@ SecretTape - I’m guessing it’s splitting hairs but in your experience do you feel it makes much of a difference hitting the T805 earliest in the capture process while the signal is still analog versus re-printing digitized audio through the unit at mixdown? Many thanks.

There will be a difference between recording straight through the T805 and sending it from converters through the T805 post-digitization. In analog tape and the T805, high frequency content in the audio signal stacks with the ultrasonic bias signal, which creates tape compression. For this reason, high frequency content greater than Nyquist will have an effect on the overall magnetization of the audio signal, even though it’s normally inaudible. Typically, there isn’t much frequency content above 20 kHz, especially when recorded with microphones, because many begin to roll off high frequencies lower than 20 kHz. That said, some sources with a lot of high frequency content will behave differently and have less overall tape compression if ran through the T805 post-digitization. Capturing more high frequency content gives a more of a tape sound for this reason. I’ve been recording with a sample rate of at least 96 kHz, and running digitized tracks through the T805, and I think it sounds great. Eventually, I would like to run a test of pre-digital and post-digital to see how large the difference is.

Intuitively I would have thought the opposite: Is this due to aliasing distortion above Nyquist? And why you suggest that you’re working at 96kHz (rather than a lower rate)?

Many thanks for the response. I’ve read it several times and am trying to digest.

I still bother to print tracks through tape at mixdown (and even recently when mastering an album) to tame and soften all that harsh HF stuff that builds up subtly, and am interested in the T805 to simplify that whole process a bit hopefully.

Yeah, it’s very counterintuitive at first. To really get it, you need to understand how a bias signal interacts with a magnetic medium.

A bias signal increases the level of all frequencies lower than itself—but not uniformly. The closer a frequency is to the bias signal, the less of a level increase it gets. For example, with a 100 kHz bias signal, a 20 Hz tone sees a greater level increase than a 20 kHz tone, simply because 20 kHz is much closer to the bias. That’s why record HF EQ is needed in tape machines and the T805—it compensates for that tapering.

Things get more interesting with complex signals. Take a snare drum: it has low frequencies from the body and high frequencies from the snares. Those high frequencies also act as a kind of “bias” signal, summing with the 100 kHz and further boosting the lows. The magnet doesn’t “know” whether the high frequency energy is intentional bias or part of the audio—it just responds to flux. So high frequencies in the audio cause a temporary level increases in lower frequencies.

That’s why there’s no such thing as a flat frequency response in magnetic recording. The spectral content of the signal itself bends the frequency response dynamically. This is what we often call “tape compression”—but technically, it’s not compression of highs. It’s expansion of lows, driven by the highs. Using compressor terminology, it’s like the low end is sidechained to the high end. That’s the magic of tape (and this new magnetic process).

Taking it back to original question: if you’re recording at 48 kHz, everything above 24 kHz is gone—meaning that any high frequency content that could have acted as an additional bias signal has been shaved off. But at 96 kHz or higher, you preserve more of that energy, allowing it to interact with the magnetic process more fully—resulting in stronger “tape compression.”

It might take a few days to wrap your head around this (it certainly did for me), but once it clicks, it’s pretty mind blowing. Frequencies are always affecting the ones below them. 200 Hz boosts 20 Hz. 2 kHz boosts 200 Hz and 20 Hz. And in between is a whole continuum doing the same. That’s why tape and the T805 smooth out transients and sibilance: it’s all part of this dynamic, nonlinear frequency interaction.

More details to come in the white paper in May

Thanks for the thorough response. Like many I’m certainly excited to read all about it in May.

I hope your hard work and risk is rewarding. 😎

Mindblowing and really well explained. I had no idea until now how tape compression works.

I'm guessing that it can't be done using a ton of bands of multiband compression (please correct my assumption if wrong).

Thanks, I'm glad it made sense. After I recreated the AC bias process in a new form, I observed this behavior, but I didn't have an explanation for it until I read the HX-PRO paper, and my mind was blown.

I tried messing with a multi-band compressor to emulate this behavior a while ago. What I tried sounded nice, but it didn't sound like tape. You might be able to get a nice high frequency softening effect this way, but it wouldn't be occurring as a natural side effect of nonlinear hysteresis in a magnetic medium so it'd be missing all the magnetic nonlinearities, which are a big part of the overall sound.

Also, any multi-band processing involves crossover filters and their inherent side effects, which are not present when recording to tape.

Something that is might be super cool about this: getting the sound/feel of tape without the record-to-playback delay. For example, when tracking vocals the talent (and engineer of course) can hear exactly what's going on in real time.

Fascinating, just read “Dynamic Bias Control with HX professional” by Jensen and Pramanik: seems they essentially perfected the “bias problem” right near the end of the road for tape. And if you’ve managed to do what I started to think about in that article, you’ve tapped into a very legitimate “tape sound” effect! Very clever and patent worthy indeed! Nice work 😁

Tease!


So, other than as a storage medium, what role does the tape itself play? Obviously I would imagine that the signal coming off the play(back) head would get another round of the signal bias which we hear a lot in older recordings where they needed to "bounce" some tracks down to make room for more tracks. Perhaps it is more complicated than just that. And, I`m probably really not explaining my question very well here, but though I (somewhat) understand what you are saying, it's still above my pay grade. It`s still a kind of magic to me what tape machines are doing. But, your explanation helped clear up some basic understanding for me. Thanks!

Yeah, there is a lot of misunderstanding about how tape works, and I think a big part of that is because the essential information is scattered. Most of my time was spent tracking down the relevant research papers. John McKnight did a great job giving a starting point into the research though.

The white paper covers magnetic tape recording to provide a working understanding of the physical magnetic process in tape recording, because it's not possible to understand how the new process recreates tape without having a basic understanding of magnetic recording itself.

As for the reproduce (playback) side: the tape is the source of the nonlinear hysteresis. It is the medium that is biased and this medium creates this tape compression effect (along with saturation). The reproduce process is actually quite linear. The voltage induced in the reproduce head is the signal stored in the tape, but it’s induced at a low level in a high permeability reproduce head, with distortion levels typically below the noise floor. The bias signal is reproduced and notched out with a bias trap (which occurs in the T805 as well) in the reproduce stage. It doesn't affect the reproduce process, except for creating additional noise in the electronics connected to the reproduce head. The magnetic aspect of the sound of tape is more or less contained to the recording stage.