• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Loosening up magnetic "stiction"

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I read this AES paper http://www.mennovanderveen.nl/nl/download/download_1.pdf
by Menno Vanderveen that was very interesting. It pointed out that output transformers have inherent low linearity at very low levels of audio input signal. Without going into all the details it shows that the worst effects occur between 20 to 1000 hz range and cause a loss of low level details. It further states that some of the holographic qualities of SE amps art probably due to a gapped OPT that has low output impedance that causes the tranformer to maintain its inductance at very low signal levels - in other words that magnetic stiction can't exert as much influence in that topology and has much more to do with the low level retrieval of sound and their holographic qualities.

Anyway, this all makes a lot of sense to me. He mentioned experiments where high frequency bias signals are sent along with the audio signals that seem to loosen up that stiction in OPTs. JLTi in Australia makes a device called a Lemm (I think) that does that, but its only sold in Australia and I haven't heard of anyone else doing it. It certainly makes sense to me as tape recorders have always used high frequency bias to make recordings if memory serves. Does anyone have personal experience with this, or thoughts on practical implementation of a high frequency bias device?
 
Yes, hysteresis is the correct term but I like stiction because it conjures up a more physical image that the average person can visualize. Hysteresis isn't commonly discussed as being a varying quantity with level so I'll stick with stiction.

Part of the problem with sizing the OPT correctly is that if you go for minimum distortion at low frequencies you inevitably need a lot of iron. And that invariably makes the problem worse at low signal levels. Compromises must be made.
 
There was a thread about this recently. You should be able to find it with a little searching. Bottom line: the effect can only be seen with really, really lousy core materials. With a pretty normal output transformer, I was unable to see it. And note the lack of that effect in MC stepup and line level input transformer measurements.
 
SY said:
Bottom line: the effect can only be seen with really, really lousy core materials. With a pretty normal output transformer, I was unable to see it. And note the lack of that effect in MC stepup and line level input transformer measurements.


I'm not so sure that's the case... take a look at some measurements from Cinemag of their 80% nickel vs. 50% nickel line output transformers. Rather stunning difference, unless I am misinterpreting here! Is there a noise difference? Or is that core-related?


Cinemag

More: Core Comparison
 
The discussion is buried in the Blowtorch saga, dating back to May sometime.

Measuring a standard EI output transformer (good quality but non-exotic) I saw absolutely no trace of this effect at the levels and frequencies that Menno was talking about. This one was a 1250:16/8/4 unit that I use in my bigger power amp.
 
The distortion from core hysteresis is in the form of additional high harmonic magnetizing current. It gets converted to voltage distortion by the source impedance. So one obviously wants to use a low source Z for driving a steel core.

Looking at the Cinemag Fig. 1 and Fig. 2, I see they have Rs = Rl, or a damping factor of 1. The steel core drops to the nickel core dist. level at about 30 X the freq. This would indicate to me that one needs a source Z 30 times lower (or damping factor of 30) to get the steel core to equal the nickel core. So local neg. feecback to the driver stage could clear this dist. up.

Using both local fdbk and a nickel core would be even better yet.

One can inject a HF bias field into the core to remove this Hyst. dist. The best way is if it is orthogonal to the main field. But this is difficult to set up. (have done it with ferrite pot cores only, by winding thru the bolt hole)

Next best is to use two cores with the main winding wound over both of them, and the bias winding wound as two separate windings around them and connected anti-phase so as to null sum in the main winding. I've tried this once and it sorta works, but the Mu balance between the cores has to be perfect or HF effects show up in the audio. DC currents can wreak havoc to the balance too.

Don
 
SY said:
The discussion is buried in the Blowtorch saga, dating back to May sometime.

Measuring a standard EI output transformer (good quality but non-exotic) I saw absolutely no trace of this effect at the levels and frequencies that Menno was talking about. This one was a 1250:16/8/4 unit that I use in my bigger power amp.


I wasn't able to locate the subject under "blowtorch".

If you read Menno's article carefully what he's really saying is that the human ear, at the threshold of audibility requires a change in sound pressure level 100000 times between 20hz and 4000hz. This isn't something he's making up! It's the ISO curve for the average human ear. He's not really taking about varying levels of hysteresis with level. What he's talking about is the extreme range of flux density required of the core in an output transformer.

A speaker is a power consuming device rather than primarily a voltage consuming device so an OPT feeding it most likely will require a greater range of flux density between loudest and softest passage than a MM xfmr. Isn't there some electrical law somewhere stating that?

Providing that is true it means one could probably create a MM xfmr with magnetic materials that is within that flux density range but far less likely you could get a permeable material for the larger range required for a OPT. Its not about the quality of magnetic material, but about the "range" of flux density the material can accomodate.
 
One further clarification: What I gather from the article is that high frequency bias might be able to "extend" that flux density range on the low volume end.

If we can get back to my original question. Does anyone have a clue how one would implement this - i.e. where would you feed the bias signal, bias signal level, frequency etc? Whether one would have to filter it out before reaching the speaker? I sure don't know how one would implement it. Even if some disagree with Menno's audio philosophy in other ways it doesn't mean he isn't on to something here.
 
smoking-amp said:

One can inject a HF bias field into the core to remove this Hyst. dist. The best way is if it is orthogonal to the main field. But this is difficult to set up. (have done it with ferrite pot cores only, by winding thru the bolt hole)

Next best is to use two cores with the main winding wound over both of them, and the bias winding wound as two separate windings around them and connected anti-phase so as to null sum in the main winding. I've tried this once and it sorta works, but the Mu balance between the cores has to be perfect or HF effects show up in the audio. DC currents can wreak havoc to the balance too.

Don


I might be mistaken (it wouldn't be the first time) but I thought there was a way of doing it simply by superimposing the bias signal on the audio signal of the primary. Am I wrong?
 
"but I thought there was a way of doing it simply by superimposing the bias signal on the audio signal of the primary. Am I wrong?"

That's the way the tape recorders did it, but then the HF bias has to be filtered out again. OK for a small signal level, but it's more of a problem at power output levels.

Orthogonal techniques avoid the problem.

Don
 
Don,
You are probably right about the greater efficacy of that kind of bias winding. But I'm sure the implementation by Joe Rasmussen, head of JLTI, uses the tape recorder type of bias. Again, he only sells it, the L.E.M circuit board, in Australia. Here's a quote from his website:

"During a discussion by Menno Vanderveen during the European Triode Festival in 2006, the problems troubling the designer of audio transformers were highlighted. This included memory (hysteresis) and permeability, which is the ability of the signal to transfer from the primary to secondary in tact. This directly related to the density or degree of magnetisation. The real problem here is that permeability is reliant on a healthy signal level. Yet the dynamic nature of music contains very low levels at the beginning and end of a transient.

Further private discussions with Menno and I suggested a solution that takes advantage of the excess bandwidth of the output transformer. A method of exciting the transformer well above hearing range and yet low enough not to cause damage further down the line, such a the Tweeter. The result is now available as the LEM, Linearity Enhancement Module, and can be fitted to just about any tube power amplifier. If you have a Tube Power Amplifier of any type, but especially Push-Pull, contact me about having the LEM installed."

I sent him an email asking if it would be possible to purchase one even though I don't live in Australia but haven't heard from him yet.

Eric
 
This topic is as old as the hills....and discussed to death back in the 1950' and 1960's.....
It has to do with the linearity of the BH loop just as it passes at FLUX Density axis = 0 ......
For those familair with FETs, there is a region called "Subthreshold region" at very low currents, this area is not consistent with the rest of the FET curves...same with the typicall BH loop for most magnetic metals...
Adding high frequency bias makes no sense.... The bandwidth of the transformer, leakage inductance and winding capacitance will squash this high frequency bias signal.... ALso the bias signal, even if it managed to drive the windings would have to be assemytrical to offset the core.... Thats why the easy solution is to offset the DC in the core for Push-Pull...this way low level signal stay away from the 0 axis....
This could also be accomplished by offsetting the number of primary turns on each side of the Center-Tap, or simply squewing the biasing.....

Chris
 
cerrem said:
This topic is as old as the hills....and discussed to death back in the 1950' and 1960's.....
It has to do with the linearity of the BH loop just as it passes at FLUX Density axis = 0 ......
For those familair with FETs, there is a region called "Subthreshold region" at very low currents, this area is not consistent with the rest of the FET curves...same with the typicall BH loop for most magnetic metals...
Adding high frequency bias makes no sense.... The bandwidth of the transformer, leakage inductance and winding capacitance will squash this high frequency bias signal.... ALso the bias signal, even if it managed to drive the windings would have to be assemytrical to offset the core.... Thats why the easy solution is to offset the DC in the core for Push-Pull...this way low level signal stay away from the 0 axis....
This could also be accomplished by offsetting the number of primary turns on each side of the Center-Tap, or simply squewing the biasing.....

Chris


The oldness of the topic may be true but when I introduced this topic I wanted to get people with "personal experience" with this. So far everyone has made an assertion backed by their "opinion" with no personal experience of having tried it.

Now, I'm not saying it works. I want to find out if it works. But so far all I've heard is assertions not backed by personal experience. Why individuals feel the need to pontificate on areas they aren't conversant in seems just an exercise in ego. This seems to be a particularly American (USA) disease, and being American, I know it when I see it.
 
exeric said:



The oldness of the topic may be true but when I introduced this topic I wanted to get people with "personal experience" with this. So far everyone has made an assertion backed by their "opinion" with no personal experience of having tried it.

Now, I'm not saying it works. I want to find out if it works. But so far all I've heard is assertions not backed by personal experience. Why individuals feel the need to pontificate on areas they aren't conversant in seems just an exercise in ego. This seems to be a particularly American (USA) disease, and being American, I know it when I see it.


Obviously you haven't been around the world much... I have and you haven't seen real ego in the engineering community unless you visit certain countries I will not mention....
I have done research in this area and have experience in solving this problem you are refering to....So I do have personal experience with this.... i don't care to go into full details on a internet forum.... I don't mind helping, but there is a limit to where I draw the line with details... especially when some of my ideas have been ripped off..I am not the only one, there are good folks around here who have been too helpfull at times, only to have someone take the idea and run with it, as in patenting things and making a business out of someone elses ideas...
I was merely pointing out that this problem has existed a long time and many have looked at solving it and there are a few valid solutions to this....
If you want answers then learn magnetics in an advanced format and do the work yourself and figure it out yourself...It is real easy to sit back and use the computer and have others provide the answers.....

Chris
 
Not interested in getting into a flame war. I was merely stating your information was unhelpful. I've worked over 20 years troubleshooting avionics for the airline industry, DMEs, Transponders, etc. I'm not out to impress as you could gather from my heading of the subject of "stiction" instead of "hysteresis". There's no reason to be errudite when simpler explanations will do, even when you know the more detailed explanation.

But I see ******** all the time where individuals try to hide their relative ignorance behind technical jargon , and always going for the one upsmanship to prove their expertise. It is definitely a more cultural emphasis in this country than others. It's so tiring, believe me.
 
The oldness of the topic may be true but when I introduced this topic I wanted to get people with "personal experience" with this. So far everyone has made an assertion backed by their "opinion" with no personal experience of having tried it.

Exeric, au contraire, I did try it. I set up a measurement exactly in accordance with the prediction and saw no trace of the effect.
 
SY said:


Exeric, au contraire, I did try it. I set up a measurement exactly in accordance with the prediction and saw no trace of the effect.


Sy,
That's good. Was it with an OPT and do you happen to remember the level of the signal at which it was measured? Also did you do a listening test with otherwise familiar equipment. That's really the final test.

Regards,
Eric
 
Yes, an OPT. Northlake C126-132. EI core, 1250:16 ohms. Primary driven balanced with a 3k resistor in each leg. 15R load on the secondary. 20Hz and 200Hz at levels down to 0.3mV.

I'm not sure how to answer the listening test question- how do I listen for an effect which doesn't seem to be there, with test signals pushing the audible threshold?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.