Showing posts with label shorts. Show all posts
Showing posts with label shorts. Show all posts

Monday, 28 September 2009

Boarding school, term 3. Muddling through the end of year tests....

Hoorah! Hooray! Yippee! Wonderful! Marvellous! Brilliant! Fantastic! (Can you tell I'm happy?)

After a whole week of head-scratching, web browsing, and advice from my brother, my partner and Nophead, I have finally FIXED my bad stepper motor driver board. Unfortunately, this was the board with near-perfect soldering, of which I was very proud. Now it looks a mess! Anyway, here is the proof it works:

OK, it's not pretty.

The voltage regulator had been tested whilst still fixed to the board (but only by its ground pad). The regulator was definitely faulty. Unfortunately, once I'd removed that, everything else in the 5V path gave readings of zero ohm when tested on a 20 k ohm setting on the multimeter with the +ve probe on the path and the -ve probe on the ground pin of the Molex-style connector. This meant there was a short to ground somewhere, still.

The first thing I checked was the power-indicating LED, as this was what I had replaced twice, once because the first one was dead (although it turned out that that was incidental to the board not working), and once because I checked it for correct orientation when the board still didn't work. The result was that this part of the circuit certainly didn't look tidy, and was a cause for suspicion. I cleaned it very carefully and checked the gap between it and its protective resistor, but it looked separated. I wondered if I had damaged the pad on the board, but without disconnecting it from the circuit there was no way to check.

Next candidate was the damaged track alongside the regulator's space. I decided to cut the track close by the 5V pin' pad,so I scraped off the protective Araldite I had applied, and scraped back to the copper track. Then I very carefully cut the track, and made sure there was no continuity across the cut. Now, using the diode setting on the multimeter, I could light the power LED, so that side of the circuit was sound.
I cut the track in a similar way towards the rear of the regulator's space. Checking on resistance setting 20 k ohm again, the part of the track alongside the regulator gave reading of 1 against ground, meaning no connection to ground, ie no short there.
O.K., they were my top candidates! And they were fine. And I now had 2 unnecessary cuts in my track!

Oh well, nothing ventured, nothing gained.....
Following the track, it goes from the regulator to the electrolytic capacitor from where it splits, with part then heading to the 100 nF ceramic capacitor and to the second RJ45 socket, and the rest going down a "via" under the electrolytic capacitor and off to various other parts of the board.
In order to test this, I cut the track in a clear section underneath the board. Testing the chip end of the circuit to ground, the meter gave 1.
Ah-ha, now the problem is narrowed down to 2 capacitors, the "via" or the RJ45 socket.

The RJ45 socket seemed unlikely to have a problem, and ditto the large capacitor. Having plenty of spare 100 nF ceramic capacitors, thanks to supplies from Nophead after I cracked (a different) one on this board right at the start, I decided to simply swap it.
Some little desoldering and resoldering later, and voila! we have readings of 1 from everywhere on the 5V path to ground. So the problem WAS the capacitor.

Now my problem was the breaks I'd deliberately put in the track....
First I used the conductive pen to fill in the gap in the track on the underside of the board. It takes 10 minutes to get conductivity through the "ink", and testing afterwards showed I had good conductivity between the copper ends.
Unfortunately, when I tried this on the top of the board, I found that I also had re-shorting to ground! Obviously the cuts on the upper side of the board reached the ground connection in the board. So, using a trusty cocktail stick (which should also be added to the parts requirements!) I could paste Araldite (ditto) into the cuts in the copper track, but while that dried, I used the conductive pen to rejoin the breaks going around the cut section. If the conductive pen only works temporarily, I shall solder wires between the track breaks.

Then I soldered a new voltage regulator in place, with the help of Blu-Tack, again. Unfortunately, I knocked the electrolytic capacitor loose, causing more burning plastic due to access issues with my soldering iron as I fixed it back on. I tested it, and the voltage regulator, and then resoldered the first (removed) RJ45 socket. I tested everything exhaustively.

And now we have Blackpool at Christmas! (For those non-UKers, that means all the lights came on.)

My partner thought the board was a gonner. I thought if it didn't work anyway, there was nothing to lose by trying to resuscitate it, and it was going to take another 3 weeks to order a fresh board from the States anyway. And it works.

(PHEW!)

Friday, 25 September 2009

Second term at boarding school!

Just when you think you've got everything settled..........

.....My brother pointed out that just because I wasn't measuring 5V out of the voltage regulator whilst it was connected in circuit on the stepper motor driver board didn't mean it wasn't working; with a short-circuit anywhere in the 5V section of the circuit, the current would flow straight to ground, meaning no voltage to measure. However, there were no obvious shorts, no solder bridging two components and I had previously checked that all the resistors apart from the 0.25 ohm ones were sound. After Nophead's advive on checking low value resistors, I noted that my multimeter read 0.5 ohm when the probes were touched together on the lowest resistance range setting, 200 ohm. The resistors gave a reading of 0.8, meaning they are rated at ~0.3 ohm, which is about right allowing for slight inaccuracy of the meter at such low levels compared with range setting.
In circuit, there was no reliable way of measuring the capacitance of any of the small capacitors, so I was going on guesswork that the problem still lay, in fact, with the regulator.

I desoldered one pin on the regulator, and then the other.
I am getting on fine now using the solder braid with flux drawn onto it from a pen. This helps the solder to reflow. I placed the braid onto the top of the pin, acrosswise if possible, and applied the soldering iron tip on top of the braid for a few seconds. I found it easiest to position the braid and hold it in exactly the right orientation, with no twisting etc, with my hand, but once held in place by the soldering iron, I swapped to holding the braid with tweezers so that I didn't burn my hand when it got hot. Access was awkward, and I decided to desolder one of the RJ45 connectors to reduce the amount of melting and burning plastic where the large screws on the soldering iron shaft would catch! I still caught the edge of the Molex connector from time to time. I cut off the used end of desoldering braid holding the absorbed solder and repeated with clean and freshly-fluxed braid until the regulator's pins looked to be clear of solder. The pins still seemed tacked to the board, possibly with sticky burnt flux, so I cleaned around them and prised them up off the board. The pads below are fine.
Desoldering the through-hole pins of the connector was easier, still, after cutting the end of the braid into a point and using the fine iron tip, pressing the braid slightly down and into the hole as the solder wicked up.

Next I tried to desolder the regulator's large ground pad - that was just too tough! I managed to remove most of the solder, but there was some visible under the centre of the pad, but the gap between the part and the board was too thin (top to bottom) to be able to slide the desoldering braid in, and although I could melt some of it, because of the proximity of other components I couldn't use the big flat soldering iron tip, which would have been ideal to melt the lot all at once and allow the regulator to be released.
Eventually my partner went for the brute force option! and created more damage than was intended! Some of the board's contact pad lifted clean away, but the remainder still has contact to ground, and there is damage to the green coating over the track that leads from the 5V pin adjacent to the side of the regulator. There still seems to be continuity everywhere (there should be), though, so I have Araldited a thin protective layer over the shiny copper that is visible.
We tested the removed regulator - it is most definitely dead! Luckily, the minimum order for these had been 5, when I (thought I) needed 3, so there are a couple of spares after all (which is a relief because this was the part that came from RS, who were absolutely useless, and from whom I have no wish to order again)!

My brother rang just as all this was done, and said that I could test this board before putting another regulator on (just in case there is still a short on the board as well) by connecting the working board to power, and connecting a 5V contact on that board to a 5V contact on this board, and a ground contact on that board to a ground contact on this one. The regulator on the first board should easily be able to handle the current for both boards together. At least the power LED should come on for both boards. And if I do Nophead's thing of getting the other LEDs to work, too, I should get proper indication on both boards as well if this board is working. I need to look that up again, but meanwhile I'm letting the Araldite cure.