(Starting here with the hot shoe extension cords problems, but which are the least of the problem. So don't miss the hot-shoe flash example, not on a cord, of same problem and same solution, Below)
There is a very serious issue causing Nikon brand flashes using zoom lenses to often underexpose TTL BL direct flash shots. There are two cases shown here, that most zoom lens can report inaccurate D-lens distance, and that hot shoe extension cords place the flash at a different distance than the lens. The manual warns about hot shoe extension cords, but it often affects regular flash shots too. However in either case, third party flashes (non-Nikon brand) are exempt from the problem. You may have trouble believing this situation can exist, especially the reason seems unimaginable, but I assure you, it does exist... and it will be very easily verified by your own tests.
Basics Note: The flash menus generically say TTL (Through The Lens metering), but the Nikon system specifically actually defaults to be TTL BL (balanced flash mode). There is also a regular not-balanced mode called TTL. The three camera metering modes (Matrix, Center Weighted, Spot) are only about ambient light - the flash system uses its own central metering system instead. Ambient metering can influence TTL BL level (balanced flash), but for flash, Spot metering shifts flash metering from TTL BL to be TTL mode (not affected by ambient levels). So for cases of flash in dim ambient indoors (level too low to matter what it is), or in camera Manual mode (Spot cannot change camera settings), Spot metering to convert TTL BL to be TTL flash can be a good thing (even though the flash system does not use Spot metering itself).
A main cause of the TTL BL direct flash underexposure problem is when the reported D-lens distance is less than the actual flash distance. Because zoom lenses report distance inaccurately, or an off-camera flash really has little chance of being at the same distance. The story is:
The case to be shown next is the speedlight in one umbrella, on a hot shoe extension cord (an old Nikon SC-17 cable here, which is same as SC-28, except for the newer shoe lock). All very straight forward as stated, TTL BL flash in an umbrella.
Four third party flashes (not Nikon brand). - These pictures were four different setups at four different times, possibly four slightly different distances, so very similar but not exactly alike. No attempts were made to compare the four tests, but is only showing that the same D-lens results always apply.
Exposure was NOT adjusted here, only the flashes and metering mode was changed as shown. The bad SB-800 result is known and expected, and intentionally created here (simply by using TTL BL and a hot shoe cable, and camera closer than the flash and umbrella). The D-lens reports the short lens distance, but the flash has to use more power for its much longer actual distance. The Nikon system imagines the distances are correct, and the D-lens distance overrides the metered value, and reduces the imagined "excessive" amount. How much underexposure depends on the difference in the actual distance and the inaccurate reported distance. Spot metering selects TTL mode, which specifically is Not affected by D-lens error. What makes this interesting is that the third party flashes are not affected either, their direct TTL BL is not affected by D-lens data (which seems a very good thing).
Center Weighted metering
Center Weighted metering
Nikon SB-800 Center Weighted metering
Nikon SB-800 Matrix metering
Nikon SB-800 Spot metering
Yongnuo YN565EX Center Weighted metering
Yongnuo YN565EX Matrix metering
Yongnuo YN565EX Spot metering
Nikon SB-800, Center Weighted metering
Nikon SB-800, Matrix metering
Nikon SB-800, Spot metering
Neewer NW985N, Center Weighted metering
Neewer NW985N, Matrix metering
Neewer NW985N, Spot metering
Center Weighted metering
Neewer VK750 II
Center Weighted metering
Neewer VK750 II
Neewer VK750 II
In this case (flash on hot shoe cord, relatively far, up and back from a reflected umbrella, but the camera and lens are much closer to subject), then the Nikon flash is always underexposed in Center Weighted or Matrix metering, but Spot metering is not. Again, this applies only to TTL BL direct flash, but the flash head in the umbrella is not tilted, so the system interprets it as direct flash. Again, the flash system does NOT use Spot metering to meter flash. Spot only changes TTL BL to be TTL metering.
Seem odd? It gets even worse below, but there is a story, and here's the deal with the pictures above. Nikon flash heads have a switch on the head tilt, so the camera knows when the head is tilted or rotated (for bounce). When not tilted, the assumption is that it is doing direct flash. And then the rules (for TTL BL) are that the D-lens focus distance serves as an over exposure guard. Because, direct flash suffers inverse square law falloff, so the background behind the direct subject is often dark. The metering system sees that dark too, and naively tries to brighten it, often resulting in overexposure of the direct flash subject. But the D-lens reports the distance where it is focused (however many zoom lenses simply report that incorrectly) - perhaps very incorrectly. Because the TTL BL system looks at TTL metered exposure, and looks at subject distance, and recomputes with Guide Number, and may say "Whoa! That's too much power for this subject distance", and it cuts back on the flash power. This could be good, if the D-lens distance were accurate (this is done because direct flash backgrounds do often meter dark, due to inverse square law). It only does this "correction" if TTL BL mode, and only for direct flash (non-tilted head), only if on the hot shoe (the hot shoe extension cable is "on the hot shoe", even in this case of being in an umbrella). Also, a flash off-camera is not likely where the lens is, but the system is not aware of that. And if no match between the reported lens distance and the actual flash position, then this "correction" can cause underexposure (above - non-tilted TTL BL flash on hot shoe cable at rather different distances).
Two issues cause D-lens error, flash not on camera, and D-lens distance not accurate in the first place.
For this reason, Nikon advises us to switch TTL BL to be TTL mode if on a hot shoe cord (flash not same distance as camera). Nikon just does not provide a convenient way to do that any more. Older flash models had a menu to override TTL BL with TTL (in current models, only SB-910 does now). But selecting Spot metering will do it (switches to TTL mode for flash), and which is NOT spot metering for the flash. For ambient it is Spot, which can be ultra important, but for flash, it is only TTL metering. Third party flashes however can simply cause the D-lens distance info to be ignored. I like that. :)
The point is, the Nikon pictures above show the problem, and the third party flashes don't. They don't have the tilt switch on the flash head. The subject in the above pictures is 1.5 feet from the lens, but is 4.5 feet from the umbrella fabric, and another 1.5 feet back to the flash head is 6 feet, so the flash distance needs a lot of power. It uses a hot shoe extension cord to the flash, so the system sees it as a hot shoe direct flash (with non-tilted head) at 1.5 feet. So the Nikon system thinks this case must necessarily be metering an excessive direct flash power for a 1.5 foot subject, and it cuts back on the flash power, hoping to prevent overexposure, but instead causing underexposure (dumb computers!) Exposure becomes a crapshoot, but wrong is the right way to bet. This is TTL BL direct flash, which is of course metered, but the inaccurate D-lens focus distance can override the metered value.
Spot metering mode is a bypass for this issue (FV Lock is another, or TTL flash mode is another). The flash does NOT use Spot metering mode, which is only for ambient, and which is only about the spot, which ignores any background ambient level. So Spot metering cannot do balanced fill flash, and so if Spot metering, the system converts the flash metering mode from TTL BL to TTL mode... and TTL mode is not affected by D lens distance, so that exposure comes out much better, using what it actually meters. Downside is, if you forget to reset Spot metering off when done, it will disrupt your next pictures outdoors.
Safe zones - Bounce flash ignores D-lens distance, since the flash head is aimed up, following a different path. The Commander ignores D-lens distance, since the system knows the remote flash may not at same distance as the lens. FV Lock ignores D-lens distance, since it is used to re-aim the camera for the final picture (we can get a very different picture with FV Lock vs without it, even if we don't re-aim). Actual TTL flash mode ignores D-lens distance (but the system is TTL BL default). GN flash mode which depends 100% on distance ignores D-lens distance (would not be accurate enough). These are good things, but can be a cause of seeing different results.
And all of the few third party flashes I have seen (Yongnuo, Neewer, Aperlite) respond this same non-Nikon way, which seems a plus to me.
Normally the flash head is NOT tilted when used in an umbrella. That can be a big problem for Nikons if TTL BL is used with a hot shoe extension cord (above example) - because the flash is simply not where the camera is. And a different factor, most zoom lenses report distance very poorly anyway - incorrectly (more detail).
Re: Nikon TTL BL System - Nikon flashes vs Third Party Flash models
But third party flashes don't report tilt/rotate, and clearly allow use of the actual metered value instead of being falsely limited. Score third party brands a big point. See next below.
It's not just hot shoe cords. Because the D-lens distance is often not accurate, it affects regular photography with hot shoe flashes too. Zoom lenses often report crudely inaccurate focus distances, which messes up direct flash TTL BL. And most flashes have no TTL/TTL BL menu, so the system defaults to TTL BL. Third party flashes (non-Nikon brand) are another way around that, shown here. This is a very dark nighttime scene, no ambient light (black without the flash). Just the hot shoe flashes were swapped out, and the metering mode switched, as indicated. This is TTL BL with a Nikon 16-85 mm DX lens at 16 mm (it's a great lens, but like other zooms, poor about D-lens distance).
Yongnuo YN-565EX, Matrix metering
Yongnuo YN-565EX, Spot metering
Aperlite YH-700N, Matrix metering
Aperlite YH-700N, Spot metering
Neewer VK750 II, Matrix metering
Neewer VK750 II, Spot metering
Nikon SB-800, Matrix metering
This seems a big problem
Nikon SB-800, Spot metering
See? Here there is not even any difference in the distance of the hot shoe flash and the camera, but the D-lens seriously screws it up (for the Nikon flash). The third party brand flashes exposed the same better result in Matrix and Spot metering. It is not about spot metering, the flash system does not even use Spot metering. Ambient metering can, but it was pitch black dark here, no ambient for Spot (and it was camera manual mode anyway). But Spot metering switches TTL BL metering to be TTL metering, which ignores D-lens distance. However the third party flashes were already causing D-lens distance to be ignored, which seems a big plus. (I usually tend to say the flash did this or that, but of course the camera does everything, the TTL flash just flashes as directed by the camera).
I realize some may think this says Nikon flash units are better when it is Spot metered. But it does Not say those words, and it is not even about Spot Metering. Again, the flash system never uses Spot metering to meter flash. Ambient metering can, yes, but Spot metering mode only changes flash from TTL BL to be TTL metering (which also happens to disable D-lens effect on direct flash, which yes, that certainly can be better when it is so wrong, a really big deal here). But the camera system NEVER spot meters flash. Flash is always metered in its own larger central zone.
But the TTL BL flash metering was overridden and held back by the D-lens distance (when the Nikon flash reported that its head was not tilted). This was the Nikon 16-85 mm lens at 16 mm. This distance was measured to be 19 feet (5.79 meters) but the lens reported it as 2.00 meters (6.56 feet). Honest, I'm not kidding, at 16 mm zoom at 19 feet, it does report only 1/3 of the correct amount (this can vary slightly around 2.00 meters, but it will never be near correct at 16 mm). What is worse, the system design plows ahead on, using the number as if it were accurate. It sure seems a very questionable plan. Nikon has to know zoom lenses cannot report accurate distance. Sure, the flash needs lots of power for 19 feet, but the system says Whoa man, that's way too much power for 6.56 feet, and it drastically limits the flash power to be correct for 6.56 feet, to prevent the supposed overexposure. So we suffer underexposure unless we select Spot metering to do TTL metering, which ignores D-lens distance. As do the third party flash brands ignore it (no head tilt switch). Spot metering has no other effect on flash (it only switches flash metering out of TTL BL mode)), but if we did have ambient here, the ambient metering could be greatly affected by Spot metering.
A few specific lens details are shown, but we cannot predict what a zoom lens will do without testing it. They react differently over their zoom range. They may even report different distances just by pressing the shutter a second time (focus seeking). It's a crapshoot.
You should be easily able to repeat similar bad results yourself. It is a design factor in the Nikon system, using inaccurate zoom lens D-lens distance data. Most of what we blame on the flash is not the flash at all. The camera does the metering decisions. The reported Focus Distance is in the Exif for all to see, and ExifTool is a great tool for this.
Again, work-arounds to ignore D-lens distance are:
But otherwise, Nikon brand speedlight TTL BL direct flash may suffer. And TTL BL is the system default.
None of the story above is about the flash, it is all about the Nikon system. However, the third party flashes do not report tilt switch, so things come out better in this case (see pictures above). No tilt switch causes the D-lens data to be ignored, which sure seems a plus. However, the mode is still TTL BL, possibly affected in some degree by the ambient level (if significant level), and of course, TTL metering is always affected by a white or dark background, so the top cord pictures above really could have routinely benefited from +1 EV flash compensation (for either/any flash doing TTL BL indoors - and of course, for all this white in the scene too). In this third party case (in lesser indoor ambient), the switch to TTL increased exposure about 2/3 stop over TTL BL. White Balance correction is also shown again. It's a preference, but due to the white background (which of course affects reflective metering), the flashes really need an additional +2/3 stop. Flash compensation is how we control automatic TTL, and it lets you watch the results, and then to do what you see you need to do. Simply do it, get what you want.
If there is a point, it is that much of what we see happen is about the camera automation, and is really NOT about the flash unit. A flash could malfunction or misbehave, but the TTL flash does what the camera system tells it to do. If you have a Nikon flash, and if using a non-tilted flash head on a hot shoe extension cord (not likely at same distance as lens reports), and you cannot get it out of TTL BL mode, at least make sure the flash is substantially closer to subject than the camera (however, without inspection, we don't know what phony numbers the lens might report at any given zoom setting). Tilting the Nikon head, selecting actual TTL flash mode, or Spot metering mode, or FV Lock will bypass the D-lens distance report. So will a third party non-Nikon brand flash.
Soap box: The more we learn about D-lens data (and how very poor it is), the more horrible it seems (at least for zoom lenses). We know of course how to easily deal with any overexposure of direct flash, but there is not much defense to the automation just deciding to screw things up by using random numbers. Only TTL BL direct flash suffers from it. We have workarounds, but ...
We seriously need two new camera menus:
Or, we can use a non-Nikon brand flash to prevent the problems. It is not even the flashes fault, it is the Nikon D-lens system, but third party flashes manage to bypass the problem (no head tilt switch to report). Their price is good too. We don't have to live like this. :)