Visual Signs of Capacitor Failure - Bulging

I'd appreciate a design review!

Scenario is, I need a PSU (500 mA, about 13.7V output after the regulator). This is a one-off project but I won't have access to it so it needs to be reliable. The ambient temperature may reach 40 deg C. The PSU will be switched on and off maybe a couple of times a day. The load may peak to 400 mA, and on average will be 300 mA.

I'm using a 14V 15VA transformer, operating from 50 Hz mains. (It's a linear PSU, for low noise reasons).

The capacitors are rated at 100 kHz, so some extrapolation is needed.
I believe the three electrolytics are selected to operate comfortably well.

I'm most concerned about whether the 47uF polymer tant is suitable or not. I've never used such a part. The polymer tant is supposed to fail in a benign way compared to normal tantalums, which is good to know, but I still want a reliable design. The TCN.. part is supposedly suitable for 'bulk decoupling of SoC').

Would you/have you used a polymer tant in a PSU? Recommended or not?
I've not constructed this, but the PCB is ordered. If I replace the cap, it will ideally need to be a similar footprint (7.3 x 4.3 mm component size). I suppose I could piggy-back a couple of capacitors as long as it also still looks tidy too (this PCB will be visible through vents).

If it's really not recommended, I could go with this 22 uF ceramic cap, it will fit. Thoughts?

I'd appreciate a design review!

Scenario is, I need a PSU (500 mA, about 13.7V output after the regulator). This is a one-off project but I won't have access to it so it needs to be reliable. The ambient temperature may reach 40 deg C. The PSU will be switched on and off maybe a couple of times a day. The load may peak to 400 mA, and on average will be 300 mA.

I'm using a 14V 15VA transformer, operating from 50 Hz mains. (It's a linear PSU, for low noise reasons).

The capacitors are rated at 100 kHz, so some extrapolation is needed.
I believe the three electrolytics are selected to operate comfortably well.

I'm most concerned about whether the 47uF polymer tant is suitable or not. I've never used such a part. The polymer tant is supposed to fail in a benign way compared to normal tantalums, which is good to know, but I still want a reliable design. The TCN.. part is supposedly suitable for 'bulk decoupling of SoC').

Would you/have you used a polymer tant in a PSU? Recommended or not?
I've not constructed this, but the PCB is ordered. If I replace the cap, it will ideally need to be a similar footprint (7.3 x 4.3 mm component size). I suppose I could piggy-back a couple of capacitors as long as it also still looks tidy too (this PCB will be visible through vents).

If it's really not recommended, I could go with this 22 uF ceramic cap, it will fit. Thoughts?

If you want low noise then put some little ceramic caps around the rectifier. Doing so is mentioned in AOE Edition 3 (maybe in the X pages book) and after I read that I measured a design I was working on. I was seeing horrible 25V amplitude HF ringing around the rectifier but the little caps (470nF) from input of rectifier to ground can fix it. Worth experimenting because caps with series resistors can be better.

Second thing, on light or zero load the max output from your transformer will be 14V * 1.1 (transformer regulation) * 1.1(mains high) * 1.1 (engineering tolerance !) = 26.4V pk. Your main caps are OK but ideally you should use a 50V cap if it has any tantalum in it (that rule (rated cap voltage = 2 * max voltage applied) comes from solid tantalum but I apply it to power caps with any tantalum).

I'm using tantalum poly's in power supplies so I hope its OK

MK

Hi Michael,

Thanks for the info! I took a look at AoE3, page 634 mentions a damping network, I have footprints for that on the PCB so I'll be sure to add those and will check out the X-chapters in case they mention further.The full schematic is shown below (I have no access to an earth connection due to this being a retrofit inside a double-insulated product). Thanks for the cap feedback, I've searched online, and found a 50V tant poly cap which is the same size, so I'll use that instead. The PCB arrives this week, so hopefully I'll have some feedback how it goes soon!

Also, just noticed that section of AoE reminds of the need for a 1M resistor across the line filter input.. I'll order a suitable resistor while I'm at it, and solder that in too.

I'll be interested to hear how well it works and how noisy.

I've just done a design with one of AD/Linear Technology  "SilentSwitcher" chips (not built any yet) so would be interesting to compare.

MK

Also, unrelated, there's some quite nice transformers on AliExpress (I cannot find them at normal distributors), the primary-to-secondary capacitance was measured to be just under 100 pF, i.e. about a third compared to a similarly-sized toroid because there's concentric bobbins with the inner bobbins containing one winding, and the outer containing the other, so there's quite a gap, and also has a shield between the concentric bobbins, that needs to be earthed. I've not tried this transformer yet. (I'm not using this transformer for the circuit mentioned, that will be a normal toroid). 

Unfortunately I've not had a chance to do any measurements : ( I was planning on using the MCC 172 to see power supply noise up to 25 kHz or so that it supports, but I've needed to finish this project quickly, and didn't have time to set this test up in an enclosure before fitting it in the end equipment (an audio amp/radio receiver).The PSU is simply used to power an off-the-shelf Bluetooth audio module (to try to make this product suitable for modern times! It is 30 years old).

The LM1086 layout was pretty much as described in the datasheet, and I used a 50V tant poly on the input (it is that black cap labelled C11).

The transformer and associated circuitry is below. The cap+resistor damping is next to the bridge rectifier.

What I'm not sure about, is the covering over the high voltage bits. It's not required for insulation purposes, but I still wanted a finger guard in case anyone opens the enclosure.

I've used UL94V-2 mylar (and impact glue), which I think is OK because it's not touching anything that could arc, and is thin enough that there's only a finite amount of burning it could possibly ever do, even if it were possible to ignite it. But I'm unsure. I'd like to obtain better material for such purposes, but would like to leave the current material in if it's acceptable.

Audio quality is finally good/acceptable! I originally prototyped with a Meanwell AC-DC module, but there was mains hum : ( (presumably because of the high transformer capacitance between windings).

I found some neat items for helping retro-fit this board. One is a chunky  M4 right-angle mount   originally intended for screw-fitted antennas, but I've mounted it on the underside, with four large heavily soldered through-hole pads, and epoxy-glue to secure it further.

The other item was a horizontal card guide.

I needed to drill zero holes to retrofit it all. All holes were created using this hole punch kit. It's excellent, has quite a lot of clearance, and easy to be precise because each punch has a protruding centre-point. Plus, the punches/dies are available as low-cost spares if they wear out. 

Also swapped out the (awful, and slightly corroded) UHF connector (Belling-Lee) for the FM antenna, and replaced with 75 ohm BNC. I'm replacing with a custom antenna which is simply a standard T-shaped antenna, but with a 75:300 ohm transformer. It might be no good for AM! But I'm hoping AM is of little interest to the user.

Antenna assembly, it is basically 75-ohm coax on the left (RG179) going to a BNC plug, and the 300-ohm antenna balanced line (feeder) on the right. The transformer is MABA-011039 and it is soldered to a scrap of board where the copper was cut into four quadrants.

I don't count myself as a power supply expert, but would also put a low ESR cap on Vout.

This particular voltage regulator expects ESR within a range, they suggest tant or alu electrolytic on the output. I'm very comfortable with using tantalum for that, even though more modern ICs can use ceramic. 

Overall I'm really happy with this IC, it has an OKish price, large easy-to-solder package, and way better than typical LM7805 etc! I will probably build another circuit with it at some point, and this time take measurements.