|
|
| Williams "Black Knight" pinball restoration. | ||||||||||||||||
| LED upgrade testing and details | ||||||||||||||||
| LED's. A lot has been said, in a lot of
places, about how they can really improve a lot of areas in a pinball machine.
As I am dealing with a Williams System 7, notorious for frying an overloaded GI
lighting circuit, they were among my top priorities for what to upgrade in this
machine. But, here's the thing, how much improvement do you REALLY see from these things? I see a lot of hypothetical numbers thrown around, but never an actual real-world test done anywhere. I'm quite interested to see what kind of improvement I can see in two main areas. One is power usage, the other is temperature. Temperature there's no way to calculate what I should be theoretically seeing in an improvement, but the light usage I can at least get a ballpark figure of what I'm looking for. The first stage of this will be to do only the GI lighting replacement, then follow up with the insert/controlled lights. The GI lighting is 65 bulbs, and insert lighting is 59 bulbs. This totals out to 124 total bulbs in the machine. |
||||||||||||||||
 |
Part
one of the test is going to be an average power used during a three
hour period. I will be using a Kill-A-Watt EZ meter, which measures
total KwH usage. The machine will be ON and in attract mode during this
timeframe, in order to see the most use from the controlled lamps and
eliminate any solenoid use that may skew the test. I cannot, however, test peak useage with this meter as the sample rate of once per second is just too slow with no data logging. |
|||||||||||||||
| Temperature will be measured in 4
places. I will sample the ambient room temperature, backbox
internal temperature, backglass temperature and playfield glass
temperature. Most of the tests will be performed with a basic IR thermometer that I've been using for a while. However, as I don't want to open the backbox in order to sample the internal temp (it will let the hot air out, and skew the results lower), I'll be using a meat thermometer inserted into the un-filled keyhole on the side of the backbox. I know, not the most scientific methods out there, but it's the best I can do on a budget. I will also be doing multiple samples of the temp over three hours. Tests will be done at the start, 15min, 30min, 45min, 1hr, 1hr30min, 2hr, 2hr30min and 3hr. I'm doing this because most home users are going to have their machines on for an hour or less, only when playing, and usually don't leave it on for too long. By three hours, temps should have settled, and will be at what an operator is likely seeing. |
 |
|||||||||||||||
 |
For the GI LED's I've chosen to use those available from
Cointaker.
The backbox will be getting 170 degree Warm White bulbs, while the
playfield will be getting the frosted dome Warm White LED's. Inserts
and the coin door GI will be going with color matched LED's.
Again, these are from cointaker, and I will use their 90 degree
lens style throughout.
All of these are a non-polarized design, featuring an internal rectifier, but no capacitor. |
|||||||||||||||
 As you can see, I only went for an
hour and a half. At that point, the GI power input connector
failed on me. It just wasn't safe to continue the test. At
a different time I had managed a full 3 hours to measure power
consumption, which meaured at .52 KwH. This works out to an
average power use of 180 watts.
Note how fast the internal temperature of the backbox came up. All this heat is coming off the power supply, the overloaded driver board resistors, plasma displays, and all the #44 lamps. In just the first 45 minutes, it's already almost at 120 degrees. For your average home user 45 minutes is not an unreasonable time for a machine to be on. Particularly if you've got guests. These numbers were not surprising at all. When you start hitting that kind of temperature, you start bumping against what a lot of electornic components are rated for. Performance starts to alter in strange ways, such as changes in resistance values, switching speeds, etc. Even bigger of a long term concern is that excessive heat contributes to component failure more than any other factor when it comes to electronics. This is why cooling fans and heatsinks are so prevalent these days in so many electronics. Yet in this system there are zero cooling fans and very limited passive venting. How do you deal with the heat? There are two methods to do so. Reduce the amount of heat being produced, or dissapate the heat that there is. Reducing the heat can be accomplished by replacing parts with alternates that create less heat. Dissipating the heat would best be accomplished with some kind of cooling fan system, and there are many suitable locations to add them in. But then you have to deal with a lot more dust accumulation, which is a bad thing in itself. |
||||||||||||||||
Here's a good photo of the connector that had failed partway through the test. This connector was brand new, with only 12 hours of run time on it. It's drastically under-rated for the amount of power that it has to deal with, and I would not be keeping it if I were keeping incandescent bulbs in the machine. I shot a temperature reading of this connector before shutting down, and it read at 331F. Ouch. It was replaced immediately following this test. |
 |
|||||||||||||||
 After the full GI
switch all the temperatures are down across the board. Just as expected.
What about total power useage? The power use over the three hour period dropped from to 0.38KwH, an average of 126 watts. This is a 27% drop in total power use for the machine in total, or 54 watts of reduction. That's not chump change, it's a serious power drop. Lastly, that GI input connector was only at 110 degrees when running the LED's. Very nice. |
||||||||||||||||
 |
||||||||||||||||
| This is
the chart showing the temperatures after swapping all of the insert
lamps to LED's, meaning that the machine is now 100% LED for the
lighting circuits. Now, things were a bit interesting here. This test was run almost a year after the previous tests, mainly because I didn't replace the LED's until fall, and I wanted equal ambient room temperatures in order to keep some consistency. But somehow the backbox temperature went up. I suspect this may be an issue with the cheap thermometer that I was using, as the temps that came from the IR thermometer showed a drop in the backglass temperature again. I had also swapped out the lamp matrix transistors with MOSFET's by this time too, which means I should have seen a lower backbox temperature. The power reduction again was reduced, as expected. The three hour power use was .22KwH, which averages out to just 73 watts total power use. That's only 40% of the power it was using when it was running incandescent bulbs. Less than a single light bulb in your average table lamp, and a fraction of the personal computer that you're reading this on. |
||||||||||||||||
So, the total numbers so far are a
59.5% reduction in power. I'm not done yet either, as there
are still two significant heat sources in the machine, which means
waste energy that can be improved upon. One is the antique power
supply design, using the waste monster LM323K voltage regulator for the
5v logic line. The other is those plasma score displays. Next
stage will be to swap those over to mode modern LED displays and a high
efficiency power supply. I really want to see how low I can get
the power use on this machine.
|
||||||||||||||||
To Be Continued!