Tape recorder project

[Analog Pitstop]

I have access to such instruments (dedicated SA with Phase noise modules + software), but not the right ones.
At home I have one that can do -136dBc/SQRHz and at work I have a few ones which can resolve -170dBc or so. Milion times more than enough for a tape machine.
Problem is, my models do not lock at 150KHz. Too low frequency. On the bright side: at least I know where to buy the right ones

But:
Even if I find such instrument, what is the use without a link to know how to interpret phase noise vs. SNR in tape recorders? Hearsay
Even if I find the instrument and the theory for analysis, what is the use without help (comments or a link?) to explain the noise bump around the signal?

So far, I observed these points about "the blob":
- it depends on tone frequency
- it depends on W&F and tape path
- it depends on rec head quality (sub-micron geometry definition of gap's trailing edge + gap length + head material)
- it depends on tape thickness, formulation and age
- it depends on tape speed
- it depends on bias frequency and its amplitude stability
- it depends on signal amplitude and noise floor
- and... sometimes it depends on how many times the tape was recorded/erased (memory). Hello SM911? What an unpleasant surprise, from you!

Counting the subjects, when I will know all these answers I can probably write a nice book.
Counting the literature sources as close to zero (so far), it also means that when I will write such a book, it MUST be one without any references!

So, then what's the point for such effort?

This thread is to help OP to do something with his x16tr machine. Originally he stated that only tape logic is busted&needed. Then this subject of tape-SNR read by SA instruments is off topic, no matter how much or less I like it.

 

[Analog Pitstop]

Though, I can write one important note from me:
- the combination of "The Blob" with Dolby or dBx will NOT result in HF sound transparency improvements! On the contrary.

 

[ruffrecords]

I have access to such instruments (dedicated SA with Phase noise modules + software), but not the right ones.
At home I have one that can do -136dBc/SQRHz and at work I have a few ones which can resolve -170dBc or so. Milion times more than enough for a tape machine.
Problem is, my models do not lock at 150KHz. Too low frequency. On the bright side: at least I know where to buy the right ones

But:
Even if I find such instrument, what is the use without a link to know how to interpret phase noise vs. SNR in tape recorders? Hearsay
Even if I find the instrument and the theory for analysis, what is the use without help (comments or a link?) to explain the noise bump around the signal?

Unfortunately I don't have a link but I am sure you can Google as well as I can.

Cheers

Ian

 

[EmilFrid]

Ok, so I basically have rec/play channels left to optimize to make this baby work, and I'm about to try a few different topologies.

My question, to all the veterans around here, is: is it worth trying to run the head in differential mode, rather than SE? I'm thinking both repro and rec, with the latter seeming more difficult to execute.

 

[Analog Pitstop]

before optimizing on this point, would not be a good idea to post a high quality photo of the heads? Additionally, a photo of their gap - down a very powerful microscope? To my non-veteran opinion, I wonder if these points are actually answering your question: worth or not.

 

[Analog Pitstop]

to be more precise, the micron quality photo of the right gap edge (plus surroundings) is needed only for the REC head (but for all tracks).
Without this info, it is a textbook risk - depending how high are your expectations.
I do not mean, just head busted or not. I mean, head in specs or not in specs (all of them). The quality of the gap will have direct influence on the quality of sound (i.e. recorded S/N at high frequencies). This is just not a static picture, it is a combination of magnetism physics plus tape movement in presence of W&F (incl. vibrations) plus discrete magnetic properties of the tape.

 

[EmilFrid]

Well, i know everything is very well head-wise, and I'm way too familiar with the head condition's influence on fidelity, so that's a non-issue. I want to focus on various amplifier topologies instead. A FET front end in the repro amp has been quite successful at least.

 

[Analog Pitstop]

not a veteran opinion, again, for your question on REC-AMP with balanced configuration: bridged Howland topology is not THAT hard to implement with Opamps. Actually is pretty simple to be done rock solid and with unconditional stability - even on the huge inductance which is driving. If you worry about <0.01% distortions at full dynamic range, however, it gets easily and quickly complex & expensive. But doable.

Interesting! Keep us posted!

P.S.
I must complain that is not clear to me why you want/expect improvements with balanced REC: if you worry about pick up noise... expect very less current in this impedance, and easily manage it by twisted pairs! Even so, why not simply place a small "standard" SE-REC-Amp next to the REC-head?
Or else, maybe you want to be sure that gap-field will be at its maximum/designed for all frequencies? If you worry for REC-head coils production imbalance: just play with reversing the SE connections, see which one gets you best SNR at 10KHz. But for dense multi-track heads I doubt that the Rec_head design is using pairs of coils: just one per track... I assume(!). Then your balanced REC-amp will really make no difference.

I wonder what the veterans will tell about this topic, from the praxis (customer) side, if any. Will be indeed very interesting! I hope just to be usable and down to practical details, not a snake-oil rant.

 

[Analog Pitstop]

A FET front end in the repro amp has been quite successful at least.

Repro side - totally different story...
But, good news, FET solutions for bridged Howlands (on Rec side) exist as well. A bit harder to realize (though apparently simpler) compared to OpAmp solutions.
I tested one about 4 years ago. Yes, indeed better. But unstable and really hard to realize (analytical distortions less than 0.005% at full dynamic): not followed it - too complex and you see new problems from every angle and from each component + PCB layout... Neh, not again for me, also not recommended to others. Basic SE with transistors was my final choice, not bothered even with Opamp. It fully depends on your needs and expectations... and your drive to discover.

 

[EmilFrid]

Thanks for chiming in. I appreciate it.

No, I'm not really expecting anything exceptional from a balanced configuration. I was just a bit curious about the difference, and if there would be any (even purely theoretical) positive aspects of such a technique. I go into intellectual masturbation mode way too easily...

The fact that the vast majority of pro - as well as consumer - machines have one leg of every tape head grounded seems to speak for the SE configuration. I read in an interview with John Stephens that he used some kind of feedback loop around the rec head, which was supposed to be his own unique design, to improve performance and fidelity. Problem is, he was kind of all over the place, jumping from subject to subject, leaving this hanging in the air. Didn't explain how the loop was configured, what it did (if anything) and why it did what it did.

 

[EmilFrid]

It fully depends on your needs and expectations... and your drive to discover.

Yes, I'm with you on this!

 

[Analog Pitstop]

well... all boils down to (actual) head gap quality and design, if it worth a ton of effort or simply nothing. Cleaning the REC-head: must be done always by the owner of studio or by the next highest paid.
Then next it depends on tape transport&path (machine design) and tape quality (?): and if you are not a big studio to secure a constant HIGH TAPE QUALITY in which you put your time and money - then it all does not worth a penny - only your hobby time!

I combine my hobby with making some/few artists happy with their mini-studio setups (yes, I know they won't be able to pay my efforts - so it is usually almost for free). No strings attached is best for me.

 

[Analog Pitstop]

Yes, I'm with you on this!

Speaking of topologies, may I ask, which one from below you would recommend?
Why?

 

[EmilFrid]

I'd say bottom one. For one, I'm not a fan of deriving base biasing from collector. And you?

 

[Analog Pitstop]

Bottom one is from Revox. Upper one is from HarmanKardon.
I have reasons to be somehow unhappy with Revox. Will change it in near future with HK, for a test (PCBs allow this). I do not have any bad stories with base from collector, if surrounding things are clean and frequencies low.

Analytically, it shows HK handles better both low end and high end.
Question to measure: how the IMDs are handled (and leakage from Bias). I am curious if HK will fare better. (I am fan of as high output impedance as possible...)
Can you make a prognosis?

 

[Analog Pitstop]

the prospect of deleting the 47uF capacitance is hypnotizing me

 

[abbey road d enfer]

I'm not really expecting anything exceptional from a balanced configuration. I was just a bit curious about the difference, and if there would be any (even purely theoretical) positive aspects of such a technique. I go into intellectual masturbation mode way too easily...

The main issue there is hum pick-up. It is addressed in many ways. HIgh-Z heads need thoroughly shielded connections. Was an issue with tube recorders, particularly because Hi-Z balanced inputs were not a viable option.
For medium-Z heads, there are many different approaches:
- Locating the low-level part of the repro amp in the headblock assembly, so an elevated signal was sent to the rest of the repro amp.
-Balanced input. Not favorable to hiss performance, coz 3dB noisier than SE.
-Nagra used Virtual Earth inputs. They claimed it solved the hum issue, but VE inputs are as sensitive to electrostatic hum fields as high-Z. It did solve the issue of head resonance, though.
-Extra shielding

That's why some manufacturers used low-Z repro heads, with a step-up xfmr in teh preamp. It also allowed to use the record head as sync head with no modification of the signal path. Of course, it's a costly solution but justified in a pro recorder.

The fact that the vast majority of pro - as well as consumer - machines have one leg of every tape head grounded seems to speak for the SE configuration.

Indeed. It works well enough in many cases. The fact that the head is floating makes the risk of ground loop irrelevant. The head must be grounded at the preamp's input.

I read in an interview with John Stephens that he used some kind of feedback loop around the rec head, which was supposed to be his own unique design, to improve performance and fidelity.

Looks like a derivative of the active guard principle.

 

[EmilFrid]

Thanks Abbey!

I appreciate your insights. So it seems that running tape heads in a balanced configuration isn't really worth exploring, if the big advantage is hum rejection. I've owned many tape machines - pro and consumer - and none of them has had any head related hum issues - it's always turned out to be PSU related in the end.

Grounding the heads at preamp input is very good advice. Seems obvious, but something I could see myself overlooking. Will take this to heart.

I actually found a (kind of messy) schematic of a Stephens, with the feedback loop (?). I'm not certain, though, exactly how it performs in real life because his opamps aren't your regular DIPs... Attached schematic for anyone curious.

As for playback, I've noticed a FET input followed by a couple of opamp stages is great. Maybe I should just focus on this category of configurations.

 

[abbey road d enfer]

I actually found a (kind of messy) schematic of a Stephens, with the feedback loop (?). I'm not certain, though, exactly how it performs in real life because his opamps aren't your regular DIPs... Attached schematic for anyone curious.

Interesting schemo. It's a bootstrapped Virtual Earth input. I guess the advantage is that it presents a light load to the source, but the common-mode impedance of both lines is low, which makes them not too sensitive to electrostatic fields.

As for playback, I've noticed a FET input followed by a couple of opamp stages is great. Maybe I should just focus on this category of configurations.

FET's were popular because of their high impedance, which made them a suitable replacement for tube electronics.
When solid-state recorders with medium impedance heads became the norm, bipolar transistors proved capable of better performance, noise wise, than tubes or FET's.
Today, low noise FET's can challenge bipolar transistors. I reckon that someone building a tape recorder today should use low-noise FET's.
The challenge is the impedance of the heads, that varies over a range of more than 1 to 100.
Bipolar transistors have an Optimum Source Impedance that doesn't match well with this constraint. FET's are almost insensitive to source impedance.

 

[EmilFrid]

Interesting schemo. It's a bootstrapped Virtual Earth input. I guess the advantage is that it presents a light load to the source, but the common-mode impedance of both lines is low, which makes them not too sensitive to electrostatic fields.

Ah, I see! Very interesting. I might just conduct some sort of experiments on this. Just for the heck of it. This build is just as much about achieving the end goal, as well as having some fun and learning a thing or two. The Stephens oddities fascinates me a lot. Not sure why. Found a huge pile of Stephens' technical documentation on archive.org yesterday.

FET's were popular because of their high impedance, which made them a suitable replacement for tube electronics.
When solid-state recorders with medium impedance heads became the norm, bipolar transistors proved capable of better performance, noise wise, than tubes or FET's.
Today, low noise FET's can challenge bipolar transistors. I reckon that someone building a tape recorder today should use low-noise FET's.
The challenge is the impedance of the heads, that varies over a range of more than 1 to 100.
Bipolar transistors have an Optimum Source Impedance that doesn't match well with this constraint. FET's are almost insensitive to source impedance.

I've been tinkering with a few different FET's and this far the best overall performer has been the 2sk170GR, which actually is the one used in some of the Fostex machines, including G-series. Next stage is an op amp configured as an IEC playback filter, which works great. From there on I'm still a bit unsure. It's been a while now, and some measurements would be good.