Announcing the SPDR two channel bag/field recorder

[Vincent R.]

1 hour ago, JFtaper said:

Multiple devices would still run off their own clocks once disconnected so im not even sure you could say this has 'real' timecode/word clock. even the best clocks out there will drift apart due to crystal variation as well as thermal factors.

less than a frame over a 24h period is sufficient. Jam/sync 1 or 2 times a (working) day and you are good. 

[JFtaper]

interesting. i know guys that do multitrack audio with multiple recorders and claim they get audible phasing in short order. a quote from a friend from another board

 

I generally allow +/- 20 ms max error, which would be +/- 960 samples at 48kHz. Some purists will say that they can hear 10ms smear but I don't find it much of a problem for rock & roll.

Sony, Tascam, Roland (Edirol)... none of these machines have tightly matched clocks between machines. After an hour, you can hear a "Flam" on almost any two recorders.
Use my spreadsheet if you would like to fiddle with some examples. You can download it if you prefer.
https://docs.google.com/spreadsheets/d/1pQGfYwPgBFFzcY5m6aRj-Zbu9HsRumLy-tJB1d8Eufg/edit#gid=583050244

 

 

[Vincent R.]

46 minutes ago, JFtaper said:

Sony, Tascam, Roland (Edirol)

There you have it, recorders with non reliable clocks/oscillators whatever. Don't believe Sony and Tascam even make TC accurate/capable devices anyways.

[Jamie Tongue]

59 minutes ago, JFtaper said:

interesting. i know guys that do multitrack audio with multiple recorders and claim they get audible phasing in short order. a quote from a friend from another board

 

I generally allow +/- 20 ms max error, which would be +/- 960 samples at 48kHz. Some purists will say that they can hear 10ms smear but I don't find it much of a problem for rock & roll.

Sony, Tascam, Roland (Edirol)... none of these machines have tightly matched clocks between machines. After an hour, you can hear a "Flam" on almost any two recorders.
Use my spreadsheet if you would like to fiddle with some examples. You can download it if you prefer.
https://docs.google.com/spreadsheets/d/1pQGfYwPgBFFzcY5m6aRj-Zbu9HsRumLy-tJB1d8Eufg/edit#gid=583050244

 

14 hours ago, JFtaper said:

I think the timecode is what makes it so expensive, and its something i would never need. Im not exactly sure (please educate me im not an ENG/broadcast guy),  but isnt 'Time Code Jam' just a reference marker? Multiple devices would still run off their own clocks once disconnected so im not even sure you could say this has 'real' timecode/word clock. even the best clocks out there will drift apart due to crystal variation as well as thermal factors. maybe not a factor with short take ENG stuff, but i want to record several hours of audio at a time, which requires post-processing to sync multiple sources

 

Yes, timecode is just a metadata stamp at the start of each file. Synchronising timecode clocks between devices is not the same as synchronising the 'recording clock' between devices (that's what genlock is for). Timecode and genlock are different things. Vincent was referencing that within the world of production sound for picture that this website is mostly aimed at, timecode clocks which drift from each other less than 1 frame per working day is sufficient for ensuring that the metadata stamps at the start of each file are in sync with each other across devices. Yes, once recording begins, the devices all record based on their own recording clocks, but for keeping the sound in sync with the image, anything less than 1 frame of drift over the course of a take is fine. For multitrack audio across multiple recorders, you need much more accuracy, as you say. Down to a single sample, really.

 

The SPDR never claimed to have genlock capability. It claims to have timecode capability, which it does have. But, no, timecode will not help you for your purposes. I used to be dabble in TS.com hobbyism myself, some years ago, so I fully understand what your purposes are 😉

[JFtaper]

yes i try not to speak out of turn, ive never used timecode only wordclock in with multiple audio devices. 

[IronFilm]

On 4/12/2019 at 10:01 AM, Vincent R. said:

There you have it, recorders with non reliable clocks/oscillators whatever. Don't believe Sony and Tascam even make TC accurate/capable devices anyways.

Tascam HS-P82

 

https://tascam.com/us/product/hs-p82/top

 

[Vincent R.]

1 hour ago, IronFilm said:

Tascam HS-P82

 

https://tascam.com/us/product/hs-p82/top

 

Ha forgot about that one. Have no idea about the TC reliability by the way.

 

[IronFilm]

I wonder if Sony ever made a recorder with timecode, quite possibly one slipped by us. There is for instance that pro mixer (the Sony DMX-P01) which even has AES, so I bet if Sony had just carried that on into the future a bit longer then I'm sure Sony would have made one with timecode. 

[Vincent R.]

8 hours ago, IronFilm said:

I wonder if Sony ever made a recorder with timecode, quite possibly one slipped by us. There is for instance that pro mixer (the Sony DMX-P01) which even has AES, so I bet if Sony had just carried that on into the future a bit longer then I'm sure Sony would have made one with timecode. 

Yeah I remember seeing that one for the first time. IBC 2003/4 or something as a prototype. Never saw it in the wild though. Also never saw one with recording functionality/TC. 

[Werner Althaus]

On ‎4‎/‎11‎/‎2019 at 4:11 PM, JFtaper said:

interesting. i know guys that do multitrack audio with multiple recorders and claim they get audible phasing in short order. a quote from a friend from another board

 

I generally allow +/- 20 ms max error, which would be +/- 960 samples at 48kHz. Some purists will say that they can hear 10ms smear but I don't find it much of a problem for rock & roll.

Sony, Tascam, Roland (Edirol)... none of these machines have tightly matched clocks between machines. After an hour, you can hear a "Flam" on almost any two recorders.
Use my spreadsheet if you would like to fiddle with some examples. You can download it if you prefer.
https://docs.google.com/spreadsheets/d/1pQGfYwPgBFFzcY5m6aRj-Zbu9HsRumLy-tJB1d8Eufg/edit#gid=583050244

 

 

I know this isn't really relevant to regular location sound work but I'd never record phase correlated material across multiple recorders, even when they are "locked", "synched" or whatever. If you have ever listened to the " Auto-align" plug in do it's sample offset thing to align phase you'll see that even one or two samples will massively mess up phase between 2 correlated microphones , that's 1/24thousands of a second, or 0.04 ms, nothing purist about hearing that. I doubt that any of the recorders mentioned here can be locked with 100% sample accuracy

[JFtaper]

fortunately there is no shortage of affordable multitrack recorders these days

[Gaf]

On 4/10/2019 at 4:27 AM, JFtaper said:

Thats because the SPDR is not a bit-accurate device, it resamples all digital inputs (among other digital processing). It does not write raw data like bit-accurate digital recorders

 

I noticed this behavior when feeding it AES from my benchmark ADC, which has very accurate meters with peak hold. With the benchmark peaking at -1 dBFS the SPDR was clipping past FSD so i knew something was wrong right away,

 

upon further investigation (comparing signal from same ADC captured by a known bit-accurate device with that recorded by SPDR), it is distinctly different.

 

The fact that it resamples all inputs should be obvious though because the SPDR will accept any input from 44.1 to 192 kHz and reclock it all to either 48 K or 96K. When matching sample rates from source device it still reclocks and processes. Its obviously more designed as a catch-all backup if you have an AES signal available, but if youre looking to capture accurate data from a dedicated ADC its the wrong tool for the job. There are prosumer larger handhelds like sony and tascam that can do this (the midgrade ones with XLRS and phnatom that start at $400 and up usually offer this). The extra gain and other digital processing involved in the SPDR adds insult to injury nfortuantely

 

I know lecrosonics prides itself on accurate clocks but its a pretty basic feature on every digital recorder out there to slave lock to a source clock feeding the device. Processing AES or S/PDIF to I2S and buffering it to disk is the simplest task in the world so im wondering why they chose to intentionally process all digital input. Perhaps to make it more universal so it would work reliably under all conditions. But dumbing it down like that is a mentality more commonly seen with consumer gear.

hi, sorry im late just landed! 
Can you reconfirm this? ive searched far and wide, contacted Lectrosonics and asked colleagues. No one can confirm that SPDR resamples AES input when it matches the incoming sample rate.
Not only that, everyone is surprised that there can be 1 whole dB difference just from that alleged resampling!! 

can this experiment be repeated and tested? Any settings on the SPDR or firmware updates that change this? I would really like to get my hands on a proper 2 channel AES recorder for my SX-M2D2 but i this is a major issue. 
Also if anyone has suggestions, feel free. 
 

Cheers

[JFtaper]

that was the case when i posted that. i doubt they corrected it but you never know. 

 

if they insist that it does not resample (not ignoring the question, but actually saying "it passes bit-accurate data without resampling"), then buy one. Its fairly trivial to compare recorded vs original file in a wave editor. inverting the recorded file over the source file should result in digital silence

[EinF]

On 12/30/2018 at 6:57 AM, Boomboom said:

Yep, memory card door out the very first time I opened it… : /  Lectro's usually better than that at design ...!

 

Hey all, as it's been a few years I checked in with Lectro HQ about any updates:

 

*Card Door: (update) Swapped out for a similar but more robust redesign. Repairs dept. also said they'd be glad to send out spares of the updated door.

 

*Limiter: Hasn't changed since 2018, except in the display: (via 2022 firmware update: )

"The SPDR now conforms to industry standards for digital audio metering: the meter always reads 0 at +0 dBFS. The analog input limiter can now be seen to engage at -5 dBFS, reserving 5 dB for limiter overhead. This behavior of the analog input subsystem has not changed, but now the meter always shows the levels relative to +0 dBFS. This eliminates the possibility for confusion caused by the meter reading differently depending on the input type (digital or analog) and whether the SPDR is making a recording or playing one back"

 

* Low Cut Filter: No shift, 35Hz is the closest the SPDR provides to turning filter off (cutting 20-35Hz range)

 

*Timecode format: No change