LaCie 20" LCD starts badly

[Mark Hennessy]

(Just spotted that I missed post #7, FWTW)

Yes, it's easiest to assume that they all need to be low-ESR types - especially as they usually do. The ones in the link look to be good.

For info, I just measured four new 150µF types (all 385V, but different brands) that come out at around 0.5Ω.

However, it's clearly been heating up. As the ESR is nice and low, then it's probably leaking.

So it the capacitance is OK-ish*, and the ESR also is OK, then I wonder what's causing the fault? Probably one of the smaller ones on the primary side, as the resistors seem OK

(* leaking caps will cause most capacitance meters to over-read. That's because they essentially put a known current through the capacitor and time how long it takes to charge. Leakage "robs" said current. Though I doubt yours would be leaking much, as it would have blown by now in that position)

I have the Peak Atlas ESR meter, and recommend it. The Chinese kits are pretty good for the money though. ESR is conventionally measured at 100kHz rather than 10kHz, but of course, the DMM needs to work out there. Naturally, the 'scope is good if already warmed up, but it is nice to have the stand-alone thing.

[ppppenguin]

That monitor is certainly worth saving. If it goes on for another 13 years I'll be very happy. At a rough guess it's probably been on for around 60,000 hours so far. Even allowing for the usual longevity of electrolytics when underrun and kept reasonably cool that's not bad.

I know about the overreading problem with DC based capacitance meters. Like you, I don't know how that big cap is measuring so well when it has clearly had it.

The 1000uF will get replaced by a 2200uF. I can't imagine this will cause any problems. I can see that that they saved a few coppers, even on a high priced product, by using a mixture of 10V and 16V for the 2200uF parts. They must have been buying in very large quantities to make that worthwhile. Looking at the other PCBs in the monitor I suspect the PSU was bought in. Possibly a standard product, slightly customised for LaCie.

When I've been ordering from RS in the last year or two I've kept on forgetting to include a little 20mm diameter speaker to replace the duff one in one of my cordless phone handsets. The handset is fine but it doesn't ring and can't be used on loudspeaking. Finally remembered! A mere £1.34.

[ppppenguin]

Some musings from a design engineer .

I might make up a simple ESR meter. There are plenty of published designs out there but as a design engineer I'll ask if they are any good. All it needs is a 100kHz oscillator and something to measure the voltage across the cap. A nice touch would be to put a current source on the osc output so it tries to pass a fixed current through the cap. Then the meter scale will be linear. That might be as simple as using a 15V rail with a CMOS 4000 osc and a series resistor to give a very rough current source. Could get 30V by using 2 CMOS outputs in opposite phase. Might need to parallel up a few sections of CMOS logic to give enough current capability. Got loads of 4000 series in stock. Power supply would be a pair of PP3 which isn't ideal but they'd live for near shelf life.

Refinements could include some inverse parallel diodes across the test terminals so that the terminal voltage never goes very high. Would want to test at less than 500mV or so to allow in circuit tests. Protection against a charged cap? Probably not as I'd be the only person using it and I can remember to discharge the cap before test. Except when i don't and have to replace the burnt bits

Might not bother including an actual meter, just plug in a DVM. My Fluke 87 is spec'ed to 20kHz but I can easily check it up to 100kHz. Or include a simple opamp based precision rectfifier so the meter can read DC.

Lash together on a bit of pinboard or veroboard and bung in a small plastic or diecast box. I take the opposite approach to Yorkie

Has anyone here built an ESR meter and so can give first hand opinions on how a particular design works? Not sure I can justify £80 on the Peak unit when I don't think it would get that much use.

[Mark Hennessy]

If least-cost is the goal, a Chinese kit is less than a tenner, and measures LCR, ESR and most 2 and 3 legged semiconductors... But that's cheating

Not sure which version you have, but I've just checked my 87V - at 100kHz and 0.5V, it's about 1.4dB down relative to 1kHz. That varies with applied voltage - it was slightly better at 5V, and at 20mV it's unusable.

Given that accuracy is not a big deal here, it'll probably be OK (but see below). Basically, it's about indication - when the ESR starts to rise, it does so fairly rapidly (increased ESR leads to increased heating which speeds up the evaporation which increases the ESR). The designs with a hand-calibrated moving-coil meter with a slightly non-linear scale are actually fine for repair work.

Before I got an ESR meter, I regarded PSU repair as "refurbishment". So if there were any signs of trouble, I'd just change all capacitors and any high value startup resistors. This gave me a 100% success rate in the shortest time possible (which mattered in a live broadcast environment). Did I change capacitors that were perfectly OK? Undoubtedly. But they would have started to play up at some point in the future, so it was "preventive maintenance". The downside of this approach was the need to keep a large stock of capacitors, but that's no problem given the other projects we did - same for me at home, too.

I sometimes measured them with a function generator and a 'scope. My old Wavetek can deliver 10V into 50Ω. Feeding that into 0.1Ω (via a 51Ω resistor) gives ~20mV. At the time, I decided that any less current might require attention to detail around noise, but obviously 200mA is fairly hefty for a battery powered device, so I talked myself out of investigating any further. Looking back, I probably gave up too soon...

Based on how my 87V behaves, you'd need some amplification after the cap to get up into the 0.6V range.

[ppppenguin]

I didn't realise there were such cheap kits available. I'm not averse to cheating either. I'm not even registered on ebay where I expect most of these things are sold. Do you have a link?

I think you're right about bulk replacement. It might lack precision but it gives the right answer quickly.

Believe it or not I've never actually serviced a switchmode PSU before. Not really designed one either unelss you count the National "Simple Switchers". Actually that's not strictly true. I did once build a flyback converter to supply voltages up to 900V or so for a vidicon camera that I was modifying. Stabilising was a long string of zeners which also gave me the intermediate voltages needed.

[Mark Hennessy]

I got this one: http://www.banggood.com/DIY-M12864-Graph...86954.html

Though I made the mistake of buying from eBay for a couple of quid extra because it was a UK seller who promised quick delivery, but actually, it was "drop-shipped" from China - perhaps even from Banggood...

It's in kit form, but it doesn't take long to assemble. Perversely, the cheaper ones are generally pre-built. Banggood also do cases for them.

It's a dangerous website. Lots of temptation

I never especially enjoyed fixing the old ones that had a habit of repeatedly releasing the magic smoke. But designers have got very good at this game, and now they tend to wear out rather than blow up

[ppppenguin]

That's astonishing value. They do a version complete with case for less than buying the case separately. http://www.banggood.com/DIY-M12864-Graph...97023.html

I see what you mean about that site being dangerous.

I suppose there's usual problem when ordering more expensive items that they may well get stopped for VAT. Don't mind paying the VAT but the £13.50 RM handling fee makes anything under £200 or so look expensive.

[ppppenguin]

Caps arrived today. Replaced all the electrolytics in the PSU and it now works fine.

The 1000uF was replaced with 220uF and a 33uF with 47uF.

I think I'll get that ESR etc meter from Banggood. At that price it's not worth mucking around with homemade.

[Mark Hennessy]

The power supply in my tumble drier blew up yesterday. I blame you entirely

That uses a LNK304PN. Looks like an 8 pin DIL IC at a glance, but has 1 pin missing to provide some clearance around the drain of the internal MOSFET. These are known for going "bang" in a big way, and so it is with mine. Luckily, replacements are only a pound.

The inrush-limiting resistor is a fusible wire-wound type, and that had done its job reasonably spectacularly. Farnell sell replacements - 1738521 is close enough. Saves money, I suppose, but I prefer a fuse followed by a conventional surge-proof resistor - that's one of the main reasons why they still make carbon-comp resistors. Of course, anything bigger would have an NTC or similar for inrush protection.

And while it's open, I might as well change the caps. Only 6 or so. The two "big" ones (4u7, 400V) would have been a bit stressed by the sudden short. Same for the rectifier diodes - just 1N4007. Let's hope the Atmel microcontroller survived...

This one is fed from a half-wave rectifier rather than the usual bridge. Not sure if it was a mains surge, excess humidity, or just "natural causes". There is a MOV across the input for surge protection.

Busy week, so not sure when I'll get the chance to fit the replacement components, but fingers crossed. Replacement controller modules are over £100. Cheaper than a new machine, but this one is a condenser, so replacement won't be cheap. Like the idea of the heat-pump types, but they're still rather expensive.

[ppppenguin]

Always glad to be of service

I don't know if ultrasonic tumble driers ever made it to market. http://www.stylus.com/ckgrdm

Thanks for your advice about my monitor. It really is a lovely display, far too good write off.