John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
(insert segue to blatant plug here) Then maybe you would enjoy listening to a show I am doing on WXYC this evening at 9PM EST. It is a 70s Underground Special, the music of the then-new FM band in the late 60s and early 70s in the US. Yep, I'm old. :cool: I plan to even play some of JC's guys!

Thanks. I'm not young either. I'm sure I'd love the music, alas, 9PM EST is 4AM my time…
 
Not the same thing this is all covered by a multi-physics model at equilibrium. The skin effect can produce fractional f effects. They are still continuous with level, the skin effect does not "stop" below a certain level.

If I put say 100nV (or whatever) on a transformer some of the results presented say mu collapses and I get nothing out. Please present evidence of this.

I am making some progress on doing 100nv signals and then bring them up to an easy to look at level. Attached is my current amplifier design. I was planning on using AD797s for the input stage. There are some old notes left on the image.

As to noise in transformers so far no one has mentioned vibration introducing undesired signals. There is a base line of vibration as low as 5mG (G is not grams!) Depending on how mechanical design this could be an issue at very low signal levels.
 

Attachments

  • Amplifier.JPG
    Amplifier.JPG
    184.1 KB · Views: 215
Actually Bob, I THREATENED Ampex with Nortronics 2 mil lam heads, IF they would not make them, themselves. Ampex (Pottberg) responded to me that the Nortronics head had terrible head bumps, apparently much worse than Ampex at the time. Ultimately, everybody went to 2mil lam heads, as they are measurably, and audibly quieter than the 6mil lam heads, I know, because I have one of each on my Ampex master recorder, and an A-B switch. A 6mil lam head would be noiser, and the actual rise in response would be closer to 5 1/2 dB/octave rather than 6 db/oct because of the losses. They had not changed the audio head design for almost 20 years and the Ampex head department didn't want the extra work necessary using 3 times as many lams per head.
 
Well, it looks like we will have to rely on Joachim to contact his colleague and report back. At least one manufacturer has 'done his homework' so to speak. I wish that I could get my boss to fork over $1000's of dollars and also pay for my time in testing every major transformer available, out there.

Yes it is: John G McKnight for those who has the AES preprints and Journals. Unfortunately, while I am pretty sure that I learned about this added noise from him, 43 years ago, as we often worked together on projects like this, his AES papers are rather complex and usually talk about other factors in tape head design, only mentioning the problems with 6mil or greater lams, in passing, even though it is actually 'Gorilla in the room' when it comes to high frequency noise and accuracy in high frequency response.

Looking through McKnight's papers, I can find no CLEAR graph that directly points to the magnitude of the problem, even though it implies it.

Ok, so you tell people to go read McKnights AES papers to learn about the high frequency noise caused by eddy current losses in the laminations of tape heads but you can't say just which paper or papers to read, and then you say you can't even find any specific reference to it in the papers you've read.

You can't even quantify it with regard to tape heads let alone the transformers we are talking about here yet you say it is the "gorilla" in the room.

This is what I am concerned about with moving coil transformers. They could present the same problem, if their laminations are 6 mil or thicker.

You can't even quantify it so how do you know it's any sort of "problem" worth worrying about?

Also, you talk as if a lamination is a lamination is a lamination.

Eddy current losses are inversely proportional to the lamination's resistivity. So what lamination materials are we talking about here?

Sowter mentioned 4 mil in his literature, but I don't know about anyone else. This converts to about 150um for 6 mils, and 50um for 2 mils. If anyone has any info on the lamination thickness of these transformers, I would be interested.

Why don't you call them or EMail them and ask?

I just got off the phone with David at CineMag and he says the laminations they're using in the CMQEE as well as most all of their other stuff is about 13 mil.

And I'd wager that's probably what Sowter, Lundahl and Jensen are using as well in their laminated core transformers.

se
 
Well, there are references to eddy current loss in the AES papers that I looked up. Unfortunately, convincing someone like you, SE, takes a more explicit graph to be cited, for you to even consider accepting it. The closest reference that I found that showed EXPLICIT graphs was 'EDDY CURRENTS IN COMPOSITE LAMINATIONS' by E. Peterson and l.R. Wrathall at Bell Labs in 1936. This was preprinted in the JAES, Dec 1977.
Unfortunately, instead of frequency, the graphs are in qualitatively configured in Greek symbols that show the right curves, but an arbitrary physical model. Difficult to rub people's noses in, because they will cry out in pain from the stress of making sense of it. Fig 2, for example, looks wonderful, but how do we put it in real terms?
If most manufacturers use 13mil lams, I should be able to measure it. The effect will be small, but real, and with the best designs, deplorable.
 
Actually Bob, I THREATENED Ampex with Nortronics 2 mil lam heads, IF they would not make them, themselves. Ampex (Pottberg) responded to me that the Nortronics head had terrible head bumps, apparently much worse than Ampex at the time. Ultimately, everybody went to 2mil lam heads, as they are measurably, and audibly quieter than the 6mil lam heads, I know, because I have one of each on my Ampex master recorder, and an A-B switch. A 6mil lam head would be noiser, and the actual rise in response would be closer to 5 1/2 dB/octave rather than 6 db/oct because of the losses. They had not changed the audio head design for almost 20 years and the Ampex head department didn't want the extra work necessary using 3 times as many lams per head.

Thanks, John.

So Nortronics DID have 2 mil lams, but they had bad head bumps. I seem to recall that you did an AES paper or two on solving the head bump problem. Would this have worked for the Nortronics heads to the point where they would have been acceptable to the folks at Ampex?

Cheers,
Bob
 
Well, there are references to eddy current loss in the AES papers that I looked up. Unfortunately, convincing someone like you, SE, takes a more explicit graph to be cited, for you to even consider accepting it.

*sigh*

Nobody's arguing about the existence eddy current losses, John.

What's being discussed is your claim of high frequency noise due to eddy current losses.

If most manufacturers use 13mil lams, I should be able to measure it.

What exactly do you plan to measure?

If most manufacturers use 13mil lams, I should be able to measure it. The effect will be small, but real, and with the best designs, deplorable.

So I guess it would be just as "deplorable" to use 99.99% copper in "the best designs" instead of 99.9999% copper?

It's nothing but a numbers game for you, isn't it?

I guess that isn't surprising given that most of the marketing out there is all about numbers.

se
 
Last edited:
Bob and Scott, you might read the October 1964 article in JAES by Keith O. Johnson on 'Techniques for Achieving Wide Dynamic Range from Magnetic Recorders' where reproduce tape head Q = 2 pi f L/Rs. A Q of 8 was mentioned at 10KHz, and I would bet that this was a 2 mil lam design. Scott might follow up with Kieth on this for clarification, then we could all be on the same track.
 
Bob and Scott, you might read the October 1964 article in JAES by Keith O. Johnson on 'Techniques for Achieving Wide Dynamic Range from Magnetic Recorders' where reproduce tape head Q = 2 pi f L/Rs. A Q of 8 was mentioned at 10KHz, and I would bet that this was a 2 mil lam design. Scott might follow up with Kieth on this for clarification, then we could all be on the same track.

So what exactly is the "track" here?

Tape heads or moving coil step-up transformers?

se
 
Moving back to moving coil phono stages, if possible, it might be useful to talk about the 2'd stage, or the IC op amp with feedback. While it is possible, but difficult to make an open-loop second stage, it is easy enough to make either a discrete, or an IC based second stage with negative feedback. The 627 is supposed to be an excellent design, so we might start with it, rather than something cheaper and potentially more problematic.
The second stage should always be jfet input, no matter what, in order to handle the relatively high input impedance that is often presented, depending on the input stage optimization. Here, the feedback loop should consist of 4 major elements, 3 resistors (precision) and one cap (1%) if possible. 2 of the resistors will set the breakpoints at 500 and 50 Hz and the third resistor should set the gain. There is some slight interaction, so be aware that final 'tweaking' might be necessary.
What I am on the look-out for is the worlds best (in subjective terms) DUAL, mini dip or surface mount, unity gain compensated, jfet input IC op amp. Anyone have any suggestions?
Now, in this position, absolute slew rate, etc, is not so important, because of the early (2KHz) rolloff ahead of the stage, that keeps the signals fairly polite and relatively easy to process. (more later)
 
Last edited:
This may be the case in some small studios.
Since I worked in recording studio and visited some others, I know that the majority of the recording studios are capable of very high quality recordings.

The level of some of the music is a different story.


What a studio was/is capable of and what they sometimes produced can be totally different 'things'. The particular studio I have in mind was one of the top international level studios in Britain. But that of course is not down to the studio management or permanent staff....it usually is down to the client and/or his producer.
 
Scott, if tape heads have problems with lamination thickness, then input transformers will as well. This is why I have gone through all this time and effort to acquaint people with eddy current loss and noise increase. PS, I have measured transformer eddy current losses in the past, and may try in future, if I have any suspicions of a problem.
I am directed to make the best possible phono preamp possible. Not the cheapest, easiest or the most politically correct one. That is why I care about such things.
 
Status
Not open for further replies.