine if you have the skill to tap into the power rails. Bigger heatsinks are likely an improvement but are still dependant on having good thermal contact.

The TO-22Os have an advantage of larger contact area but are intended for through-hole mounting. As replacements for surface mounted regulators you would want to careful not to move the unit too often to avoid any risk of tearing the tracks due to vibration.

Either approach should work provided care is taken to avoid excessive physical movement.

[stevebow]

As prl mentioned, the thermal adhesive has also been a problem for some. I have been one of the lucky ones, I guess. There was very minimal PCB damage due to the adhesive when I removed the original heatsinks compared to others (just a few small dark stains), as can be seen in the photos in the OP.

Before pulling the heat sinks off the regulators, I applied downward pressure to the A and B tuner regulator heat sink, and low and behold the picture stopped pixellating. As soon as I removed the downward pressure, pixellation commenced immediately. I repeated this a number of times to satisfy myself that it was either a dry joint or hairline crack causing the issue.

It would be interesting to know what the damage to you PCB is, if any of the tracks have been damaged by the adhesive, but my money is on a hairline crack like mine. Do you put your T4 into deep standby when not in use?

Just a few comments on removing the original heatsinks. The SMD packages are quite fragile and the adhesive quite strong, so there is a possibility you may crack the 3.3V regulator case when removing the heatsink. Unless you have a hot-air rework unit/station, you might struggle to replace the SMD regulator with just a soldering iron. If you aren't in a position to possibly replace the SMD regulator, you may want to think twice before proceeding.

The adhesive is quite strong a bit of a bitch to remove. I used a single-sided scraping blade and carefully worked my way from left to right. This is because (you may have noticed) that the heatsinks are at a slight angle due the LHS 1.25V regulators being thicker than the RHS 3.3V regulators, so there's a tiny gap on the LHS to work the blade in to.

If you do crack the 3.3V regulator and have the means to replace it, and still wish to use a SMD part, I'd suggest using a case style that matches the other regulator (TO-252), so any heatsink fitted will mate flush to the devices. See my posts in T4 - Tuners A/B have died for how that went.

Good Luck!

[stevebow]

The TO-22Os have an advantage of larger contact area but are intended for through-hole mounting. As replacements for surface mounted regulators you would want to careful not to move the unit too often to avoid any risk of tearing the tracks due to vibration.

True, but the largest heatsink weighs in the order of a few grams - you'd really have to be throwing the unit around to be lifting tracks.

[Paul_oz53]

True, but the largest heatsink weighs in the order of a few grams - you'd really have to be throwing the unit around to be lifting tracks.

Agree.

Probably the biggest risk would be caused by poor soldering technique cooking the pads thereby weakening them.

Personally, I like the TO-220 approach.

[brianh]

.....I'm a little late to the party, I know, but I've just carried out the conversion and I wanted to add my 2c worth....

Firstly, many thanks to Steve for this excellent conversion, I'm very grateful he's taken the time to put together this information.

When it comes to removal of the old regulators, I started by nipping through the legs - I quickly realised that the mechanical shock on the leg is quite significant.... The profile of the legs being quite short and also fairly solid, don't provide for much shock absorption - even with very fine nippers I could feel quite an amount of sharp mechanical shock being transmitted to the board when the leg parted, and was concerned that the mechanical shock could be enough to cause the land to separate from the PCB.

I think a better way is to work at cracking the regulator case just above where the legs enter - it's quite brittle and should separate without a lot of force, and when enough of the case is removed to reveal the leg, it can be de-soldered and removed quite easily.

Once the rest of the case is removed it also leaves most of the backplate revealed, which makes it possible to get a lot more heat more evenly into the whole backplate, rather than working with the tiny amount normally visible when the regulator body is intact.

In order to attempt this you still need the right type of soldering iron - I used an 80W temp controlled iron and it took a good number of seconds before the solder holding the backplate would flow enough to allow removal, but with care I was able to do that without damaging the board.

EDIT: - I've added a picture that might help to show the approach - you can see the whole of the backplate revealed on the right hand 3.3v regulator after cracking off the casing:

After cleaning up the board ready for the new parts:

[alfons]

Hi Brian,
Thanks for your tips on removal of regulators. I've been holding off removing my damaged ones, but looks like all of them have now given up the ghost so now I have no choice but to get in there and do the deed. I'll certainly follow your advice to remove the regulators.

Thanks again to you and all the other members for providing such helpful information.

Regards
Alfons

[peteru]

The standard way to get those regulators off the board would be to pre-heat the entire PCB and then desolder using hot air. The PCB is designed to act as a heat sink and uses a lot of copper planes stitched using the vias to conduct heat away.