[?]

• The Crest CRE-55465 Switching Power Supply
• Performance 23 Dec 01
• The Guts
• A Cold Start Problem 03 Jan 02

Crest (AristoCraft) has released a new DC power supply to power its Train Engineer system or any other control system that can use 12 or 22 volts of DC raw power. This supply can provide all the power that a medium to large layout could ever need.

Unlike the older ART-5460 Ultima 10 Amp Power Supply, this is a truly regulated power supply. The older Ultima used a big iron transformer with a simple rectifier and filter to provide an adequate source of DC power. It was larger and quite heavy and sometimes made a fair amount of noise. Its output voltage would sag some at full load, but overall it did pretty well.

This new supply is a significant improvement. It uses modern switching power supply design to provide a constant, regulated output at 22 volts DC up to a rated 13 amps. It operates virtually silently. The loudest noise is the soft whirr of a cooling fan.

The front panel contains a power switch, an output voltage selector switch, a 20 amp meter (accurate this time) and banana style output terminals. There is an included DC heavy gauge power cable with lugs that fit under the terminals. Note that you can use banana plugs in the output terminals, but this is NOT recommended. A typical banana plug is good for only 3 amps and it WILL burn at 10 amps. Don't even try it.

The rear panel contains the fan, the AC input connector and a 115/230 volt selector switch. While the supply came packaged with a US style 115 volt power cord, it uses the worldwide standard power jack so a country specific cord could be substituted. On my unit, the voltage selector switch was set to 230 volts as delivered (probably as a result of testing). US users should be sure to look at it and set it to 115 volts BEFORE plugging the unit in.

Note that there is an internal fuse, a standard 3AG type. Also note that the grounding screw has a nut on the other side that grounds a wire from the circuit board. If you loosen the screw, it'll remove the preload from the screw and then nothing will happen. You'll have to open the unit up to tighten it again. There should be no need for this grounding point anyway as the unit is grounded through its 3-wire cord.

[ Top ]

I did some performance tests of the supply and I find that it really is a constant voltage supply with a very clean output. At no load, it produced 22.1 volts with 4 mV RMS ripple. At a load of 10 amps, it had sagged to 22.05 volts with 13 mV of ripple (mostly a 120 Hz feedthrough component as a result of full wave rectification of the AC line). The supply appears to switch at about 75 kHz but it didn't make enough output ripple to reliably trigger my scope.

With a heavier load, it put out 17 amps at 21.9 volts even though it is rated at only 13 amps. This is 372 VA, more that you'll find in most model railroad power supplies. I didn't test it for long because my test loads were starting to smoke. I also didn't test it at the 13 volt output setting.

I rearranged my load resistors to be able to handle the steady state dissipation at 15 amps and I found that the power supply just sat there and ran at 330 VA output without complaint. The exhaust air leaving the slots on the top of the unit got perceptibly warm.

At a load slightly higher than 17 amps, the power supply goes into short circuit protection mode. If the load is steady, the power supply will shut down for about a half second and then come back on for another half a second. It'll repeat this until the overload is removed. This is a good response to a short, but it could wait longer for the retry to reduce the average current delivered to a short. This way, it won't deliver enough power to melt or burn whatever is shorting it out, but it will also recover in a reasonable period of time. A delay of 5 to 10 seconds before a retry might be better. This response is good for protecting the power supply, but it may not protect the load.

Under a total short circuit (output leads shorted together), the power supply goes into its shutdown sequence also however, it seems to take a little less time to detect the short and shutdown. Again, it could wait longer to retry.

The average current supplied by the CRE-55465 in a fault condition (track short) is a little high because it retries so quickly after being shutdown. If the supply was used for constant track power, an output fuse would be advised. The CRE-55401 10 amp controller will work with the CRE-55465 (or any other DC power supply) and adds more reasonable overload protection. The CRE-55401 will shut down at a load of about 11 amps and it will wait several seconds before turning itself back on. If the short is still there, it will shut down again.

The meter is accurate to within a needle width, a substantial improvement over the older Ultima.

[ Top ]

Crest claims that this supply uses "computer control", but this isn't really true. It is a conventional switching supply. It uses mostly analog circuitry with some digital components. Nonetheless, its a very modern design.

AC power enters through the connector on the back and is wired to the front to the switch and then back to to the rear corner of the printed wiring board. The toroidal inductors and some other parts at the rear part of the board are the AC line filter to keep the supply from sending switching noise back out the power leads. There is a full wave bridge rectifier operating at line voltage and two large filter capacitors to create an internal high voltage DC bus. At 115 volts input, this bus will run at about 170 VDC. At 230 volts AC input, it'll be 350 VDC. You don't want to be poking around in there when this thing is plugged it. These voltages can kill you.

Under the rear heat sink are two switching transistors, probably wired in a forward configuration. The power transformer is in the middle of the board. This transformer can be quite small for the power it can handle because of the high switching frequency. Under the front heat sink is a low voltage rectifier. The yellow toroid coil wound with heavy wire and the capacitor next to it are the low voltage output filters. It doesn't take much of a capacitor to filter 75 kHz. Power is then routed through the meter and to the output terminals. Along the front left side of the board are the control electronics that sense the output voltage and generate the switching signals to drive the main switching transistors.

While inspecting my unit, I noticed that one of the leads on the meter was loose due to an improper crimp and the lug had some signs of heating due to a less than ideal contact. I recrimped the lug and pressed it back on the meter lug.

[ Top ]

The unit that I received did had a problem. It ran fine for a few days, and then it refused to start up. The power light would come on but the fan wouldn't run and there was no output. I sent the unit back to Aristo and it worked fine for them. When I got it back, it wouldn't work for me.

It was quite cold (50 degrees F is considered cold in Los Angeles) in my garage the evening that I got it back. On a hunch, I blew some warm air into the ventilation slots with a hair dryer. After about 10 seconds, it popped to life and seemed to work fine. After that it continued to work. I let it sit for about an hour to cool down and it wouldn't start again. The unit obviously had a cold start problem which is common in electronic equipment. The design hadn't been tested for cold starting and when Aristo test it, it was in their nice warm shop so it worked fine for them.

Both Aristo and the manufacturer of the equipment in Taiwan were able to reproduce the problem on other units by chilling them. After a few days of circuit analysis, Aristo's engineers determined that cutting a jumper labeled JP10 (located next to the main power transformer) and resoldering it to the second pin on the transformer fixed the problem. I made the change on mine and after that, it would start every time, even after being left in the freezer until it stabilized at about 20F. This modification changes the response of the overvoltage crowbar which was apparently tripping before unit could the unit could begin to regulate. This kind of problem is called a "race" condition where two things are happening at nearly the same time. The response is determined by which thing happens first.

There are perhaps 200 units in the field that may or may not have this cold start problem. Whether any particular unit has the problem or not depends on the results of the "race." Aristo says that all further units will be modified before shipment and any unit in the field will be repaired if necessary.

[ Top ]