There are two ways of dealing with this old equipment… Replacing it with something new is the preferred way. When it becomes difficult to locate spare parts, when it breaks down too often, just replace it with new gear. For much of the equipment this is the usual answer and is often a major part of the job.
Some equipment is not so easily replaced. When replacement would require wholesale redesign of a system it becomes more of a challenge. Sometimes the only choice is to keep that old gear running.
This is the case with our servo amplifiers. Twelve amplifiers supply the power that drives the telescope, one amplifier for each motor. Eight amplifiers and motors drive azimuth, four drive elevation. Three hundred and seventy tons moved by twelve relatively small DC motors. While much of the telescope control system was recently replaced, it was decided to keep the old servo amplifiers.
You might notice that these servo amplifiers are just a wee bit critical.
My current problem does not concern the amplifier itself, but the power supply. Each amplifier has a dedicated power supply paired with it, a unit about the size of a toaster that converts the AC supply to DC. These units were built by Inland Motor decades ago, and have not been manufactured in this century.
The documentation calls this black box a FCU-100-30 filter control unit, we call them power supplies. Besides a high current rectifier and a couple beer can sized capacitors each power supply has some additional functions. This includes a 28Vdc regulated power supply and some control circuitry for turning power on and off. The unit also shunts the back EMF generated by the motor while bringing a nearly four hundred ton telescope to a stop.
In this power supply are a couple circuit cards, one of which is known to have a few problems. The manufacturer apparently knew about this, at least twice the circuit card was revised to fix some of the issues. There are examples of three revisions in the units we have.
Over the decades of operation at the observatory these circuit cards have been repaired many times. They have become quite hacked up, torn solder pads and traces where components were removed and replaced, small burn marks from overheating components, wire jumpers used to patch the damage.
We just need to replace the boards. Of course these things have been out of production for decades, the manufacturer long ago bought out by a larger company. They occasionally pop up on eBay, but the units you find are just as beat up as ours.
What we do have is a complete set of drawings. A schematic, a bill of materials, everything. They may be old, barely legible scans of the original documents, but we have them. Schematics are simply worth their weight in whatever precious item you name when trying to repair old gear.
Just re-layout the PCB and manufacture new parts… Easy!
New printed circuit boards take much of a weekend to lay-out. This part is just fun, I love PCB layout. Creating a PCB layout is the most complex puzzle I know, with dozens of things to consider with each copper trace drawn across the screen.
A few of the parts are not manufactured anymore. Motorola MOC5008 optocouplers? No longer made, out of stock everwhere. Motorola does not even exist as it once did. But Toshiba TLP104’s will work just fine. New layout? May as well convert everything to surface mount.
The drawings are occasionally difficult to read so I cross check against the real board, checking component values and ratings. Where possible I keep the original reference designators to make this easier.
These new units may be in service for decades to come. Choosing higher rated parts and long lifetime rated parts is also a good idea. I do not need to penny pinch like the original manufacturer building thousands of units, a few extra dollars for the good capacitors is cheap insurance.
Those design flaws in the original version? Mostly overheating of the linear regulators. Answer? Just replace them with modern switching regulators. I delete the old LM338 linear and replace it with an LM2593 switcher. The old design allowed for 1.6 amps of continuous output, so a two amp switcher will do nicely.
The details in a PCB layout seem endless, you have to check everything thrice. Even after doing so you just have to accept that there will be some small error. Hopefully something you can live with. Eventually you zip up the CAM files and send them off to a board house with a credit card number.
Do you need to build a few of these? Here are all the documents from the old Inland Motor schematics to the new PCB gerbers. The board files are in Eagle if you need to edit them. With a few component value changes, mostly as found on the old schematics, these boards should work for most of the FCU-xxx-xx series.
It is always a joy to get the PCB’s, a small box of tech magic in the mail. I see these fiberglass boards with copper tracery as a form of functional art. A well designed PCB is art, more beautiful to me than much of what you will find in the local galleries.
It is enjoyable to spend a hour or two soldering components to the PCB, always a soothing task. Better yet when the board powers up for the first time. I spend a while probing the PCB with the meter, finding the right voltages in the right places.
Now there is the challenge of installing into a FCU chassis and testing it to insure all of the functions work properly. Then installing it in a telescope? That will happen next week, I have two FCU power supplies missing these boards, waiting to be repaired.