Magnetic Tape Harmonics / Modulation

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Bo Deadly

Well-known member
Joined
Dec 22, 2015
Messages
3,266
Location
New Jersey, USA
I've been testing my ReVox B77 recorder and I think I have at least one channel setup properly. I'm using new LPR90. I noticed something weird. Consider the following frequency response plots:

tape.png


These are 1kHz, 10kHz and 1kHz+10kHz tones respectively, all -3dBu output. As you can see, the 1kHz and 10kHz are pretty much what you might expect. There's some side band splatter in the 1kHz that is odd but I don't think it's incorrect. You can see a little 10kHz bleed as well but that's just because I didn't use a fresh part of the tape. The 10kHz plot looks good. Great actually. I had no idea magnetic tape recording was that good.

But then you can see that if it's a mix of both tones, all hell breaks loose. There's wild harmonics that appear to be from the modulation of the 10kHz on top of the 1Khz.

Is this normal? Is this part of the "tape sound"?

UPDATE:

As discussed, 0dBu out is too hot for 10kHz so I have repeated the 1kHz+10kHz test at 0dBu, -10dBu and -20dBu which are shown here respectively:

TapeIMD.png


Note that I did not set repro level on this machine (I don't have a ref tape). So the trimmers are just at 12'oclock. I'm still fiddling a lot with this machine (getting major scratchy noise from relay on right channel).
 
What speed is the tape running at? I would expect it to be terrible at 7.5ips and much less terrible at 30ips.
 
PRR said:
-20VU at 10KHz is a LOT for tape. Few real sources will do that.
The level in these plots is actually -3dBu output. You just can't see the peak very well in the plot. But that's output. The VU meters on the B77 are on output only. I know that with 0dBu in and with the input level control at ~4, I get 0dBu out.

As fo r the level actually hitting the tape, that is way less just because of the EQ alone. So I'm not sure what you mean exactly.

But I will rerun the test at -20dBu output. I suppose it makes sense that I'm just cropping the HF riding on top of the LF.

Does anyone know if the sideband splatter around the 1kHz tone is normal?

The machine sounds very good. I'm just trying to understand the noise / distortion performance of this. I've never owned a proper reel machine before so I'm not sure what's normal.

The machine is normal speed 7.5 IPS.
 
Those plots show the intermod products at about - 40. Thats about 1% distortion. Tape is normally characterised as having distortion of 3% at peak operating level. As others have said, 10 kHz at full level at 7.5 ips is unlikely. You need to repeat the test 20 dB down. And buy a high speed motor for your Revox.
 
+1 to what others have said...  too damn hot for tape  :eek:... at HF.

You can trick yourself with single tone testing but that's why they invented two tone IMD (SMPTE)...  IMD testing will reveal nonlinearities.

It is instructive that they used 60Hz and 7kHz because even 7kHz was very HF for early tape machines (early everything). Not only did they consider 7kHz HF, they mixed it in 12 dB lower than the 60Hz.

==

Back in the 70s I modified my old SMPTE IMD analyzer to run 19KHz and 20kHz (1:1), but clearly not for tape machines.

JR
 
When you buy an MRL test tape for 7.5ips the high frequency tones are printed 10dB lower than the reference level for the reason you have run into. They are printed at reference level for 15ips and 30ips.

You should be aware of the actual flux level nW/M on the tape. dBu doesn't really tell you what you need to know.
 
In the days of analog multitrack tape,  sources with high energy at high frequencies ,such as overheads on cymbals you typically dialled the levels to tape way back from normal .
A guitar with its own distortions will probably sound fine recorded with as much level on tape as you can get , where sources with lots of HF  will sound bad recorded this way .
I guess the distortion, or more like the annoyance factor of the harmonics it produces is not linear across the spectrum , so you need to premtively correct for this by making sure  HF's dont saturate the tape .

Its true also that these effects are less prounounced at higher tape speeds , the more ips/cm of tape fly past the heads , the more hf energy you can print to tape without it becoming objectionable sonically .

In the old days you used a test tape ,like the one from MRL to ensure basic compatibillity from one studio to the next , now in modern times if say you want to use a two track machine as a kind of serial effect or bounce the sound through it to digital its unlikely your ever going to need to play the tape on another machine ,
I dont have a proper test tape to set up my machine , so I apply noise with equal energy across the audio spectrum to the inputs making sure levels to tape are well down -20db range , then slowly tweak both record and playback amp eq controls until the widest flatest response was obtained , bias adjustment  also changed the HF peak a bit ,  I was able to eq things out flat within a fraction of a db to 19khz on my Studer , If I accepted a slight tip up at the top end I was able to get usable bandwidth beyond 20 khz , 15ips
Test tapes come with all kinds of issues unless there very carefully handled , I found the frequency generator/ realtime output level alignment method you can probably get a better sound out of your machine with the particular grade of tape you use , of course  you do loose to some small degree cross compatibillity ,but  instead of worrying about some notional  levels to tape ,your setting the record and replay electronics to give the flattest overall response input to output (off tape) .

Id very much like to try out modern  FFT/Spectrum software to look at the how the harmonic content of tape changes with frequency and level ,instead of worrying  about millimaxwells and nanowebbers you could just tweak things  for widest bandwidth and  most favourable harmonic spread  .

















 
The only way to align a tape recorder is to start with a reference tape. Period. Willi Studer will be turning in his grave at your heresy of owning a Studer and not owning a repro tape!
 
The pre emphasis and de emphasis EQ for tape is roughly equivalent to the RIAA EQ for phonographs.  The high end being put on tape is up roughly 20dB above reference level at 20K.
 
In reply to Radardoug ,

If you need standardisation yes of course you do I mentioned that above , if you dont have archive material ,never need to play back the tapes on another machine ,or are just throughputting to daw , adhereing precisly to the eq curve or reference level doesnt matter much at all , and that allows you to squeeze out a little more performance with your particular combination of tape formulation and record/replay electronics .

15khz +1/-2db was the original spec on the  studer ,  I'm able to tune it to 19khz +0/-0.5db by this method , I had of course some  basic 1khz  reference tones on tape at a specified level to set my  playback gain  in the right ball park  ,but apart from that your  fine tuning the record and replay eq for best performance with the actual tape your recording on while your monitoring/measuring  the output of the reproduce head .

A 10khz signal off an old ,if even well cared for test tape is never going to be the most reliable reference , I stand by what I said if you have no need for strict level and eq  interopperabillity with other machines, tuning the record and replay electronics to the actual tape formulation your going to use gives better results . Even more so nowadays  we can easily measure  the distortion in real time off the tape and tweak  record/play chain for best performance ,

Reference levels for tape came about in a very different time ,since then formulations have changed and old curves for frequency vs level vs distortion may not apply or be meaningless , adhereing to the standard tape eq charachteristics  curve if it serves no real purpose  means  your probably throwing away performance with a modern tape formulation .


 
So what do you do, tweak repro and record until you are happy? Thus ending up with a machine all over the place? As you say fine for you, but maybe not good general advice. Bear in mind a lot of people who read your post may not be technically qualified to play with the machine as you do. So it better to give them the right advice, which is to use a repro tape to set up repro.
 
You guys aren’t that far apart. Use a test tape to set up the machine if you are a normal person.

If you want to tweak to tape formulation or individual tape for ultimate performance print tones when you are done and all is well.
 
> As fo r the level actually hitting the tape

The machine is a black box. I never had to care what the level on the tape was. I know that at economical speeds in the top of the audio band you CAN'T be anywhere NEAR 0VU on the meters (maybe you thought "VU" was a typo?).

In fact the record EQ is flat to about two octaves from the very top. With no record EQ, response would slump in the top. Record EQ starts with a rise and often adds a resonance in the top octave. At 10KHz and 7.5ips you are very much IN this boost.

And yes, professional alignment MUST start with a calibrated reference tape to set the Play. Trimming both Record and Play against each other gives a machine which may play its own tapes flat but will be *wrong* for any tape brought from another machine.
 
That’s called a flux loop. That’s how MRL sets up their machines. I’d start with a test tape.
 
squarewave said:
As fo r the level actually hitting the tape, that is way less just because of the EQ alone. So I'm not sure what you mean exactly.
the question here is what's the magnetic level, as in reference to nominal magnetic level (185nWb/m). Typically, hitting tape with nominal level will produce about 1-3% THD at 1kHz. Due to preemphasis (about 20dB at 10kHz) THD at nominal level at 10kHz produces enormous THD, that is masked by the limits of audibility (who actually hears 20kHz?).
However, in the presence of two tones, there is also enormous intermodulation distortion; that is what you see, with products at 10kHz +/-n.1k
 
I wasnt suggesting an alignment should be attempted without at least some known good reference signal off tape , there is a lot of what ifs of course , mechanically especially  , I spent a lot of time ensuring the physical running of the tape across the heads was just right before I ever  touched the controls , the machine itself came with about 50 radio studio program reels with tones printed at the head of the tape , the machine is also full track mono ,which simplifies matters a lot .

The main thing I wanted to point out is with modern software based audio test kit you can fine tune your own machine well beyond whats possible with a vu meter and pre recorded test tones ,yeah your using the reduction method ,so its tedious to and fro ,minute  adjustments here and there , you see the cumulative effect of all the adjustments both physical and electronic on the fft/spectrum and re-tweak for best a couple of times ,with care and providing mechanicaly and electrically your machine is good you can probably exceed the basic spec of the machine by a good margin , its something thats well worth doing even if you have the MRL tape .

In the old days in my studio we would have had to hire in someone with an A.P or Neutrik analyser to do that , so it only happened after a major service interval  where the heads were lapped and transport was done , tweaking a 2" 24 track is like tuning a piano in many ways .

Conciliatory notes greatly appreciated Paul(Gold) ,your a gentleman and a diplomat ,Sir

Happy St Patricks Day all ,



 
Tubetec said:
The main thing I wanted to point out is with modern software based audio test kit you can fine tune your own machine well beyond whats possible with a vu meter and pre recorded test tones ,yeah your using the reduction method ,so its tedious to and fro

I must object to this.  Even if you have all the time in the world you can’t set up a deck properly without known tones on tape (ie. MRL) or flux loop.

It’s even more important on stereo or multitrack decks for obvious reasons of desiring each channel to sound the same.  With out a MRL you will be hitting input harder on some channels where you’ve “guessed” too low on you repro setting,  And you’ll be under on channels where you’ve guessed repro too high, and this will result in more noise from the medium and change the point of saturation etc.

It’s like if you found a nice old wristwatch and just wound it up and set the time without a reference.  It’ll offer a certain amount of valuable information, but will be wrong and you won’t be at your meetings with other people on time...

Even  just borrow a MRL to align the deck’s playback and then do a record alignment and make a “house” tones reel.

Also, earlier in the thread people mention tape not handling high frequencies at high VU.  This is not true.  But I think it is being confused by the fact that high frequency TRANSIENTS barely move a VU meter.  A sine wave at 20khz at operating level is fine. But if you see a HF  transient moving the needle that much, you will definitely have distortion!
This is a limitation of VU ballistics, not the tape medium.
 
chops said:
Also, earlier in the thread people mention tape not handling high frequencies at high VU.  This is not true.  ... A sine wave at 20khz at operating level is fine.
Although I agree 100% with the beginning of your answer, regarding the need to use a reference tape for alignment, I must disagree on this last point.
You may record a 20kHz  tone at operating level and get a good reading at playback, but the recording process will distort heavily.
Of course, the 40kHz and 60kHz harmonics will be killed when played back, but if you record program that contents about the same HF energy, you will notice instantly that the signal is not correct, partly due to electronics distorting, head and tape reaching saturation, and self-erase, that is the process by which heavy HF content acts similarly to the erase head.
 
Back
Top