Variable Audio Recording Delay

I've been combing forums and the interwebz for days now trying to figure this out, and I'm finally throwing in the towel and asking for help.  The search terms are all to common in other situations to really figure out whats going on.  This is all audio, no midi.
 
I'm using X3.  I've been getting my new studio situation set up, and all is working well except for some problems syncing with my metronome. At first I thought I had completely lost my rhythm!  Which was kind of funny, as I've been out of the business for a while, but I have a percussion degree! lol.
Anyway, I did a lot of troubleshooting to try to get it working right, and eventually had to settle on manually adjusting my Recording Latency Adjustment.  Seemed like it would work, though I was somewhat concerned at my audio-loop back test showing about 1800 samples (about 40ms) off while my control panel shows: "Effective Latency at 44kHz/stereo: 5.8msec".  After trying it out a few times with the new setting (doing most all of my testing with a simple output-to-input loop back), it seemed better.  But, now I'm finding that I have a wide variability in the synchronization of my recording.  Sometimes, the manual delay I entered is perfect, though more often I'm seeing a variability of 5-25ms in where my audio lands. 
 
At first I thought it may have to do with processor load, but changing my buffers from min to max and everywhere in between (in both Sonar and my interface's control panel) seem to hardly affect it at all.  And yes, I would run the wave profiler after every adjustment.  from what I understand that makes sense that it wouldn't matter in the recording timing, even though my computer runs flawlessly at minimum latency, as the wave profiler is supposed to adjust for these.
 
Please guys, help me out :(  there is no way I can get my reputation up and running in the new city I moved to with a random 25ms delay on every cut I make.  I can of course just use a recorded click track, but the metronome sync isnt the issue -- its the recording lag being different EVERY TIME I press record.  --- interestingly, though, it tends to be consistently one of five different amounts of lag, tending to be perfect for my manual adjustment, or about 250, 500, 750, or 1000 samples.  These are consistent values, and the cuts that line up with those delays line up with eachother perfectly.  (add about 1800 to those for the absolute amount of delay I'm getting in a system that should, I thought, be self-adjusting)

Not sure its relevant, but keep in mind I'm showing minimal latency in the preferences menu and no problems with dropouts or processor load.
 
My setup is a Tascam M-164UF (not super pro, I know, but was an affordable multi-input interface and gives me lots of options doubling as a mixer, and is generally very quiet), windows 7 64-bit, 8gb ram and a snappy processor.   I have tried using various driver setups (MME, ASIO4ALL, etc) but none of them seem to function properly for this audio interface other than the WDM.  I have tried just about every combination of toggling 64-bit processing, buffers, mmcss, dc offset removal, sampling rates and bitrates, multiprocessing engine, and im sure a few more.   My recording clock is set to the mic input/output on the Tascam.
 
 
Thanks so much for anyone that even takes the time to read my wall of text!  I'm pulling my hair out over here!
 
-gwen
 
 
 
 
 
 

You can download a 64-bit Windows driver (as well as Mac and 32-bit Windows) here.
 
Also, TASCAM is one of the companies with known audio interface issues if you have ASIO4ALL installed on your computer. It's not enough to disable it; you need to uninstall it.

quick question -- would getting usb ports dedicated specifically for my interface, say with a PCI USB card, be a good idea?
 

Thanks, I'll keep that in mind in this situation and the future as well :)
 
So I got that utility (freaking awesome!), and it's showing roughly the same thing.  The jitter isn't as common, on the reading as it seems to be in sonar, but it wil occasionally show me different readings as well.  I removed every USB device except the mouse and keyboard, and tried a few different ports (I did see a difference in latency's with different ports -- so at the very least I've got a good new trick). 
 
Since I was showing a 30ms measurement latency with a 4.35ms buffer, I decided to try it out on a different computer.  I got almost identical results (about 7 samples different).  The other computer is showing a bit of inconsistency as well.
 
I got this mixer/interface used, as its a discontinued model and can't afford to drop 1500 on an Onyx 1660 or whatever...  sounds like it might be a hardware problem?  =/   Might have to downgrade to a little interface box on the desk for a while.  bummer...

Good job checking out all the variables. Strange that CEntrance is finding so much "hidden" latency. I can't imagine how something would be wrong with the hardware to cause that. You'd think it would have to be a problem with firmware or drivers. But if that's all up to date and compatible with your O/S (Win7?)... I dunno. When Googling for info I found this forum member's signature showing he's using an M-164UF; you might try pinging him:
 
http://forum.cakewalk.com/Profile/67841/

Yes, I meant PM. I sent one for you to see if he's around and wants to weigh in on the thread.

Cool. Thanks, Tony. Maybe let her know what driver version you've got installed while you're at it.

gwen/Dave-  My Tascam driver is the one that came with the M-164UF, so it's old.  I'm still away from my studio and won't return till tomorrow.
 
(The M-164UF driver is available on the Tascam website.)
 
Tony

There isnt some kind of fancy setting somewhere that would have it automatically run a silent pingback every time recording starts and offset the cut by that much immediately after is there?  lol
 
 
PS:  Just to clarify for one of the replies, I have removed all USB devices other than my keyboard and mouse and tried many different ports.  Its definitely something I'll remember in the future because I shaved off some time, but it didn't begin to touch this problem :(
 
-gwen

Lifeismusic,
 
Some of the symptoms you describe sound to me like DPC latency issues, usually caused by drivers for other components on your system, such as networks. In addition to all the other good things that have been recommended and you have done, you may want to download and run LatencyMon, which will tell you if your system is suitable for processing real-time audio, and identify the drivers that cause the highest DPC latency, among other things.  If it detects unsuitably high DPC latency, you can try disabling the component whose driver is identified to see if that improves the issue.

gwen
On Vista, W7,W8 there is something called the driver store. WHen you install a USB device the driver is put into the store. I have encountered situations where multiple drivers for the same USB are contained in the store. If the manufacturer properly writes a new driver, when it is installed it replaces the old one in the store and there are no problems. Unfortunately not all manf's write correct drivers. I design USB devices and have to deal with this constantly and it causes all kind of problems.
If you have installed the same driver multiple times or updated to another driver without uninstalling the old one, it might be a good idea to get rid of ALL the old ones. 
Microsoft does not make this easy but there is a rather easy way to tell if multiple drivers are installed.
Open a command prompt as admin.
type in
pnputil.exe -e
hit enter
You will get a list of all  drivers in your store. There may be like 30+ depending on your system
 
Here is just two fro my laptop.

Published name :            oem5.inf
Driver package provider :   E-MU Systems
Class :                     Sound, video and game controllers
Driver date and version :   10/07/2010 6.0.01.0001
Signer name :               Creative Labs Inc

Published name :            oem1.inf
Driver package provider :   Realtek Semiconductor Corp.
Class :                     Sound, video and game controllers
Driver date and version :   07/28/2009 6.0.1.5904
Signer name :               Microsoft Windows Hardware Compatibility Publisher
 
Find Tascam in the list then see if there is more than one Tascam listed.
You can scroll up/down in the command window.
You could just post it if you want and I will take a look.
If there is, then it might be worthwhile to make the effort to remove all the dupes and
start fresh. This is kind of a pain and is why I posted the above. But I can point you to the process if need be.
 
Michael
 
 

gwen... FYI, my Tascam driver is indicated to be version 2.1.4.0 in my computer Device Manager.

dang!  i about jumped out of my chair.  my device manager shows the right driver, I thought you meant the tascam control panel at first where it shows 2.01.  Thanks a bunch though Tony :)  hopefully I won't have to bug you anymore!
 
-gwen

Welp, USB cable made the latency a little less but I'm still seeing variability.  The 25ms latency I have over what the control panel says it should be (I imagine that is monitoring latency and it would be roughly double to bounce it back again through the mixer and to the computer).  Interestingly, now I'm showing identical latencies on both my recording computer and my laptop (30.17ms on a 4.35ms setting in the tascam control panel).  I guess my next step is to install the DAW on the other computer and see if I get the same variability when bouncing back my metronome.
 
This computer was sitting for a couple years while I was away from home for work and I do seem to be having a small bit of trouble with some things on it (a certain USB wireless mouse works fine on my laptop, intermittently on my boyfriends computer, but not at all on this one -- also a fresh install of windows is giving me long file-access problems as if my harddrives have gone a bit bad).
 
If you get the chance Tony, could you run that CEntrance utility and tell me what it gives you?  Maybe let me know just how you're routing the signal back to your computer, too, I might be adding extra jumps I don't need.
 
Thanks again guys :)
 
-gwen

here is an example of my recording variability;   imgur.com/lZEB1OF
 
 
First track is straight recording metronome, the rest include track one so I can see the difference, if any, between straight computer bounce back and replayed audio bounceback.
 
I've also tried it where I silence the previous tracks instead of muting them (maybe muting keeps the computer from ever loading the sample, while having the gain turned inaudible would still add for the factor of computer stress?)
The result is about the same:    imgur.com/uQ65e18
 
After doing this countless times in different ways, it does seem that there is a general trend (though any one image like this would give a wrong impression as to how frequent or infrequent it is) of the latency being its shortest with the first track, having a few that are more longer but identical, then getting longer as more tracks are added.  However, this is not consistent, and they still tend to fall in the exact same places.  Interestingly, after an audio dropout (they're rare but do happen), as seen in the last recording of the first image, the track returns to it's lowest latency even with all of the other tracks present.  Usually.  Even the first track isn't very consistent.  It almost seems as if the system is building up stress which causes the triggering of the latency to get more likely.
 
Hopefully this will not be repeatable on my laptop.