Too much tape hiss when recording (R2R)

Have you actually measured your signal to noise? Just listening for hiss is not a valid measurement. If you're expecting it to sound like a typical digital "silence", then you're going to be disappointed.

It should be about 72/70dB NAB Wtd and 67/65db Unwtd. at 15"/7.5" measured at 3% THD @1kHz. DBX adds about 30dB Noise Reduction.

The specs are here: Tonbandmuseum - Sie sind im Bereich: Tascam ATR-60-8-Docs
 
Have you actually measured your signal to noise? Just listening for hiss is not a valid measurement. If you're expecting it to sound like a typical digital "silence", then you're going to be disappointed.

It should be about 72/70dB NAB Wtd and 67/65db Unwtd. at 15"/7.5" measured at 3% THD @1kHz. DBX adds about 30dB Noise Reduction.

The specs are here: Tonbandmuseum - Sie sind im Bereich: Tascam ATR-60-8-Docs

Ah! That is rather better than my crude result by just playing OP's clip in Samplitude. I got -5dBfs for the tone and the noise fell to -60dBfs so a S/N of 55dB and that is really where I would have expected an OR machine to be? Yes, a top bllx Studer carefully setup with a really good tape will be 10dB better.

Op might have a distorted bias waveform? Tricky to tell without a decent scope and that is the buggerooter of tape recording! To do it well you need to spend about AS much on test gear as the bloody recorders! As Rob alluded to, bias should really be set for best distortion.

One test would be to measure the replay noise of virgin tape and then after just a silent pass with the taps shut so you are only getting the basic noise of the recording amplifiers plus DC noise from a wonky bias oscillator.

Dave.
 
I expect you already know this, but the obvious is often worth repeating. In analogue recording generally it's important to have the programme material at the highest possible level, i.e. as far above the noise floor as possible. That's much less of a problem with digital. And on top of that, noise is cumulative as you layer up different tracks. And AYK a lot of us outside of classical music deliberately push the level into saturation: it 'juices up' the sound in a way digital doesn't. I could also go on and on about the psychoacoustics of human hearing (but I won't), especially the phenomenon of 'noise masking' that goes on in the brain and is little understood. So I'll just say 'can you really hear the noise in an intrusive way when the music is ACTUALLY PLAYING?' (as opposed to empty or quietly recorded tape). We are all a bit prone to getting carried away with the Holy Grail technicalities of noise and dynamic range; but the only test that really matters is the musicality of the finished product: just listen to the incredibly obvious hiss at the very start of Marvyn Gaye's 'Heard it through the Grapevine' . Does anyone care, or even really hear it after the first 4 bars? No. That's neural noise masking for you. And one of the greatest cuts of popular music ever made.
 
Ah! That is rather better than my crude result by just playing OP's clip in Samplitude. I got -5dBfs for the tone and the noise fell to -60dBfs so a S/N of 55dB and that is really where I would have expected an OR machine to be? Yes, a top bllx Studer carefully setup with a really good tape will be 10dB better.

Op might have a distorted bias waveform? Tricky to tell without a decent scope and that is the buggerooter of tape recording! To do it well you need to spend about AS much on test gear as the bloody recorders! As Rob alluded to, bias should really be set for best distortion.

One test would be to measure the replay noise of virgin tape and then after just a silent pass with the taps shut so you are only getting the basic noise of the recording amplifiers plus DC noise from a wonky bias oscillator.

Dave.


I saw similar results, but I didn't know if the posted sample was from a single channel, or from 8 channels at the same time. Obviously, if its a single channel, he has some room for improvement, but if it's an 8 channel blended sample, it's probably right in the range. I don't remember for certain, but I think 8 channels would be somewhere close to 9dB noise increase. Plus the S/N is measured at 3%THD, not at 0dB which is less than 1%. If the signal was recorded at -5dB, then there's unused headroom that could be added.
 
I saw similar results, but I didn't know if the posted sample was from a single channel, or from 8 channels at the same time. Obviously, if its a single channel, he has some room for improvement, but if it's an 8 channel blended sample, it's probably right in the range. I don't remember for certain, but I think 8 channels would be somewhere close to 9dB noise increase. Plus the S/N is measured at 3%THD, not at 0dB which is less than 1%. If the signal was recorded at -5dB, then there's unused headroom that could be added.

We may be at crossed decibabbles here Rich! I saw -5dB 'Full Scale' I have no idea how that relates to OP's signal voltage? (think he said +4dBu?)

But yes, if he is not referring levels to 3%THD then he is short changing himself noisewise.

I am just happy to be done with it all!


Dave.
 
A Technician can only test and adjust the machine according to the service manual and the specs that are there. I have done it many times but then once in a while a complaint comes in that is based on abstract testing and usually by a person with very little knowledge of the subject. Then you need to explain to them the right way to test these types of equipment.

No they are not perfect and the reason the commercial tapes do not have hiss as most of them used Dolby A that cost at least $1000 per track.
You can not compare consumer level equipment to that which is 10 of thousands dollars more expensive. However if you know what you are doing you can come out with a very close result. I am afraid you will need to use dBx for the unit if you are looking for dead silence. Or go off and buy a rack of Dolby A processors.
 
A Technician can only test and adjust the machine according to the service manual and the specs that are there. I have done it many times but then once in a while a complaint comes in that is based on abstract testing and usually by a person with very little knowledge of the subject. Then you need to explain to them the right way to test these types of equipment.

No they are not perfect and the reason the commercial tapes do not have hiss as most of them used Dolby A that cost at least $1000 per track.
You can not compare consumer level equipment to that which is 10 of thousands dollars more expensive. However if you know what you are doing you can come out with a very close result. I am afraid you will need to use dBx for the unit if you are looking for dead silence. Or go off and buy a rack of Dolby A processors.

Percisely^. My son started off building tracks with a Teac A3440 at 15ips. The machine was easily good enough for say two 'bounces' but as he got more adept and versatile and wanted more complicated tunes the cracks, well HISS, began to show (no NR you see). A stop gap was a first generation Hi Fi VHS recorder. He could bounce a mix to that and send it back with no noise penalty (IIRC VHS HF tape was about 16 bits so we got about a -80dB S/N.

But the dynamic range of CD was always in our ears and so we decamped to computers!

Dave.
 
The one detail I see missing in this thread is the detail in setting bias. The OP stated -3db bias. That’s not enough detail for me. It has to be -3dB overbias. You back off the bias until you reach a peak, then increase the bias until it it goes down 3dB. You can have -3dB overbias, or -3dB underbias so you could be on the wrong side of the bias adjustment. Also most of the information in this thread is accurate. Noise is additive, the more channels you sum together the more noise you’re going to have. Mixing in the old days we’d mute the faders until we knew a part was coming in. I’m also assuming you’re recording at 15ips.

Virgin tape will always have less noise than a tape that has been recorded on. The bias is the culprit when recording. As stated earlier, it’s true function is to correct linearity in the recorded signal and reduce distortion. Noise specs are a certain sized window in a correctly aligned machine. You must record with your levels hot.

May I make a suggestion if your machine allows you to do it. Once I got past using noise reduction, I used to align my machines for a variation of a non standard 320nWb/m alignment. I used to do this with Ampex 456. When setting your repro levels use -2dB reference instead of 0 when adjusting the reference level. Do it for 1k & 10k. If you’re really intending on going into the tape thing, learn as much as you can about reference levels, vs. the 250nW/m tape. There’s no need to buy multiple repro tapes for different levels. This will move your signal to noise window up a bit towards more distortion/compression & less noise. The record part of the alignment is still exactly the same. So you’re recording a bit hotter and the noise is reduced but you’re reducing the playback levels a bit. If you’re getting a Studer, definitely look into the high level alignments like +6 or +9.
 
When setting your repro levels use -2dB reference instead of 0 when adjusting the reference level. Do it for 1k & 10k.

Don't do this on a machine that you plan to use with DBX though as the DBX will greatly exaggerate the distortion and it will sound bad.
 
Back
Top