John Curl's Blowtorch preamplifier part II

[Joachim Gerhard]

I bought that particular Lundahl transformer because it has 4 primary and 2 secondary windings. That way it is very flexible. Step up ratio can be 1:5, 1:10 and 1:20 and impedance can vary over a wide range. I though that was very handy because i could experiment with a variety of arangements. I also consulted with the late Allen Wright and he recommended amorfous core. I also had a chat with Brian Sowter and he is more conviced about mu-metal as core material. I have reported that here before. So sound quality per se was not the only reason i bought this transformers. Sorry if i cunfused anybody here.

[john curl]

I want to thank everyone for getting me off my duff and searching the internet to put some more input here. I ran across an analog tape recorder app note by Dale Manquen dated 2001, more than 30 years after I left Ampex.
They CHANGED the transformers and the heads to 2mil laminations in subsequent models of the MM1000 and the AG-440 of BOTH the heads and THE TRANSFORMERS, just like I asked them to, in 1968, and they virtually laughed at me then. Of course, I went and PROVED my contention with both Q meter and noise with frequency measurements, and rubbed their nose in it, as I am prone to do.
Years later, they upgraded. Nice to see people catching up, even if they could have gained a small advantage on the competition, years before, IF they would had only given a darn.
Now, if I were a betting man, I would BET that Sowter uses 4 mil lams. Ampex perhaps 35 years ago, went to 2 mil lams, which will measure better AND contribute less noise at frequencies between 10K and 20KHz. If you are over 50, maybe you don't care, but if you are 18-35, you might certainly care. This is an ADDED hiss from the eddy current losses, due to thick laminations. 6 mils used to be standard, perhaps in 1950. 2 mils was possible, and commercially made in 1968. 1 mil was also possible, just more expensive and delicate to work with.
Now what are the lamination thicknesses of the audio input transformers talked about here? Sowter, Jensen, Lundhal, etc. Anybody know? If not, why not?

[Tom Danley]

Hi John, all

I had observed something once which may be worth tossing in as a ‘for what it’s worth”

About 15 years ago I was hired to work on improving an output transformer for a 250 Watt tube amplifier. One of the things I did was to casually explore the distortion the transformer added.

Using an HP-3562a network analyzer and the two port vector subtraction approach (where the distortion from the source is excluded) I found a consistent thing.

If one viewed the spectrum with the same start / stop points, the “shape” of the distortion envelope (the shape set by the peaks of all the harmonics) stayed more or less constant regardless of the input frequency.
If I remember right, the top of the hump for that envelope was in the 200-300 Hz neighborhood and this being a 25 lb transformer, an E/I stack, M-6 I think with a very small signal low corner F.

What I saw was all the other transformers I tested have a similar general hump shape.
To be clear, this is not caused by the normal inductive corner but (I think) due to the sort of incremental nature of the magnetic material at a very low level.
Since you guys are getting serious about transformers, I thought this might be applicable.
Best,
Tom Danley
Danley Sound Labs

[johnferrier]

This is indeed a well known phenomenon. If you start to do something that is new to you, be it playing the violin or running a steeple chase, you have to do it over and over again so that the brain can develop the motor patterns needed. The more you do it, the quicker the brain can call up the required patterns and the more automatic this becomes. More 'brain power' can than be diverted from the process to tackle other tasks. This is a typical process found in most living creatures and is called 'learning' . . .

and at other times called 'burn-in'


.

[DF96]

Given the high efficiency, and therefore low attenuation, of a transformer its thermal noise will be low. I can appreciate that thermal noise is not just due to winding resistance but also core losses - any loss mechanism will contribute thermal noise.

Excess noise normally means noise over and above thermal noise. What are the sources of excess noise in a transformer? Perhaps leakage inductance contributes indirectly by allowing coupling between the transformer and its EM environment? In that case, noise will depend on the environment so perhaps should be treated as 'EM soup rejection ratio'.

[Steve Eddy]

Originally Posted by Joachim Gerhard

I bought that particular Lundahl transformer because it has 4 primary and 2 secondary windings. That way it is very flexible. Step up ratio can be 1:5, 1:10 and 1:20 and impedance can vary over a wide range. I though that was very handy because i could experiment with a variety of arangements.

Yeah, it's fine for experimentation, but it's one of the things that have turned me off on the Lundahls.

Most of their transformers are of the "Swiss Army knife" variety, forcing you to strap them externally from their shields which can make them more susceptible to noise pickup.

I also consulted with the late Allen Wright and he recommended amorfous core. I also had a chat with Brian Sowter and he is more conviced about mu-metal as core material.

I'm with Sowter on this one. Perhaps because I never cared much for freeze dried coffee.

se

[Steve Eddy]

Originally Posted by john curl

This is an ADDED hiss from the eddy current losses, due to thick laminations.

Eh?

Would you care to explain how eddy current losses in the core produces added hiss?

se

[john curl]

Why don't you try to figure it out for yourself, Steve?

[jan.didden]

Originally Posted by Tom Danley

[snip]What I saw was all the other transformers I tested have a similar general hump shape.
To be clear, this is not caused by the normal inductive corner but (I think) due to the sort of incremental nature of the magnetic material at a very low level.
Since you guys are getting serious about transformers, I thought this might be applicable.
Best,
Tom Danley
Danley Sound Labs

Tom,

Are you aware of Menno Vanderveen's AES paper on exactly this subject?

jan didden

[scott wurcer]

Originally Posted by Tom Danley

Hi John, all

I had observed something once which may be worth tossing in as a ‘for what it’s worth”

If one viewed the spectrum with the same start / stop points, the “shape” of the distortion envelope (the shape set by the peaks of all the harmonics) stayed more or less constant regardless of the input frequency.
If I remember right, the top of the hump for that envelope was in the 200-300 Hz neighborhood and this being a 25 lb transformer, an E/I stack, M-6 I think with a very small signal low corner F.

What I saw was all the other transformers I tested have a similar general hump shape.
To be clear, this is not caused by the normal inductive corner but (I think) due to the sort of incremental nature of the magnetic material at a very low level.
Since you guys are getting serious about transformers, I thought this might be applicable.
Best,
Tom Danley
Danley Sound Labs

I'm having trouble picturing what you are describing. If the B-H curve had steps I would expect sudden changes in distortion components. And as we discussed before a step at 0 would mean below a certain input there would be no output.

Jan could you send a copy? That might make it clear.