Jeremy Chadwick · @koitsu2009
2nd Oct 2018 from TwitLonger
Post-analysis of my dead Dell 2407WFP monitor
I've had my Dell 2407WFP (specifically the 2407WFPb, revision A03) since roughly 2006. Despite its well-documented quirks and high input latency due to the Genesis FLI5962H-LF video controller, and some general ghosting issues, its served me very well. I've used all of its inputs -- even composite, for when I needed a monitor for my Apple IIGS. It's also one of the last monitors to use an S-PVA panel, known for excellent colour, deep blacks, good viewing angles, and without ridiculous problems like IPS glow, while offering an impressive 1920x1200 resolution (16:10 ratio). In short, I've grown attached to it. If you aren't familiar with this monitor, TFTCentral has a great review of the 2407WFP-HC, which is similar enough to the 2407WFP that you can get a feel for what it brought to the table back in the mid-2000s: http://www.tftcentral.co.uk/reviews/dell_2407wfp-hc.htm
Several things began happening over the years which finally came to a head a couple days ago when the monitor stopped powering on. These are the things I noticed and thought peculiar but never had an answer for until now:
- The top of the monitor would get hotter than I remember. While the 2405FPW and 2407WFP series have always run pretty hot given use of W-CCFL backlighting, this felt abnormally hot.
- Some grease or oil on the back of the monitor, particularly around the AC jack.
About 2 weeks ago, the monitor began exhibiting its first death throes: occasionally when coming out of standby, it would show some randomly-coloured static on some scanlines, random input LEDs would be lit, and none of the buttons would work. Physically power-cycling the monitor fixed the problem. I worked around the issue by disabling monitor standby in Windows 7 and opted to just turn the monitor off manually.
That "solution" worked up until 2 days ago, where one night when powering off the monitor, it went black like usual... but then proceeded to turn itself back on, displaying a series of randomly-coloured vertical lines across the entire screen, along with random input LEDs lit. Like before, no buttons worked. But this time, physically power-cycling it did nothing to fix the problem. I left it unplugged for about an hour, after which it never powered on again. :(
Once I disassembled it and did some sleuthing, the cause of death, as well as the peculiarities, became quite obvious:
Video board
-------------
Examination showed 7 of the 29 electrolytic capacitors were bulging. This is never a good sign, and for whatever reason was particularly common in electronics during the early-to-mid-2000s. All the caps were "Elite" brand. Of the 7 bulging, 5 were 220uF 25V, and 2 were 220uF 35V -- the remaining 22 were all mostly 47uF 10V or 25V. "Elite" caps are known for low-cost mass manufacturing, and not intended for long-term use.
Because of clearance and limited PCB room, plus the fact I don't like soldering components which are extremely close together (especially on a board with SMD/SMT components on it too!), replacing all 29 caps is not feasible for me to do. However, replacing those 7 is certainly doable, and I plan to replace them with high-quality Nichicon-brand caps.
Also I should note that the flex PCB responsible for the buttons is connected with a ribbon cable to the video board. So, odds are, the weird behaviour I saw (visually) and lack of button functionality was likely caused by failing components on the video board... but might have been exacerbated by something worse:
Power board
-------------
Most 2405/2407 monitors die due to a leaky 220uF 450V capacitor on their power board. The leakage is usually quite obvious, coming out of one of the legs. Mine, however, looked perfect. In fact, aside from signs of extreme heat (a couple resistors near the top of the board had a certain look to them; not cracked, but certainly getting there), the board looked OK.
But when I got to the bottom of the power board near where the AC jack was, I found what looked to be oil or grease covering everything. In fact, some of it had even managed to sneak its way over near the DVI-D connector on the video board. It was when I examined the metal cover/casing that things became much more clear:
The power board houses several VRMs which are attached to heat sinks. The heat sinks are quite thin (thinner than they should be, IMO). To help dissipate further heat, Dell decided to place thermal pads between the tops of the heat sinks and the metal cover/casing. I imagine this varies monitor to monitor, but on mine, the thermal pads are glued/affixed to the metal cover/casing. The power board also has two transformers on it; for whatever reason, the assembly plant chose to affix a thermal pad to one of the transformers, rather than the metal cover/casing. I can't explain why they chose not to put one on the other transformer. Furthermore, there's one part of the board where 3 VRMs are attached to a single thin heat sink.
The thickness of the pads is about 4-5mm -- thick enough that they act as a "buffer" or "conduit" between the metal cover/casing and the heat sinks (or transformer), held together with pressure. But as anyone familiar with thermal transfer knows: no surface is perfectly flat. So to minimise this and provide better thermal transfer, an agent must be used: the most common being thermal grease (thinner) or thermal paste (thicker). Dell or the assembly plant opted for thermal grease, not paste, and a particularly a cheap low-quality type of grease. Most thermal grease doesn't hold up for more than 5 years, and usually becomes a sticky goopy mess when it gets old. Hmm... grease... oil... hmm...
So what happened exactly?
As said above, the 2405/2407 series run pretty warm due power requirements and use of W-CCFL backlighting and general power requirements. As the VRMs got hot, so did the heat sinks, thus the thermal grease. All of this worked great when the grease and components were new.
But as things got older, the grease became less viscous (i.e. thinner, more liquid-like) and began to "leak out" from between the pads and the heat sinks/transformer. And because the boards are obviously positioned vertically, the grease began to drip downward onto the metal cover/casing, AC jack, and everything else in that area.
This in turn created a cascading failure: now that there was less thermal grease between the pads and the heat sinks/transformer, there was less thermal transfer, which meant the heat sinks and a transformer, and surrounding components, got hotter. That meant more grease leaking out; rinse lather repeat. And because heat rises, the top of the monitor became hotter than where it historically had (back of the monitor, e.g. metal cover/casing).
All of this is the result of short-term design choices. Better choices would have been as follows:
- Using a different approach to power PCB thermal transfer, if at all possible. Using external PSUs might seem like an option (many vendors do this), but a lot of people like myself prefer integrated PSUs. If you have to use an external power adapter, then make your monitor stands include a place/pocket to hold the adapter!
- Use substantially larger heat sinks.
- Use high-quality non-conductive thermal paste (ex. Arctic Silver Ceramique), not grease. Non-conductive is important for safety: if the paste ever became less viscous and started to drip, you wouldn't want conductive paste causing electrical short in your PSU, leading potentially to a fire.
- Use high-quality electrolytic capacitors (e.g. Rubycon, Nichicon, Suncon, Chemi-Con, Panasonic).
- Better assembly Quality Control (re: missing thermal pad on one transformer).
So while I absolutely love my 2407WFP, had better design choices been made, these monitors would last longer. I ask readers to consider the MSRP when the 2407WFP launched in 2006: around US$800. These were not cheap monitors -- we were paying for quality as much as features. And while hardware can't last forever (20 years would be considered incredible), the above design appears to be neglectful considering the price.