Samwha BL capacitors, a strange cable, and a little about modifying drivers on an assembly line

Following up on my review of buying ceiling lights, I decided to write an article about modifying their drivers—or, more precisely, about some interesting capacitors from Samwha, different types of electrical cable, and the details of working when you have a lot of boards to modify.
I hope you find it interesting and useful.


So, if you remember, I bought a number of ceiling lights. But since I'm a bit lazy and don't want to have to repair them later, I decided to get ahead of things and modify them right away.
The plan was to replace at least:
1. The current-sense resistors, which I wrote about last time.
2. The input and output capacitors with better-quality ones.

For the modifications, I first wanted to buy the familiar Samwha RD capacitors, which I usually recommend for power supplies. But when I visited the website of the Kharkiv store Kosmodrom, I happened to come across the BL series, which was new to me. According to the specs, these are specifically designed for ballasts. They cost 8 UAH each, or about 0.3 dollars, which suited me just fine. For the output capacitor, I decided to use Samwha as well, but the WL series—I wrote about those in my DAC review.

I stopped by the store after work and bought 25 of each type: 24 for the drivers and one extra, just to make the numbers come out right :)


First, the output capacitors: Samwha WL, with low ESR, rated for 50 V, 100 µF, ESR 0.25 Ω. The originals were 47 µF, 50 V — link, priced at 2.5 UAH, or 9 cents.
There were also 100 µF, 63 V versions, with half the ESR but almost twice the price. I passed on those not because of the price, but because of their size—they wouldn't fit on the board.


Some reference information from the datasheet.

General information.


Key specifications; the ones I bought are highlighted in red, and the 63 V versions in green.


I hadn't tried the BL series before. They're said to be specifically designed for electronic ballasts, and to have an extended service life of up to 10,000 hours at 105 degrees and full load. They cost 8 UAH, or 0.3 dollars — link.


I decided to keep the same capacitance as before, 10 µF. The capacitors are rated for 450 V rather than 400 V, though, and their cases are considerably larger. And don't forget that increasing the capacitance also increases the inrush current when the unit is switched on.
Some information from the datasheet.



I tried to find the ESR data in the datasheet, but for some reason it isn't listed in the table, even though it's mentioned in the heading.


And a few measurements.
1, 2, 3. Samwha WL, 100 µF, 50 V; specified ESR at 100 kHz is 0.24 Ω. Unfortunately, my meter only measures at 7.8 kHz, but I measured in every available mode just in case.
4, 5, 6. Samwha BL, 10 µF, 450 V; measured ESR at 7.8 kHz is about 1.5 Ω.


Now let's compare them with some other capacitors:
Original 10 µF, 400 V.
Jamicon 10 µF, 400 V.
Samwha BL, 10 µF, 450 V (results above).
Samwha SD, 22 µF, 400 V. This was supposed to be a 10 µF, 400 V capacitor, which I use very often, but somehow I unexpectedly ran out of them.
Teapo 10 µF, 450 V.

Interestingly, the Samwha BL 10 µF, 450 V is the same size as the SD 22 µF, 400 V. These capacitors are quite large, so keep that in mind when using them.


1, 2, 3. Original capacitors.
4, 5, 6. Jamicon 10 µF, 400 V; quite old, they've been sitting around for a long time.
7, 8, 9. Teapo 10 µF, 450 V; same story, I still haven't found a use for them.
10, 11, 12. Samwha SD, 22 µF, 400 V.

In my opinion, the difference is quite significant: the ESR is about three times lower than with the originals. The originals also show a considerable dependence of their parameters on frequency. Teapo really surprised me: in terms of ESR, it turned out to be even worse than the “Chinese” capacitors, though I suspect age is to blame.


First, an experiment: replace the capacitors on one board and see how everything fits and whether it gets in the way of anything.


I had to bend the leads; here they almost touch the case. Later, I started bending them a little differently.


The board could be installed in either of two orientations, but now it only fits with the components facing down.


Power it up—it works, which is a good start :)
By the way, I forgot to mention that when I tested all the drivers, I powered them through a 150 W incandescent bulb, just to be safe. It didn't interfere with operation, and in the event of a short circuit, there would have been less risk of damage.


But we still have 23 more drivers to modify, plus a set of resistors and capacitors.


First, take all the boards out of their cases and put the cases somewhere out of the way.


Now I'll tell you a little about doing the modifications “assembly-line style.” I used to build various devices in small batches of 20–100 units, and the assembly process for a batch like that is quite different from building one or two devices. The key is to break the work down into separate operations, which speeds things up considerably compared with modifying or assembling each board individually.

First, desolder everything we're going to replace—in this case, the current-sense resistors and capacitors.


Then take either a sharpened matchstick or a wooden toothpick and clear the holes for the capacitors. Some people use a desoldering pump, either a regular one or one with a built-in heater, or special copper desoldering braid. I have a pump—two, in fact, both with heaters—but I don't use them. I don't use braid either; I prefer an ordinary matchstick. It never runs out, and there's always one at hand.


At this stage, the board should be free of “unwanted” components and ready for the next step. While clearing the holes, check for solder blobs or anything else that looks odd.


Next, I grab an ordinary piece of laminate and a very ordinary soldering iron, lay out the boards in the same orientation, and solder in the current-sense resistors. For some reason, people often mistakenly think you need a special soldering iron to work with small components. Sorry to disappoint, but I get by just fine with a basic “Solomon,” paired with a compact soldering station that keeps the temperature stable. I hardly ever use modern T12 tips, either.

In general, it's best to break the work into steps so you have to change tools as little as possible during each modification or assembly step. For example, here I lay out several boards and pour the resistors onto the table. I hold the soldering iron in my right hand and tweezers in my left, and don't put them down until I've finished with all the boards I've laid out.


Next comes installing the capacitors. You'll need tweezers to form the leads and flush cutters to trim them after installation. The small capacitors stay in place on their own; the large ones do too, more or less. But I handled them the same way as in the previous step: inserted the capacitors, trimmed the leads, placed the board with the leads facing up, and repeated this a few times. Then I picked up the soldering iron and solder and soldered everything I'd prepared. Very convenient.
Again, I don't use imported flush cutters; I don't like the ones available for sale. I ended up using the locally made “Kommunar” ones. Anyone from Kharkiv will understand. In short, they're excellent tools, but harder to find in other cities.


And that's the main stage done: the capacitors and current-sense resistors have been replaced on the boards.


The second stage is to replace the input cable and shorten the output cable a little.
I hadn't planned to do this at first; I just decided it would be better. Also, I could have made the first stage a little easier by desoldering the cables right away.

I bought a lot of different cable. At first, I just went to the nearest hardware store and bought some ШВВП 2х0.75 and 2х1.5, since I was almost out of cable at home. But I didn't like it, so I had to go to the market after all, where there's a much wider selection.


So, here's what I ended up with:
1, 2. 10 m of 2х0.75, which for some reason is marked 2х1.0, and 6 m of 2х1.5 (in two pieces). The first cable's marking is visible in the photo; the second is marked “Premium ШВВП НГ 2х1.5 GOST 7399-97 2018Г.” 2x1.5 seems okay, but 2x1.0 has very stiff conductors. I even thought it was copper-plated steel, but the conductors aren't magnetic.
3, 4. Cable from the Zaporizhzhia plant, three 15 m lengths: 2x0.75, 2x1.0 and 2x1.5. At first I told the seller I wanted the more familiar cable made in Odesa, but only when I got home did I notice he'd given me ZZZCM cable. How does that even happen?
The cable is decent, fairly flexible, and doesn't go stiff even in freezing weather.
5, 6. This one's more of a "collector's item". I found a piece of cable at home from a manufacturer I didn't recognize, K86, ШВВП 2x1.0 2013Г. No other information is available. The cable feels decent.


Now for the tests. I won't measure the cable's cross-sectional area directly; instead, I'll measure its resistance and calculate the actual cross-section. A cable with a cross-section of 1 mm² has a resistance of about 17.2 Ω/km, so we can work out what we've actually got.
To improve accuracy, I connected the conductors together at one end and measured at the other. That gave me twice the cable length.
1. 2x1.0 (sold as 2x0.75), 10 m length, 17.2/(0.82/20x1000)=0.42 mm²
2. 2x1.5, 3 m length, 17.2/(0.0955/6x1000)=1.08 mm²
3. 2x0.75, ZZZCM, 15 m length, 17.2/(0.738/30x1000)=0.7 mm²
4. 2x1.0, ZZZCM, 15 m length, 17.2/(0.565/30x1000)=0.91 mm²
5. 2x1.5, ZZZCM, 15 m length, 17.2/(0.378/30x1000)=1.36 mm²
6. 2x1.0, unknown cable, 0.82 m length, 17.2/(0.0388/1.65x1000)=0.73 mm²

Those are the actual figures, and in some cases they differ noticeably from the stated values :( Maybe I should get a couple of meters of cable from different manufacturers and run a test?


Next, we desolder all the wires. I didn't do this with every board; I took 10 at a time. There was no point taking more: the next step needs more room on the table, and the boards would just get in each other's way.


I shortened the wires to the lights a little. It would have been nice to replace them too, of course, but they come with connectors already attached, and replacing the connectors didn't appeal to me, either technically or financially.


I had a dilemma over which input cable to use, 2x0.75 or 2x1.0. It turned out their outer dimensions are exactly the same; the only difference is the conductor cross-section. Even the wire insulation has the same diameter. So the 2x1.0 cable has thinner insulation on its conductors.


Since I planned to have the housing clamp the cable with its plastic strain relief, I chose the cable with thicker conductor insulation. I didn't want to clamp the whole cable; the housing latches wouldn't hold properly.


I used a knife to score the outer insulation and then removed it, leaving one side longer than the other.
I measured the wire length needed for soldering to the circuit board, then cut the remaining wires to match.
I stripped the wires, again leaving more insulation removed from the longer ones and less from the shorter ones.
I tinned the short wires; the photo shows the difference in cross-section compared with the original wire.

By the way, I've often come across the misconception that cable cross-section is chosen solely based on load current. That's not quite right: in some cases, a larger cross-section is also chosen for greater mechanical strength.


Everything will be connected using Wago terminal blocks. I bought quite a few for the lighting, but I didn't want to clamp stranded wire directly, so I fitted ferrules to the cable. Since I was using insulated terminal blocks, I used uninsulated ferrules.


The second stage is complete: all the boards have been reworked and the wires soldered on. After that, wash the boards with alcohol or acetone. If using acetone, make sure it doesn't get into the transformer.


I could have stopped here, but I'd already got carried away and decided to coat the boards with protective varnish as well. Since using an aerosol isn't so convenient with this many boards, I took a 5 L canister of varnish Plastik 70 out of the cupboard.


I poured the varnish into a small container; about 50–60 mL was enough.
I coated the boards by fully immersing them in the varnish.


I quickly coated all 24 boards and went off to do other things, since the varnish is not only very volatile but also has a strong, rather unpleasant smell.


About 6 hours later, I came back and repeated the process, this time laying the boards face down to give the underside a better coating.
After a while, I repeated the process a third time. That used up the rest of the varnish in the can.


Unfortunately, because I laid the boards with the capacitors facing down, the appearance of some of them suffered a little. I don't think it's significant, though; the important thing is that the boards are better protected.


The underside looks great.


Most of the lights have already been moved elsewhere, but I kept a few at home for testing, and now they came in handy. The light looks yellow, but in reality it's noticeably whiter.


And that's the end of the rework. Now it's time to install everything on the ceiling. I hope it lasts a long time; at least I did my best to extend its life. I liked the capacitors. Time will tell, of course, but their stated and measured parameters were quite satisfactory, and Samwha hasn't let me down so far.

That's all. I hope this was useful, and as always, I'm happy to answer questions.