Topic: Sound Devices MixPre-3 and 6, Part 6

[rick.lang]

Paul, thanks for the feedback!  I didn’t know about the local calibration of the MixPre-10 II to an external Timecode clock.   Interesting my second 5.5 hour test didn’t change from that early loss of a part frame.  So it’s still just barely less than a frame but didn’t get any worse in the last 4.5 hours!

I don’t know if there is value in shooting 23.976 fps as I’m shooting 24 fps but I’ll do as you request of course.  Then I’ll also test jamming the Tentacle to the MixPre clock and using that Tentacle to jam the other Tentacle.  Will leave both Tentacles on the two cameras and see how they all fare after that method.


Sent from my iPhone using Tapatalk

[jerryfreak]

I wish I knew there was an update coming.Unfortunately as of yesterday I don’t have most of those cards anymore as i had to return them to Amazon. (Yes I’m a bad Amazon’er but unfortunately I can’t personally subsidize hundreds of dollars to test their firmware).

I did a successful recording this weekend but I’ve kind of long since given up on recording anything over six channels of 24/48 with the mixpre. I encourage you guys who still have cards to test them with the new mixpre firmware at 8 channels of 24/192 so we can see if the problem is indeed resolved.

[rick.lang]

Jerry, I need to record 6 tracks so I’ll test that to see if it works this week.  If that works, I’ll go for 8.

I have the only recommended card, SD32GB.  I’m also not adverse to trying my WISE SDXC 128GB card as it should be fast enough even running in UHS-I mode.  It’s 285MB/s in UHS-II mode.  Of course raw speed isn’t as important as suitability to the task.  WISE are tested against theIr own Blackmagic Pocket Cinema Camera 4K which is where I normally have used it without issues.


Sent from my iPhone using Tapatalk

[jerryfreak]

Jerry, I need to record 6 tracks so I’ll test that to see if it works this week.  If that works, I’ll go for 8.


Sent from my iPhone using Tapatalk

Great, Rick! we have a thread for results here

http://taperssection.com/index.php?topic=191314.msg2311409;boardseen#new

If you can try to test all the way to the end of card as it seems that a lot of failures seem to increase on the back half of the cards’ capacity

[Paul Isaacs]

A few points if you choose to do intensive SD card testing ...

- Start from a clean slate i.e. low level format your SD card using SD formatter, then quick format the card in MixPre v4.01 before running the first test. Subsequent tests you can simply do a quick format in MixPre v4.01

- You should see significant improvements across the board but we cannot guarantee that every card on the market that quotes a sufficiently high write data speed will work. We have seen cards whose internal house-keeping/garbage collection algorithms can cause them to temporarily bottleneck the data flow.

- We do not recommend microSD cards in adaptors as the adaptor itself is a potential weak point in the data path.

- Understand that an SD card itself can go faulty or be fake. When we do our approval tests, we run those tests across multiple cards of the same model to ensure consistency before we approve.

[rick.lang]

The first firmware 4.01 test ran for 6.5 hours, losing part of a frame in 15 minutes and going over a few hours when it lost less than two frames by the end.  Next test losing a slightly larger part of a frame in ten minutes and I’ll monitor this for several hours.

What’s good is that I may no longer see those chunks of a second or more lost suddenly.  This gradual loss is similar to before 4.1 but doesn’t get ridiculously bad.  Still losing Timecode so quickly means I can’t use it.  On a real shoot, I’ll have it taking TC from the Tentacle continuously when I can...

... MixPres are calibrated to an atomic clock. I suspect that so are Tentacle's but don't know that for sure. Run the test again with both devices set to 23.98 and if you still see that level of drift, contact our tech support team to arrange a calibration check...

Paul, I set all devices to 23.976 fps; the test to jam the Tentacles from the MixPre Gen Timecode vía the stereo out port ran for 9 hours and appears to be flawless.  Both Tentacle devices match the Gen clock.  No need to calibrate or blame the Tentacles or MixPre for inconsistencies.

Currently running (over an hour) the test of MixPre with Gen Timecode jam synced to a Tentacle.  Monitoring that via the Jam screen, there is no difference detected at 23.976 fps.

So you must be relieved as I am.  I’ll let it run longer but doubt there is a problem under 4.01.

However, I still want to verify if running at 24 fps now performs properly.

I can do a low level exFAT erase/format on my Mac (Disk Utility) as requested before stress testing 6/8 tracks armed writing 192KHz 32bit Audio.  All I plan to use at this time is 6 tracks and 48KHz 32bit audio but I’ll do the stress test using your own approved SD32GB card.  In practice if I was really running those rates, I’d test with the Wise 128GB card I have.


Sent from my iPhone using Tapatalk

[jerryfreak]

A few points if you choose to do intensive SD card testing ...

- Start from a clean slate i.e. low level format your SD card using SD formatter, then quick format the card in MixPre v4.01 before running the first test. Subsequent tests you can simply do a quick format in MixPre v4.01

are you talking about the software found here: https://www.sdcard.org/downloads/formatter/

or is 'SD formatter' a Sound Devices app?

When we do our approval tests, we run those tests across multiple cards of the same model to ensure consistency before we approve.

Should we assume right now that there are no cards other than the 32GB Sound Devices branded card that work reliably with the unit?

[rick.lang]

Running 4.01 under 24 fps and previous difficulty remains.  Running for an hour and the same pattern is there: MixPre Jam synced by Tentacle and loses Timecode sync before long, very close to a list frame by one hour and it might stay that way for many hours of accurate Timecode if you ignore the early problems.

Why is 23.976 so good and 24.00 flaky?  24 is an integer; it’s precise.  23.98 or 23.976 or 23.976024976024... that’s imprecise at some point of accuracy and yet that gives the right Timecode for Tentacle and MixPre.  What’s hard about 24? I’ll revert to drop frame 23.976 filming for the foreseeable future until Sound Devices solves this in release 4.02.

Edit
Five hours running and same result with 4.01 for 24 fps.  Lost less than a frame but still enough to require manual sync.  FAIL 24.

Sent from my iPhone using Tapatalk

[rick.lang]

Success in the stress test recording MixPre-6 II under 4.01 with 6 tracks armed and the left and right mix tracks at 192KHz with 32bit audio. Filled the branded SD32GB card with only a minute’s recording time left on the card.  So no worries about getting past the halfway mark of the card’s capacity and no worries practically filling the card! 

I’ll complete the formal results chart but need to rush this post as I’m updating the iPhone at the same time. 


Sent from my iPhone using Tapatalk

[Paul Isaacs]

Running 4.01 under 24 fps and previous difficulty remains.  Running for an hour and the same pattern is there: MixPre Jam synced by Tentacle and loses Timecode sync before long, very close to a list frame by one hour and it might stay that way for many hours of accurate Timecode if you ignore the early problems.

Why is 23.976 so good and 24.00 flaky?  24 is an integer; it’s precise.  23.98 or 23.976 or 23.976024976024... that’s imprecise at some point of accuracy and yet that gives the right Timecode for Tentacle and MixPre.  What’s hard about 24? I’ll revert to drop frame 23.976 filming for the foreseeable future until Sound Devices solves this in release 4.02.

Edit
Five hours running and same result with 4.01 for 24 fps.  Lost less than a frame but still enough to require manual sync.  FAIL 24.

Sent from my iPhone using Tapatalk

There is no reason I can think of why 24fps would present a problem. Please will you contact our tech support team to troubleshoot this in more detail. Easier than doing this over a forum. At this point, it is too early to cast whether this is a mixpre problem, tentacle problem or something else.

[Paul Isaacs]

are you talking about the software found here: https://www.sdcard.org/downloads/formatter/

Yes

Should we assume right now that there are no cards other than the 32GB Sound Devices branded card that work reliably with the unit?

No. Many other cards will work reliably, it's just that officially approving them is a long and intensive process.

[rick.lang]

There is no reason I can think of why 24fps would present a problem. Please will you contact our tech support team to troubleshoot this in more detail. Easier than doing this over a forum. At this point, it is too early to cast whether this is a mixpre problem, tentacle problem or something else.

Okay.  I’ve been in contact with Tech Support but action by the analyst was pending completion of our discussions here.  I’ve kept Support informed.


Sent from my iPhone using Tapatalk

[jerryfreak]

found my lucky 32gb sandisk pro micro, the only card that has never failed me. have resumed testing and will do updated tests on the other cards i still have with 4.01 FW

[rick.lang]

Tested 29.97 fps failed slightly at about the 15 minute and by one hour was almost a complete frame behind. I stopped the test.  It’s the same pattern.

Sadly 25 fps with 4.01 suddenly went from perfect to an even frame behind as I watched it! About 40 minutes into the recording.  Nothing gradual about it.  I killed the test.  Maybe it was a one-time thing and might go well for hours, but I wouldn’t trust Timecode unless you are using 23.976 fps. 

If you want or need to shoot one of the problematic frame rates, rather than jam the MixPre to a Tentacle, you just tell the MixPre to take its Timecode from a tethered Tentacle.  Costs you money if it means you need an extra Tentacle but you’ll have worry free syncing in post or anywhere else.

I’m going to let 23.976 fps run for hours and see if it remains good.  It should because it’s done longer than that before.


Sent from my iPhone using Tapatalk

[rick.lang]

I retested Timecode at 23.976 fps to see if that frame rate is still flawless.  Over nine hours and it’s still good Timecode, no drift.  I’ll call it a day.


Sent from my iPhone using Tapatalk