Topic: Tascam DR-70D 4-channel audio recorder (Part 6)

[soundsinteresting]

I do not think this unit has any "input transformer" to my knowledge, that was in the good old days with dynamic microphones.
Dr 70 has an equivalent input noise of -120dbU specified which should leave plenty of room for a god microphone.
The problem is that most condens microphones has a fairly high selfnoise that is amplified with the signal and that sounds horribly of course but that can only be better with a less noisy microphone.
if you have a less noisy microphone you will get closer to the limit of input amplifier noise of course but that is part of the game, but for most users i think the microphone is the big pronlem and typically these are low ouput mics so S/N is the limit here. so the best mic for DR70 is low noise high output which then may become a problem at high sound levels :-(.

[soundsinteresting]

I do not think this unit has any "input transformer" to my knowledge, that was in the good old days with dynamic microphones.
Dr 70 has an equivalent input noise of -120dbU specified which should leave plenty of room for a god microphone.
The problem is that most condens microphones has a fairly high selfnoise that is amplified with the signal and that sounds horribly of course but that can only be better with a less noisy microphone.
if you have a less noisy microphone you will get closer to the limit of input amplifier noise of course but that is part of the game, but for most users i think the microphone is the big pronlem and typically these are low ouput mics so S/N is the limit here. so the best mic for DR70 is low noise high output which then may become a problem at high sound levels :-(.

One further test.
I put the Tascam into LOW gain at MAX (+21dB)
I entered +21dB as gain with SD702.

Then i made a reference input (exciter) into the microphone input ch1 at ref level of 94dB (1kHz) at a sensitivity of 44mV.
Tascam now shows -5dB dBu at LCD
SD702 shows -24-26 dBFS (-4-6dBu?) at VU meter LED

I stored the 2 files and loaded into Audition.
File from Tascam shows a 1 kHz level at -5.4dB
File from SD702 shows a 1 kHz level at -22.8dB

Apparently SD702 have a preference of dBFS and Tascam dBu and the difference is probably supposed to be 20dB maybe because of an inaccurate gainsetting.
I would prefer that they both showed the same (-5.4dB) but how can i accomplish that?

[voltronic]

I do not think this unit has any "input transformer" to my knowledge, that was in the good old days with dynamic microphones.

I haven't ever cracked open my 70D, but Jim Williams, who designs mic preamps and other audio equipment has, and if he says they have input transformers I'm inclined to believe him.  Hopefully someone else here will be able to chime in on how common (or not) input transformers are on these types of recorders.

I know for a fact that my 2006-built FP24 (AKA SD MixPre) has input transformers, as does the currently-produced MixPre-D and the 633 (says so right in their manuals).  The SD 664, 688, and the entire 7-series recorders do not, however - they are "active balanced".

Here's some info explaining the two from Sound Devices:
http://www.sounddevices.com/tech-notes/output-balancing-topologies

The link to the Jensen article linked there was broken, so I found it here.  I realize this is from a leading transformer manufacturer so the fact that they call a transformer-balanced input the "best" way may be a bit biased but the explanation between the different implementations is presented in a nice, clear way.
http://www.jhbrandt.net/wp-content/uploads/2014/11/Interconnection_of_Balanced_and-Unbalanced-Equipment.pdf

[soundsinteresting]

I am not going to open my DR70 to find any input transformer so if you have information that say so, that is how it is, but i have not seen that mentioned anywhere. Tascam use more energy to state that input amps are 5532 :-). Both could be true.
Anyone here that has a clue why i have to fight dbFS/dBu implementations to have my test between Tascam/SD verified above?

[aaronji]

Decibels are ratios, so that they are always with respect to some reference value. dBu is with respect to 0.775 V and dBV to 1 V, so they are independent of the gear. For dBFS, the reference is the maximum signal that won't lead to digital clipping, so it varies depending on the recorder.  Not sure if that's helpful?

[soundsinteresting]

Decibels are ratios, so that they are always with respect to some reference value. dBu is with respect to 0.775 V and dBV to 1 V, so they are independent of the gear. For dBFS, the reference is the maximum signal that won't lead to digital clipping, so it varies depending on the recorder.  Not sure if that's helpful?

Thankyou aronji.
I am slowly moving to some fundamental understanding.
What really annoys me is that my documentation with stored files is different and dependant on which recorder used. For the same signal i have a difference recorded of 17.5 dB for these two recorders, it could look like the 12dB mentioned in article below due to the relation between dBu/dBFS. My brain says that stored files should be the same if same gainsettings are used with same mic. Anyone agrre on this?
i read this and maybe that gets near for some sort of understanding but is not suggesting a solution. do I really need to adjust all my files recorded for this dependency?
http://www.soundonsound.com/sos/may00/articles/digital.htm

Is it correct to say that tascam display could be showing DBu or dBFS, in this case the same where SD has a dBFS LED display where 0 dBu is the same as -20dBFS?

[2manyrocks]

There was a post somewhere back in the thread of a 70d vs 702 comp, but there wasn't much discussion about how it was done. 

One huge difference between the 702 vs the 70d is the 702 metering.  The 702 has a long and very visible metering scale which is invaluable when it comes to setting recording levels.

[yug du nord]

To my ears, the 70D does not have transformers.  Weight of unit suggests to me that there are no transformers.  Cost of unit suggests to me that there are no transformers.
So, IMO.....  there is nothing that suggests that there are transformers.
But JW is a tech guru.....  so if he claims that there are tranny's, it's hard to disagree......  but I'm trying.   

I'm just a yahoo with aging ears, aging muscles, and a tight wallet.   

[voltronic]

They could be very small?  I don't think it would be anything hefty made my Lundahl or Jensen like you find in pro gear.  I might post the question on GS, and either Jim or a couple other engineers may chime in.

[voltronic]

OK, I'm a dope.  What Jim actually said, speaking in reference to his modification:

Noise is dominated by the input transistor pair at higher gains so the opamp noise is washed out. At lower gains the opamp noise dominates and it's very quiet now. The sonics are also far better with the added slew rates and decreased THD. I also added tiny NOS Wima polyprop film bypass caps to the phantom blocking caps, another large improvement. The transients are now excellent when they were damped stock.

 

Also, speaking about the recorder in general:

They use a set of SOT-23 input transistors as used in many mic preamps. Those dominate the noise floor at the High+ settings. If you stick with the High gain and use high output condenser mics, it is silent. Several have had those opamps replaced, all heard the improvements.

[mfrench]

So, is this latest firmware stable enough to move to?
I've been running the OE stock firmware (1.00 0037) that came with mine since I got it, being a bit nervous to move away from something that didn't give me any grief out of the box.

thanks

[aaronji]

For the same signal i have a difference recorded of 17.5 dB for these two recorders, it could look like the 12dB mentioned in article below due to the relation between dBu/dBFS. My brain says that stored files should be the same if same gainsettings are used with same mic. Anyone agrre on this?

That's the thing I was getting at earlier:  there is no fixed relationship between dBu and dBFS, unlike dBu and dBV (dBu - 2.22 = dBV).  dBFS will vary from recorder to recorder.  The maximum mic input for the 702 is spec'ed at 10 dBu and for the 70D at 0 dBu, so, ostensibly, that should translate into a 10 dB difference in dBFS right there (I say ostensibly because the relationship between the maximum input and 0 dBFS isn't always as straightforward as you might expect).  There are very probably other differences in how the two recorders handle an incoming signal of a given voltage, as well.

Now that you have an idea about how the two compare, why not just adjust the gain settings prior to, or after, recording to account for the difference?

[jbell]

Updated mine and have not had an issue

So, is this latest firmware stable enough to move to?
I've been running the OE stock firmware (1.00 0037) that came with mine since I got it, being a bit nervous to move away from something that didn't give me any grief out of the box.

thanks

[soundsinteresting]

For the same signal i have a difference recorded of 17.5 dB for these two recorders, it could look like the 12dB mentioned in article below due to the relation between dBu/dBFS. My brain says that stored files should be the same if same gainsettings are used with same mic. Anyone agrre on this?

That's the thing I was getting at earlier:  there is no fixed relationship between dBu and dBFS, unlike dBu and dBV (dBu - 2.22 = dBV).  dBFS will vary from recorder to recorder.  The maximum mic input for the 702 is spec'ed at 10 dBu and for the 70D at 0 dBu, so, ostensibly, that should translate into a 10 dB difference in dBFS right there (I say ostensibly because the relationship between the maximum input and 0 dBFS isn't always as straightforward as you might expect).  There are very probably other differences in how the two recorders handle an incoming signal of a given voltage, as well.

Now that you have an idea about how the two compare, why not just adjust the gain settings prior to, or after, recording to account for the difference?

Thankyou Aroonji for taking your time on this.
The reason it is aproblem for me is of course that I use recorder for something it was not intended for.
Let me give you an example:
I want to verify selfnoise for the recorder and for my capsules. For the capsules I could use an amplifier and a voltmeter, but for the recorder it is abit more complicated as you have no access internal to analog amplifiers but have to rely on internal digital processing.

With Tascam  I have a display calibrated to dBu where max display is 0 dBu which is also the clipping level an therefore no over head is built in (as far as I know).
With Sd it is now clear that display is calibrated to dBFS where 0dBFS is dBU + 20dB or 0 dBu is equal to -20dBFS.

If you use recorder in a normal way you may set your amplifiers to show signal at let us say -16dB, that would be -16dBu with Tascam but with SD it would be -16dBFS which is not the same (+20dB) as Tascam.
If you record a file and on both and take the file into Audition or similar you would see that files are identical despite that they were recorded with different gain settings. This has now been verified by SD support that this is how Sd is calibrated.

My problem is that I use recorders to measure abolute (noise levels or signal) and get wrong results at verification because the dB's in audition are not the same as the the numbers stored in SD file (relative values).

I make a setup where a signal (or noise) is introduced to input amplifier at both items. I try to set same amplification 63dB (max) on Tascam and 63dB on SD. i would then expect the same output on both amplifiers before digitized.
If this signal was -63dbU Tascam would show max display and be near clipping but SD is showing -20dBFS (0dBu) which is ok sofar.
Problem for me is that the stored digitization is not calibrated to dBu but dBFS so when i compare the two files in audition, the file from SD is 20dB lower than Tascam file (same gain settings). i could of course compensate for this by adding 20dB amplification, but it is a bit worrying when small signals are used and whether you can trust the digital representation without degradation.
This has also ben verified by SD support to be correct, so that is how it is, but not very user friendly.

This confused me first because the selfnoise seemed to be far better with Sd than Tascam (could be), but when compensated for by 20db, suddenly Tascam seems to be very close to performance of SDxxx if measurement is considered reliable .

The first file attached is a screenshot from AUDITION showing Tascam 70d selfnoise (equivalent input noise) with input shuntet by 150 ohm resistance (resistor equ noise is 0,22uV/-130dbU) and max gain of 63 db which can be subtracted from measured value on output. Doing the same with SD would show a spectrum that is 20dB too low (absolute values) which should be compensated by 20db for comparing.

Second file attached is a screeshot showing spectrum from recorder noise measurement, red is Tascam and blue is SD702. Sd spectrum was compensated for as explained above.
If this measurement is considered reliable which I claim sofar, you may be surprised that Tascam has lower levels of noise at low frequency than SD from input amplifiers but overall is slightly worse but not that bad. (Input is 150 ohm resistor + 10uF) Interesting?

[morst]

CORRECT!

To my ears, the 70D does not have transformers.  Weight of unit suggests to me that there are no transformers.  Cost of unit suggests to me that there are no transformers.
So, IMO.....  there is nothing that suggests that there are transformers.
But JW is a tech guru.....  so if he claims that there are tranny's, it's hard to disagree......  but I'm trying.   

I'm just a yahoo with aging ears, aging muscles, and a tight wallet.