Or how I learned to start worrying and hate the tape machine.
I’ll go through the process of adding another 8 tracks (ADAT) to an audio interface so we can record more tracks simultaneously. On the way I’ll cover some other points of interest like jitter, word clocks, and reference levels.
One interesting thing I’d like to point out upfront: Different converters/preamps can introduce different latencies even when in “sync”. More about this below.
Here is the setup. The album I’m mixing was recorded on 16 tracks of tape. The recording engineer is a real analog guy and can’t do the transfer to digital.
At the time I only had 10 analog inputs on my interface (RME Fireface 800). My first thought was to A/D convert in batches. The first batch would have tracks 1-10 the second 11-16. Can you tell I’ve been working in digital? Then I remembered something about tape from back when I owned a Tascam 424. Tape does not play back at a consistent speed. You can’t hear it pitch up and down like vibrato. It doesn’t end the song a quarter note flat. But if you try and do 2 transfers, as I suggest, good luck lining them up. This could be an even bigger problem if you have mics that are supposed to be in phase but are now in two different transfers. I confirmed this with the recording engineer who said he’d tried it. Then he said don’t try it.
Plan B. Sync my Fireface up with another analog to digital converter adding tracks. Sounds expensive for a one time use. My Fireface was around $1800. Another quality box would be in the same price range I thought. Maybe. After reading the review in Sound on Sound as well as some positive remarks on Gearslutz I bought a Behringer ADA8000. I know Behringer is not perceived as high fidelity. That doesn’t guarantee this unit sounds bad. Best of all it was around $300 which is super cheap for 8 channels of preamp and converter.
Can I hear a difference? Yes. The Fireface sounds warmer to me. Or if you want to spin it in favor or Behringer… the ADA8000 sounds cleaner. Is it better or worse then my Fireface? I can’t tell. Imo different isn’t necessarily better or worse. Take a listen and tell me what you think.
You can listen on this player or download by clicking the downward arrow on the right end of the player. If you really want to nerd out load them in your DAW so you can loop a small section and listen to each carefully.
Here is how I organize the rest of this process. Somewhere at the end of this post I will go through the specifics of this particular gear.
- Choose your master device.
- Make data connections.
- Make sync/clock connections.
- Designate devices as master or slave via software or hardware.
Choose Your Master Device: Digital audio systems are kinda like Highlander. There can be only one! (master device) Which one should you pick? Briefly, the one with the better clock. RME has an excellent reputation for building quality products. So I’m just going to set my Fireface as the master. But what’s a clock and why does it matter?
When a device takes samples (recording) or plays them back it needs to do it at very regular time intervals. Hopefully everyone has seen that comparison between an analog waveform and a digital waveform. If you haven’t let me show you why I don’t have a career as an illustrator.
Figure 1: This is a simple analog waveform. It is continuous. Every point in time and amplitude is defined.
Figure 2: Digital audio samples this waveform at regular time intervals. Blue vertical lines show where samples are taken.
Figure 3: Now we take those sample values and reconstruct the waveform. The original wave is shown in black. The blue “steps” are the digital reconstruction. Note that the reconstruction has the same basic shape as the original analog wave.
Figure 4: But what if we don’t sample at regular time intervals?
Figure 5: Then we reconstruct the waveform with these samples taken at irregular time intervals. Jitter! Some intervals are too close together. Amplitude does not change much over that time. This produces the flat areas of distortion. When time intervals are too far apart you get big vertical drops in the waveform because a large amplitude change occurs over that time. I didn’t spend tons of time making figure 4 map exactly over to figure 5 but you get the idea.
Your clock determines the time intervals. If your clock is inconsistent you get the jitters. Er, I mean, jitter is what happens when your clock is inconsistent. Can you hear jitter? You sure can.
Which brings us back to picking a master device. It’s called the master because we will use it’s clock as a master clock. All other devices will be “slaved” to that clock. Again, all things the same, use the device with a better clock as your master.
Make Data Connections: The only way to get 8 or more tracks of digital audio into my Fireface is through ADAT. (It also has one input for S/PDIF or AES/EBU formats. But these are 2 channel formats and I need 8 channels.) Technically it’s called ADAT optical since the connections are optical Toslink cables.
My rule of thumb is not to buy super cheap cables. Some people will argue that digital is digital and super cheap cables will work fine. But there are other considerations. I’ve read of Toslink cables with ends that won’t fit or break off. I’m not convinced super cheap fiber optics can’t effect data transmission. I also would not buy cables any longer then I needed. All signals degrade over distance.
Make sure you take the protective covers off the ends of the Toslink cable and connect the output of the slave device to the input of the master device. Since these are optical connections be careful not to muss up the ends of the cable.
Make Sync/Clock Connections: This is the connection that passes the clock signal from master to slave device. There are 2 ways for me to do this for ADAT. I can run it via an ADAT connection (Toslink optical cable) or a BNC word clock connection. BNC is a common coaxial cable connection.
I sniffed around to see if either was preferable. As I could not find a strong case for either I choose Toslink.
So just connect the ADAT output of your master device to the ADAT input of the slave device.
Designate Devices as Master or Slave: Remember you can only have on master device. Everything else must to set to slave. The ADA8000 has a switch on the pack panel. The Fireface makes the switch via software. Check the manual for your device if you are unsure.
Here are the specifics for the Fireface and ADA8000.
ADAT optical is always 24 bit and mostly 48k. You can use 96k if you really want but your bandwidth stays the same. So you only have 4 tracks at 96k, not 8.
- Make sure it is in fact set to master. This is found in the Fireface settings software. Clock source should be set to internal. That’s RME’s term for Master.
- The Fireface constantly looks for input clock signals. If it finds one it will automatically sync with it. If you use the Fireface as master make sure not to send any word clock signals to it.
- Make sure ADAT is set up as an input. The Fireface allows you to limit the bandwidth on the Firewire bus if you want to. For example you can limit input to just 8 analog inputs. I set bandwidth to all channels to be sure adat is received. I’ve never had the need to limit traffic on my Firewire bus. Again this is in the Fireface settings software.
- My DAW seems to drive my interface’s sample rate. Changing my DAW’s sample rate will change my Fireface’s control panel setting to the new sample rate. This may be a foolproof design but I would check to make sure you are at 24bit/48k.
The Fireface has a nice set of LEDs that indicate what is connected and if it is in sync.
A steady green light as seen here says ADAT 1 is connected and in sync. A blinking light would indicate signal with no sync.
- The ADA8000 does not use a software control panel. Instead you use a switch on the back of the unit. There are 4 postions: 2 master and 2 slave. In the slave postion you need to set it to ADAT as seen here.
- There are 2 indicator LEDs on the front of this unit. You want it to look like this. Only the green locked light should be on.
We are sunked!
It would be a great idea to do a test now. This truly great idea was never followed by yours truely and causes a delightful work flow stoppage during the tape transfer. I suggest you make sure all your inputs properly route to your DAW tracks and you can record to them.
ADAT 1 did not get routed where I expected it to. 10 analog inputs and 2 S/PDIF inputs put the first ADAT at input 13. I hadn’t considered the S/PDIF inputs. My DAW’s input weren’t labeled so it appearted I was missing 2 tracks.
Even when in sync the two devices were out of phase! This is probably the most interesting thing I discovered. What do I mean by this? think about adding a constant of 31 samples to one converter.
I was told this was nothing to sweat unless it puts mics out of phase. For example: You have two in phase mics on a snare. One is converted via the Fireface. The other the ADA8000. Now they are out of phase.
I disagree. When I time adjusted the 8 tracks that came in 31 samples late it was noticeable. Take a listen.
Here are the tracks are they arrived.
Here they are after I slid the Behringer tracks back 31 samples.
How did I tested this? I split a mono cable and ran these into the Fireface and ADA8000.
I tapped my finger on the end of the cable producing some pretty violent transients. Then I zoomed way in to the smallest units possible, samples. I made sure that both cable paths were the same. Even if they weren’t I don’t think it makes a difference. If memory serves right electricity (err, wave propagation) travels at nearly the speed of light.
Getting levels from the tape machine: After routing ADAT (from Behringer ADA8000) into my DAW we were ready to transfer. In other words, record 16 tracks of analog into my DAW. Which brings us to delightful tech snag #23.b. Levels.
So we all know thou shall not clip digital. Great. But thou should also not record super hot at 24 bits. 24 bits has so much range there is no need to track hot. You will actually sound better with peaks down at -6db or -10db. This may be news to you. If it is and you want to learn more I suggest reading Bob Katz’s book Mastering Audio. Or you can go into a pro studio and see where they track at 24 bits. I shoot for average (RMS) levels of -20db.
Can I get a -20db? (whut whut). No I can’t. I can get a flagrantly violating constant clip.
The Fireface has 3 reference levels (+4dBu, -10dBV, and High Gain) you can set in software. I like to think of this as a place where you tell your hardware what level of signal to expect.
Let’s see if I can break these down without talking about logrithmic ratios and reference voltages.
+4dBu is a hotter signal used for most pro gear. This gear is balanced. Think XLR or TRS connections. Balanced cables employ a pretty cool trick. You can read about it here: http://www.mediacollege.com/audio/balanced/cable-balanced.html If you don’t like reading here is the breakdown. Balanced cables have less noise. Balanced cables have a higher signal (about 6db greater then unbalanced). Balanced cables are what professionals use. Ok, there are some notable exceptions. Instrument cables, like your guitar cable, are usually unbalanced.
-10dBV is a weaker signal use for most consumer gear. Think unbalanced like RCA connections.
High gain is just in case you get a really hot signal and need to put it in the proper range.
Nerd think: I have a hard time believing balanced cables would always be 6db or double the signal strength of unbalanced. If you read the above link you will see this is from summing 2 signals. But what if the 2 signals are from a passive device? You can’t get something from nothing. It seems to me a passive device such as an sm57 mic would have the same level balanced or unbalanced. However, the balanced signal should still have less noise.
We finally got the levels cooled off and in a good place with an unholy union of RCA (-10dBV) output from the tape machine and choosing the “high gain” input reference setting on my Fireface. I just wish we hadn’t used Hosa cable. Note we are using the lowest level connection type and setting the Fireface to expect the hottest signal it can. Damn that tape machine had a hot output. Makes me wonder if it was out of spec or needed some type of adjustment.
Levels good, connections routed, roll tape! Roll tape #2. Done. It’s late. Whiskey, cigarettes, no coffee, I’m tired.
The next phase is what I call the fish net phase. You know when you download a great mess of free illegal software overnight? Then the next day you go and see what you actually got. Astute and correct readers might say ,”Why didn’t you listen to it there and make sure it sounded good?” You weren’t there man! Also I did do some listening. But what I’ve found is you can still miss things. One time I got 30 tracks for mixing and thought they sound good. But after a few listens I realized there was buzz in every track. The kind of buzz you get from running audio over unbalanced lines in a noisy environment. Also the vocals were distorted and the room had major resonance issues. But that took time to hear.
The nest day I hauled the net onto shore and took a good listen. Sounds good. F yes. Let’s mix ‘er!