Its interesting but seems to make a big leap of faith from 'a difference is perceived' to 'fuzzy distortion is the likely culprit'. What's missing is testing the hypothesis that audible transparency is reduced owing to the presence of "fuzzy distortion". I'm left wondering why no listening tests were reported. I offer my own alternative hypothesis for loss of transparency which is RF-induced noise modulation, its a bit simpler to understand than MOJH's one
It also helps tidy up why cables sound different at the same time100 pV / rt Hz http://www.thinksrs.com/downloads/PDFs/Manuals/SR554m.pdf
As a published spec in an industrial environment its probably real. There a number of interesting specs in the manual linked. 2 channels for $1,000? Too cheap for this market.
As a published spec in an industrial environment its probably real. There a number of interesting specs in the manual linked. 2 channels for $1,000? Too cheap for this market.
Jlsem: No (but I have that book, courtesy of you, and loved it); in audio, you have to juggle several parameters and how to balance noise against noise rejection, bandwidth, and distortion. My approach, the calculations, and the results are documented in an article you're no doubt familiar with- I optimize bandwidth and transient response, then make sure that the added noise compared to source thermal noise hasn't degraded more than a dB or so. Others have different approaches, and there's several excellent phono stages out there where the noise can't be heard in any conceivable home listening situation.
Interestingly, with a tube or JFET input, the equations all show that a ridiculously high level of step-up is optimal for noise in an audio amplifier, since the ratio en/in blows up for devices with extremely small in. Problem is, as you know, that the bandwidth and distortion go to hell at those high ratios.
SY, take another look: Page 3, 2nd column, the para just after equation # 15.
Malcolm says: "Never the less, both calculations yield results of only a few electrons".
In the version I am familair with, the "hifi" version with less math, he says: "in the order of one electron".
SO I ASK AGAIN: is it possible that "only a few electrons" can create a flip of magnetic thingies in a transformer, hence allowing accurate trasmission of such low signal levels?
It is a VERY SERIOUS Question, IMO and IMO is more important than the ultra low noise possible with a transformer input.
Answers from any technical heavy will be appreciated.
Regards, Allen
Allen, he's not talking about the signal from the cartridge, he's talking about the electron charge transfer into the transistor base, an internal mechanism. Let's get a feel for the signal numbers:
Take an MC with a typical output, say 0.2mV at 5 cm/s and 1kHz. The cartridge is likely to have a self noise of -70dB below that, which is about 0.03uV. Now, for argument's sake, let's pick an effective load of 100R. That means the signal current is about 3 x 10-10A.
By definition, an ampere is about 6 x 1018 electrons per second. That means our signal current is about 1.9 billion electrons per second or 1.9 million electrons per cycle. That's not enormous, but it's certainly bigger than "about one."
Take an MC with a typical output, say 0.2mV at 5 cm/s and 1kHz. The cartridge is likely to have a self noise of -70dB below that, which is about 0.03uV. Now, for argument's sake, let's pick an effective load of 100R. That means the signal current is about 3 x 10-10A.
By definition, an ampere is about 6 x 1018 electrons per second. That means our signal current is about 1.9 billion electrons per second or 1.9 million electrons per cycle. That's not enormous, but it's certainly bigger than "about one."
van den Hul doesn't sell transformers.
His points on shielding are well taken. Very tight balance is also important for noise rejection. It's hard to get a quiet phono stage when the raw supply is in the same cabinet.
I spoke at length with Mitch Cotter today about audio transformers. He agrees that a low effective resistance is possible, and the transformer that I have selected so far leaves a little room for improvement.
He also independently (of VDH) warned me about ground loop problems and related hum pickup with transformers. We shall see.
He also independently (of VDH) warned me about ground loop problems and related hum pickup with transformers. We shall see.
Input transformers are quite the complex brew with more options than a Chinese Restaurant menu. Also interesting is the effect of different source impedances on transformer response. The chart for the SRS box shows quite a difference between 0 Ohms, 1 Ohm and 10 Ohms source impedance (the range that MC cartridges represent). With a smaller ratio there will be less variation.
While the lowest noise will be had with a transformer the real limitation is the surface noise of a disk. Even a very quiet disk (fresh cut and unplayed lacquer) will be noisier than any of these numbers. A good design will start with the complete envelope of requirements of the application and then balance that against what the technology can support. Aside from getting the basics right (no hum, correct RIAA, enough gain for the system) its all about the designers particular mix of tradeoffs. The same as a good Chef.
While the lowest noise will be had with a transformer the real limitation is the surface noise of a disk. Even a very quiet disk (fresh cut and unplayed lacquer) will be noisier than any of these numbers. A good design will start with the complete envelope of requirements of the application and then balance that against what the technology can support. Aside from getting the basics right (no hum, correct RIAA, enough gain for the system) its all about the designers particular mix of tradeoffs. The same as a good Chef.
I should be pointed out that it was not just the transformer's S/N ratio that got us on the track of attempting to REMOVE the transformer from the input audio path of many components.
It was really low frequency distortion, and the addition of high frequency noise due to eddy current losses in many commercial transformers, that led the way. While the RIAA tended to roll of the high frequency noise from both the record or the transformer, the NAB or IEC tape equalization did just the opposite.
This is where I found myself, back in 1968, looking at the transformer noise problem.
One way to evaluate this is with a Q meter that measures the ratio of inductive impedance to the EFFECTIVE resistance of the transformer with frequency. I used a GR unit for my measurements. The main transformer that we were looking at was a 10:1, or so, Beyer transformer from Germany, that looked pretty good from the outside, and was well shielded. However, on the Q meter, I saw bad things. The effective Q at 20KHz or so was only about 1 or 2, as I recall. This meant that the effective resistive noise at high frequencies went up several times over its DC value. Not good for tape recording. This is due to the magnetic materials and the THICKNESS of the laminations that made up the core assembly. They used 6-10 mil laminations as I recall, AND I could easily see the problem with my Q meter.
This was a VERY GOOD reason to both improve our tape heads AND remove the transformer, if possible, from magnetic recorders.
Getting back to phono, this problem would not be so pronounced, but it is still real.
For phono, another problem becomes VERY SERIOUS. This is the ultra low frequency garbage put out by the record warps and passed through the transformer. This could be deadly, and what I will look at, primarily, with my transformer sample.
Initially, the challenge of making a transformer eliminator to make direct connection with a MC cartridge possible, was considered almost impossible. In fact, I was virtually 'shown the door' at Ortofon in 1967 for even considering it. It had been 'proven' by their best engineers that it was impossible, I was told.
Well, it wasn't impossible, just impractical. With the JC-1 released by Mark Levinson in 1973, I got just revenge.
It must be said that at that time, the transformers used, were inferior to what we can get today. Now, we have true competition, given that low noise transistors and jfets are getting rare and expensive, and transformers are improving.
It was really low frequency distortion, and the addition of high frequency noise due to eddy current losses in many commercial transformers, that led the way. While the RIAA tended to roll of the high frequency noise from both the record or the transformer, the NAB or IEC tape equalization did just the opposite.
This is where I found myself, back in 1968, looking at the transformer noise problem.
One way to evaluate this is with a Q meter that measures the ratio of inductive impedance to the EFFECTIVE resistance of the transformer with frequency. I used a GR unit for my measurements. The main transformer that we were looking at was a 10:1, or so, Beyer transformer from Germany, that looked pretty good from the outside, and was well shielded. However, on the Q meter, I saw bad things. The effective Q at 20KHz or so was only about 1 or 2, as I recall. This meant that the effective resistive noise at high frequencies went up several times over its DC value. Not good for tape recording. This is due to the magnetic materials and the THICKNESS of the laminations that made up the core assembly. They used 6-10 mil laminations as I recall, AND I could easily see the problem with my Q meter.
This was a VERY GOOD reason to both improve our tape heads AND remove the transformer, if possible, from magnetic recorders.
Getting back to phono, this problem would not be so pronounced, but it is still real.
For phono, another problem becomes VERY SERIOUS. This is the ultra low frequency garbage put out by the record warps and passed through the transformer. This could be deadly, and what I will look at, primarily, with my transformer sample.
Initially, the challenge of making a transformer eliminator to make direct connection with a MC cartridge possible, was considered almost impossible. In fact, I was virtually 'shown the door' at Ortofon in 1967 for even considering it. It had been 'proven' by their best engineers that it was impossible, I was told.
Well, it wasn't impossible, just impractical. With the JC-1 released by Mark Levinson in 1973, I got just revenge.
It must be said that at that time, the transformers used, were inferior to what we can get today. Now, we have true competition, given that low noise transistors and jfets are getting rare and expensive, and transformers are improving.
WRT HF noise and Q, did you adjust the transformers response to
perfectly flat in the passband and Bessel rolloff with proper transient
response?
Have you tried the Jensen JT-346-AX. It has about the lowest LF distortion
I have seen from a standard MC TX .
T
Last edited:
Assuming you have understood his paper better than me, I ask my question again with different numbers:
Can a transformer ACCURATELY pass on the signal change from (say) 1.9 x 10pr9 to double that, 3.8 x 10pr9?
There must be a lower limit where a traffo simply stops working accurately, or doesn't work at all.
Any thoughts? Any theory? Any tests?
Regards, Allen
- Status
- Not open for further replies.