Plumbing Issues:
The easier way to install an intercooler is to use a Ford supplied compressor outlet/Y-pipe.  If you want to use the Ford parts, then this article by Terry Lang is a good place to start. I would not use the black rubber flex fittings though, as they are not rated for the temperature or pressure that this application will subject them to. Also, see my comments further down about mounting bottom of 'cooler to frame.

I didn't like the fact that the factory piece routes hot and cold air through the same aluminum casting.  The whole purpose of adding an intercooler is to deliver cooler air to the Y-pipe, and using a common casting is a step in the wrong direction. If you have seen pictures of a Banks install, then it should be pretty obvious where my inspiration came from.

Instead, tubing and a "megaphone" were purchased from Headers by Ed and used to make the pieces shown in the photo above. The megaphone was 2-4" diameter by ~25" long. This was cut in half, and the large end was creased lengthwise in two places to produce the transition section of the Y-pipe.  The small end was used to produce the diffuser section following the compressor. The taper included angle is approximately 5 degrees which is the theoretical optimum for a round cross-section diffuser.  2-1/4" OD x16 ga tubing was used for the bend between the compressor and the diffuser.  The 2-1/8" ID is a good match for the compressor outlet. As is obvious in the photo, there is generous clearance for servicing the fuel filter.

Nothing is square to anything else, and there are no good datum points/planes to take measurements from. The whole thing was and exercise in cut-and-try.  Lightly tack all welds and check fit before final welding.

Here are some photos of the Y-pipe. The arms are made from 2" OD x 18 ga. tubing.  The braces/brackets that secure the piece have not yet been added at the time I took these pictures:
View 1
View 2

I silver brazed (45% Ag alloy) some rings cut from 2-1/8" tubing so that the silicone tubing could not blow off the end. With this feature, there is no need to over torque the hose clamps, and risk damaging them, the hose, or the tubing. The tubing is stretched fairly thin on the outside of the bends, and is pretty easy to deform. The rings add welcome strength to the vulnerable ends.

I also glued (JB-weld) similar rings to the plenums that feed air into the heads. Sorry no photos.  If you try this, sand the paint off, and clean the bare metal with brake cleaner first. Wedge toothpicks under the rings to keep them from sliding down while the epoxy cures. I know, I know, JB-Weld is "hillbilly engineering". Just wait 'till you see where else I used it...Oy!

The lead photo shows that the tubing over those rings is green instead of OEM orange. This is Gates "Duron" coolant tubing. It is Silicone, Heavier than OEM, and I bought 3' stick (Westfleet or McMaster-Carr) for $18 less than the $76  Ford wanted for a pair (~3" each) of OEM tubes. The thicker tubing needs slightly larger clamps. Buy LINED clamps where you get the tubing.

The '99+ intercooler tubes drop down as soon as they are clear of the engine. This makes for a tight squeeze past the power steering pump on driver side, and there is a lot of coolant hoses, and battery cables on the other side. I bought some 3" ID elbows from Baker Precision. These allowed the ductwork to turn upward right behind the radiator, greatly easing the fit.

Each side needs two flexible fittings in the duct work, so that the engine can move on it's mounts, and the intercooler can move with the cab. The elbows above serve as one of them, and (clearly visible in lead photo)  I used a Donaldson "hump hose" (p/n P532962) for the second one.  Donaldson makes exhaust and filter stuff for large trucks. If you can locate a dealer near you, they have all kinds of other goodies that can be fitted to Powerstroke pickups.

Here you can see one of the hump-hoses, and the special constant tension clamps used with them. (that is a Breeze p/n showing) Note also the 3/8" compression elbow I added to the coolant vapor line so that it would clear the hump hose.. I drilled out the pipe thread end, and sweated a piece of semi rigid copper in to serve as a barb for the 3/8" rubber hose.

Mounting The Intercooler Itself
Some fellows just weld a couple of bars on the frame for the intercooler to rest on.  The top end of cooler is then bolted to the body.  Consider that the body is mounted to the frame by rubber isolators, and you see that the two move independently to some extent. Even though the cooler comes with rubber grommets ant the top,  and pads for the lower mounts, I didn't think this was a good idea.  Instead, I made these mounts for the lower end of the cooler to rest on. The three mounting bolts are 1/4"-28, and the body is tapped, so no nuts are needed.  This also gets the inlet & outlet spigots of the 'cooler up higher. This means that the lower U-shaped body piece (below radiator) does not need to be cut at all. This part of the body is highly stressed, and it is not a good idea to be weakening it...especially right next to the front cab mounts.

At the top mounts, others have used 3/8" bolts. I found that 7/16" bolts were a better fit in the bushings.  I tapped some 1/4" thick plates for the bolts, and ran them in from the back.  I had to run a die down the bolt shoulders, as they were only threaded for the first inch. The plates kept the bolts in position, and spaced the cooler so that it would clear the upper part of the hood latch brace. As you can see, the bolt holes fell on a body seam, so I milled a 1/16" deep step on the back of the plates so they would sit true.

The lower (crooked) end of the hood latch brace was cut off and replaced by a straight piece of 1" angle, with a piece of 1/4 X 1 slotted to fit over the stud on the bottom of the radiator support.

New brackets were fabricated for the lower mounts for the AC condenser. Same shape as the originals, but flat rather than dog legged.  I found a couple of scraps of 16 ga titanium in the scrap box.... If I'd bought the material, I would have used 1/8"  Aluminum instead.

Modifications To The Intercooler
The intercooler comes with cast spigots which have a 3-1/4" OD. I found no source for elbows which would mate to this. So I cut the casting off, leaving only about an inch of the spigots. These were then bored 2-3/4" through, counter bored 3" X 3/4" deep. Then 3" OD 6061-T6 tubing was glued (JB weld again!) into the casting. The photo shows two of the four 5/32" hardened steel dowels I used "just in case" the JB weld needed some help. I bought some real (not zinc) aluminum brazing rods that didn't work out. I couldn't get the casting hot enough, and was afraid I'd disturb the header-tube brazing if I kept after it.  The tubing stock was a little over 3" OD, so I turned it down, leaving a rough turned surface...the epoxy should have gotten a pretty good bite.

The stock intercooler spigots point upward and outward. To reduce trimming of the body, I installed the tubing square to the core. I forget the exact dimension, but I allowed just 3/8" clearance from each side of the radiator. This put the bored holes slightly inboard of center on what was left of the cast spigots.

Even though this tubing has a 1/4" smaller OD than the casting, the ID is actually about 1/8" larger...the 6061 alloy is much stronger than cast Al, so no problem. I did do a little porting & blending  just past the end of the tube...every little bit helps.
It is impossible to photograph, but I also rounded the casting internally where it transitions from the spigot to the plenum. This had to be done using a mirror, but mostly by feel.  Mounted points will quickly fill with aluminum and stop cutting, but a carbide burr on a die grinder made short work of this part of the job.

While I was turning the casting end of the tubing, I also cut a shallow groove at the other end. I glued a couple turns of aluminum mig wire into this groove, to serve as a bead to keep the elbows from sliding off.

Controlling The Cooling Air
Since my new spigots tubes were longer than the stock casting, I did not have to jump through hoops to get the 'cooler far enough back to get clamps on them.  Easier fitment, but now there was a pretty big gap behind each side. To allow the fan to suck as much air as possible through the core, I decided to fit 1/8" neoprene  skirts to the backside of the cooler. The rear of the skirts fit neatly into the 1/4" wide slot on each side of the radiator core.  To mount these, I made some "combs" (visible in background of this photo) by milling  some 7/16" slots into  2x1/8" hardware store aluminum strip The teeth fit neatly between the tubes, and allow greatly increased area to glue (JB weld again...I should buy stock) them to the headers.
I am particularly proud of how I clamped the combs to the headers while the glue set...5 arches where cut from yardsticks (cheaper than lath stock) and loaded with weights, creating strong outward force at the ends. The lights in the photo are providing heat so the epoxy will cure faster.  The rubber is rather soft, so it is sandwiched between the comb, and a second strip of aluminum with 1/8" pop rivets through the whole stack.

A third strip of rubber  is mounted to the top radiator support on the body, and forms a flap sealing the top of the radiator. It is carefully notched to seal around the hood latch.

The bottom gap of the cooler is not sealed. This is how the condenser and radiator were originally set up, and I thought there might be reasons for it.

Making It Last
If you clicked on the link to Baker above, then you might have noticed that silicone rubber fittings are NOT oil resistant. The Oil breather on the Powerstroke is connected to the inlet to the turbocharger compressor, and a fair amount of oil mist is thus thrown into the intake. 94-97 owners are all to familiar with the softening that occurs to the turbo inlet boot, and the 2" ID tubes at the intake plenums.

To protect the expensive silicone fittings used in this project, I have installed a Racor CCV-4500 crankcase ventilation filter system. It just fits between the fender and brake booster on a bracket I made. If you install one of these filters in this location, take great care to mount it low enough the hood doesn't hit it when closed, and that the hinge/spring levers don't snag the outlet hose.  It would be easier with some elbows at the filter, but the filter has oddball threads, and I could find no suitable elbows. By designing the bracket to mount the filter turned about 30 deg. I didn't need the elbows.  I don't know how well this filter will work, but Racor has a good reputation, and it is made for exactly this purpose....I'll update this article when I have enough miles on it to make a fair judgment.

The lead photo shows that the ductwork is just bare steel.  That needs to change. I want to slightly alter the diffuser piece, (hot side)  to provide more hood clearance, then I think I will have it ceramic coated. On the cool side, powder coat should work OK, but I might get that ceramic coated too,  just so it all matches...Or maybe my scots will come out, and I'll just rattle can it all with exhaust paint.
 

Performance
So how does it work?  With my uprated injectors and chip, I could push EGT through 1350F on hard acceleration even empty on level ground. Maximum EGT would have been even more, but I back off at 1350.  Now I am not exceeding 1150, even working at it.

Top end power is great, and lower end of rpm/boost range seems unaffected, but there was some lag in mid-range response due to lower intake pressure. Modifying the IDM (Injector Driver Module) helped, but more work is needed in this area. I tried plumbing exhaust pressure to the MAP (manifold absolute pressure) sensor, but that was a severe overcorrection...great response, but way too much smoke for me. I have a couple ideas I want to test, but I may be shopping for a custom chip if that doesn't work out.
 
 

NOTE:  These answers specifically apply to intercoolers installed on Powerstroke diesel engines.  The effect of an intercooler is different on a gasoline engine, and I'm not talking about them.  Also, other diesel engines will need to compensate for the intercooler in ways different from a powerstroke.  In some cases it may be simpler.

What is it?
An intercooler is an air-to-air heat exchanger plumbed between a turbocharger and the engine it is feeding air to. It looks like a big radiator, but has intake air on the inside instead of coolant.

What does it do?
The turbocharger heats the air as a natural consequence of compressing it.  With modifications, the powerstroke turbocharger will easily make over 30 psi of boost pressure. On a 80 degree day, the air will leave the compressor at around 350 degrees Fahrenheit.   The intercooler reduces this temperature to near ambient (depending on how well it is working) and in so doing the density of the air delivered to the engine is greatly increased, so the engine takes in a greater mass of air with each intake stroke....it breathes deeper as it were.  With more air in the engine, more fuel can be introduced, resulting in increased power.  More fuel can be added without the intercooler, but the high exhaust temperatures which result will seriously shorten the life of the engine.

How much power does it add?
In my case, adding the intercooler allowed me to safely run higher output fuel injectors which give approximately 70 HP increase to the engine.  The fuel system of a diesel must be modified to benefit....Actually there is a small efficiency improvement due to reduced pumping losses, so there can be a slight (probably only a few HP) power gain without fuel system changes.

But my Drag racing  buddy (who runs low 9's) said you get about 1% more power for every 10 degree drop!
The intercooler drops over 200 degrees, so it should add at least 20% all by itself!
Your buddy doesn't race with a Diesel engine, does he?  Power comes from fuel. Carburetors and FI systems increase fuel automatically in response to cooler (denser) air. Diesels don't.  Diesels are different...just listen to one if you don't believe me!

Why didn't Ford install intercoolers on the '94-97 Powerstrokes?
In these model years, Ford limited the power of the engine to reduce warranty claims on the the ZF-5 (manual) and E4OD (automatic) transmissions installed behind it. At this de-rated power level, an intercooler was not needed. Beginning with 1999 model year, the transmissions were upgraded, the engine unleashed, and the intercooler was needed to maintain safe operating temperatures.  Also, the powerstroke was introduced AFTER the 94-97 body-style was designed.  This engine was something of a shoehorn job, so omitting the intercooler eased the tight fit some.  The later trucks were designed around the whole package, so fit was not an issue.  Note that Ford still hasn't figured out a way to fit an intercooler into the Powerstroke powered E-series vans.

Why not just buy a kit?
This intercooler is one of the few good deals available at the Ford parts counter.  The cost of the intercooler was about 1/4 of what the popular kits cost.  The kits might be worth it, though, if you lack tools, time, or patience. Also, there are other parts needed to install it, which add up.  The silicone couplings in particular are expensive, to the tune of around $50 EACH....since you need four, that is $200 for some simple rubber tubes. (DON'T substitute rubber though, it won't take the heat and/or pressure!) Even if you buy all the parts new from ford, you will still save several hundred dollars over the cost of a "bolt-on" kit.

"Bolt on" kits are available from ATS, Banks, and Hypermax. (The URL for Hypermax was not working when I wrote this, but maybe they will fix it) The ATS is probably easiest to install, but least efficient, as it is smaller and uses smaller tubing for the duct work. The Banks kit is probably the most efficient, but also the most expensive.

Is there a downside to installing an intercooler?
If the PCM (engine control computer) program is not changed to account for the presence of the intercooler, then performance can suffer. The 94-97 computers monitor boost pressure and calculate allowable fuel based on the pressure.  Some people think that the intercooler restricts the airway, resulting in a pressure drop. This is not true. Though the intercooler drops a small amount of pressure, most of the "lost boost" is due to _reduced_ rather than increased restriction: Since the intercooler shrinks the air, it presents less restriction to turbocharger outlet, and lower boost pressure results, despite the fact that the airflow (mass flow) has actually increased.  The boost is "lost" prior to the intercooler, as well as after it.

Because lower pressure is measured, the PCM withholds fuel, and the result is lethargic acceleration.  This is commonly referred to as turbo-lag, but that is a misnomer. True turbo lag is caused by the rotational inertia of the turbocharger. Because the intercooler allows the turbocharger to deliver the same amount of air at a lower turbo speed, it can actually reduce true turbo lag.  "under fueling" would be a better way to describe the problem, but I'm not going to waste my time on that battle:  Since everyone refers to this as turbo lag, I might as well join the bandwagon..

Is there any way to cure/reduce turbo lag?
Yes! Several in fact.  Special chips (TS Performance) will solve the problem, by correctly accounting for changed density vs. pressure function.  Hotter injectors will do wonders. Modification of the IDM (injector driver module) to increase injector drive voltage will help. Smaller A/R turbine housings will decrease lag.  The pneumatic signal (pressure) to the MAP (manifold absolute pressure) sensor can be altered, or the electrical signal from the MAP to PCM can be altered.  Modifications to exhaust system which reduce back pressure will also be helpful.

All racers know that water-air 'coolers work better, why are you wasting your time with air-air?
Water-Air coolers can work well in gasoline engine applications where boost is only used in short bursts....drag racing in particular.  If the engine will be boosted for long periods at a stretch (quite normal for diesel applications) then  water-air cooling is of no real advantage. A radiator larger than the A-A cooler would be needed to maintain a supply of cool water, and with two extra interfaces, the W-A system will be less efficient at cooling the induction air.  For a Diesel, I would only consider W-A cooling in marine (duh!) applications, or for a racing or pulling application.