I just purchased a new Yongnuo YN685 to replace a trusty but old Canon 580-EX (the original, not the II version) that is misfiring and may be on its last legs.

After playing with it for a few minutes I seem to have found a major problem with severe underexposure in bounce mode at longer focal lengths. My question is whether this is a sample defect, a design or firmware issue, or it's just how this speedlite behaves. As an event photographer I use bounce mode almost exclusively, so if this is just the way it works it is not usable for me.

I have obtained identical results using two different units purchased through completely different channels, so I'm leaning towards a firmware programming problem.

Below is a summary of my research, including sample images. I can make the RAW files available if necessary


I'm using the YN685 with a Canon 5D3 and 24-105mm f/4 L lens. At 24mm the exposure, in an appropriate bounce situation, (white ceiling) looks reasonable and about the same as what I get with the 580-EX.

As the lens is zoomed in the exposure gets progressively worse (under exposed) so that at 105mm, in exactly the same location, distance and bounce configuration the image is between 2 and 3 stops under-exposed. I know the flash has plenty of power because it produces the correct exposure at 24mm, and I can force the correct exposure at 105mm if I set the flash to manual 3/4 power instead of ETTL.

My 580EX has no problem with this situation at any focal length.


The following images demonstrate the problem. They were all taken on the above-mentioned 5D3/24-105 on a tripod with the flash mounted to the camera hot shoe. The camera was set to manual exposure 1/125 at f/8 with the flash head pointing straight up to a white 10 foot vaulted ceiling. The images were shot in RAW and then imported into Lightroom and immediately exported as 900x600 JPEGs with no manual editing applied.

The file names have the form


where [Flash] is the speedlite name, [FlashMode] is ETTL or M075 for "manual 3/4 power", [FlashZoomSetting] describes whether the flash zoom is manual or automatic, and the focal length, and [LensFocalLength] is obvious.

The first two show the results with my Canon 580-EX speedlight. When in bounce mode the flash head zoom goes to its maximum and the display reads "--", so the filename doesn't include [FlashZoomSetting].

580EX_ETTL_FL24.jpg Lens Focal Length 24mm

enter image description here

580EX_ETTL_FL105.jpg Lens Focal Length 105mm

enter image description here

The next two show the corresponding results with the YN685, demonstrating the extreme under-exposure at 105mm.


Lens focal length 24mm; Flash Zoom AUTO mode 24mm

enter image description here


Lens focal length 105mm; Flash Zoom AUTO mode 105mm

enter image description here

Next an of image with the flash in MANUAL power mode (instead of ETTL) to show there's not a problem with flash output.


Lens focal length 105mm; Flash manual power 3/4, zoom AUTO 105mm

enter image description here

At this point I surmised that maybe what is happening is that the firmware is unaware that the flash is in bounce mode, and when the flash zoom is at 105mm it reduces power because the zoom would be concentrating the light into a smaller area. So I shot in ETTL with the flash zoom in manual mode at the widest possible setting, 20mm. But that didn't produce anything different, so either my theory is incorrect or it's ignoring the manual zoom setting.


Lens focal length 24mm; Flash ETTL, zoom manual 20mm

enter image description here


Lens focal length 105mm; Flash ETTL, zoom manual 20mm

enter image description here


Flash  Flash  Flash   Focal   Exposure
 Mode  Power   Zoom  Length   Result
-----  -----  -----  ------   -----------------
ETTL    Auto  A  24      24   OK
ETTL    Auto  A 105     105   2-3 stops under
Man      3/4  A 105     105   OK (note 1)  
ETTL    Auto  M  20      24   OK
ETTL    Auto  M  20     105   2-3 stops under

(Note 1: just to show a correct exposure is possible in bounce mode at 105mm -- i.e. the flash has plenty of power).

  • \$\begingroup\$ No, but I'd be extremely surprised if that had any effect. Will try tomorrow and update the post. \$\endgroup\$ Feb 2, 2016 at 6:32
  • 1
    \$\begingroup\$ At least one user reports the evaluative/average metering issue with the YN-685 in the comments here: flashhavoc.com/yongnuo-yn685-speedlite-announced \$\endgroup\$
    – Michael C
    Feb 2, 2016 at 9:30
  • \$\begingroup\$ @inkista If you post your comment as an answer (with the link and some more detail to avoid being a "link-only" answer) I'll accept it. You are correct that it's probably too much to expect a flash costing 1/4 the equivalent unit from Canon to perform the same. That, along with the zoom motor noise, convinced me to bite the bullet and spring for the 600EX-RT. \$\endgroup\$ Feb 3, 2016 at 3:45
  • 1
    \$\begingroup\$ @JimGarrison Yeah, mean to write an answer first, but got lazy. :) Comments now converted to an answer. \$\endgroup\$
    – inkista
    Feb 3, 2016 at 17:43
  • \$\begingroup\$ @JimGarrison, btw, congrats on the 600EX-RT. Nice piece of gear. Only drawbacks are if you want to trigger studio strobes along with the -RT stuff (sigh), or use 2nd-curtain sync over radio. At which point, you may have to go with 3rd party triggers like (sorry) Yongnuo or Phottix. \$\endgroup\$
    – inkista
    Feb 3, 2016 at 18:25

2 Answers 2


This is a known bug with Yongnuo flashes (notably the YN-685's predecessors, the YN-568EX and YN-568EX II). TTL tends to be inaccurate and will underexpose unless you switch the metering mode from Evaluative to Average (see: this DPReview discussion, where one person claims Yongnuo support themselves suggested switching to Average metering).

This is, after all, a $100 reverse-engineer flash of cheap Chinese manufacture. While it may have some feature parity with Canon's flashes, it will always lag behind on copy and component consistency, backwards/future compatibility, warranty service, and resale value. They're great for hobbyist usage, less so for professional usage. While their QA has gotten noticeably better since the days of the notorious Strobist review of the YN-560, the low-low price has gotta come from somewhere. And if it's a new model and you're an early adopter, you may still be an inadvertent beta tester. Yongnuo has many quirks, and the fact that aficionados can tell you where to find manufacture date codes and which dates align with which silent updates is only one sign of this.

You may want to look into Godox's Wistro (bare bulb) and Ving (speedlight) flashes, Phottix's Mitros+ speedlights, or (if you want to go manual-only speedlight) LumoPro's LP180R for pro hardiness and reliability—not to mention better trigger interoperability (Yongnuo's got three separate triggering systems—none of which work together) if you still want to go third party vs. OEM. The Flash Havoc blog is a good source for what's out there.

Most folks purchase a YN-685 simply to have a cheap YN-622C off-camera slave. I love my Yonguo gear for off-camera, but for on-camera bouncing, I'll grab my 580EXII first, every time.

  • \$\begingroup\$ You are likely correct that "early adopters" function as beta testers at that price point. I finally got a response from Yongnuo service and have sent them a link to this question and also to the review I posted on Amazon. I hope they appreciate the amount of work I've put into documenting this for them :-) \$\endgroup\$ Feb 3, 2016 at 18:25
  • \$\begingroup\$ Jim, Did you ever get a response from them? \$\endgroup\$
    – MichaelH
    Aug 31, 2016 at 19:19
  • \$\begingroup\$ @MichaelH, might wanna move your comment to the question. As-is, I'm the only one getting a notification. \$\endgroup\$
    – inkista
    Aug 31, 2016 at 19:28

I had exactly the same issue as above. After sending one unit back and getting a replacement I was having the same problem. The fix for me anyway was to go in to the camera external speed-lite control in the setting menu / ETTL II meter and set this to "average" as opposed to the default "evaluative". Now getting perfect exposures in bounce flash scenarios. :)

  • \$\begingroup\$ You're just restating the basics of the accepted answer. This is, at best, a comment, not an answer. I know you can't comment yet, but abusing the system will just attract downvotes. I.e. Please don't post comments as answers. \$\endgroup\$ Aug 3, 2017 at 18:58
  • \$\begingroup\$ Sorry Jim, new to this site whilst looking online for an answer to a problem I've been having. I've just seen a thread with a lot of unnecessary chat without a concise answer - I've not deliberately "abused" the system I don't even know what you are talking about as visitor with downvotes. Just trying to help other folks with the same issue! \$\endgroup\$
    – Steve
    Aug 3, 2017 at 19:19
  • \$\begingroup\$ When you say "I had the same issue as above", what does the above refer to? The original question, or a different answer? Understand that Stack Exchange sites do not necessarily present all answers in the same order to all readers, depending on the number of votes received. Things change. Answers should be self-contained and address what's in the question, and be independent of other answers. Q&A sites like Stack Exchange are different than typical internet forums; in fact, they shouldn't be thought of as "forums" at all. \$\endgroup\$
    – scottbb
    Aug 3, 2017 at 20:00
  • \$\begingroup\$ Wow, I'll be sure to never post here again trying to help. Its pretty clear I'm replying to the original post. \$\endgroup\$
    – Steve
    Aug 3, 2017 at 21:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.