Phono Preamp. With transformer or fully active

With one SUT you can't use all MC cartridges, such phono stage is not universal for all MC. It will work for some of them nicely, but if your cartridge impedance is very low (2 Ohm) or very high (40 Ohm) you can't use one SUT for both of them.

I'll differ from Ralph and agree with chakster 100% on this one.   The thing everyone misses in this discussion is the transformer needs a specific amount of inductance to assure low end extension based on cartridge internal impedance.  If you take a 2Ω cart and use a SUT designed for 40Ω, the LF extension will be down into and past record warp and tonearm resonance and the low impedance drive will cause an underdamped top end (HF peak in response). Depending how close to the audio band this happens, the results can be problematic.  Going in the other direction and using a 40Ω cartridge for a SUT designed for a 2Ω source typically will not have a full bandwidth bottom end and an over-damped top end.   The general solution to this is to design with enough inductance for the highest impedance cart and rely on a network on the secondary to fix all the other issues that arise from any mismatch.  The other typical solution is to use a dual or tapped primary that allows for different turn ratios to better suit a wider range of cartridges.  

For given source and load characteristics a wide bandwidth transformer can be wound to match the situatu

dave

I might have been a little unclear above when I typed:

I do not consider the use of the rather heavy secondary load in addition to a Zobel as a sonically acceptable solution.

That should read:
I do not consider the use of the rather heavy secondary load in the form of a Zobel as a sonically acceptable solution.

I will add that the load from the Zobel is at higher frequencies only.

Jensen named the parts of the Zobel Rdamp and Cdamp which clearly tells its purpose. I do not know of any phono stage inputs with those parts intentionally in place. It would be interesting to see the response without the Zobel in place to see what it is hiding. I for one do not have a problem with a Low Q 2-5dB peak nearing 200kHz and find that any resistance or network added to tame it does more harm than good.  The 4722 or other vintage mic transformer turned SUT showing the peak in the 20-40kHz range is another situation altogether.

http://www.intactaudio.com/images/SUT%20white%20paper/4722%20vs%20emia.png

dave

Ralph...

Loading the transformer is well-known to control issues in the high end- the 'underdamped' part to which you refer.

I do not consider loading of the secondary of a transformer an acceptable way to control transformer behavior.  Of course there will always be some minimal load and with a known source impedance the transformer can be designed to work into that minimal load.   From a purely subjective viewpoint I have always found that reasonably well behaved unloaded transformers have sounded better than transformers forced into submission by loading.  If you start with a couple of unknowns loading suddenly becomes the only thing left in your toolbox to get acceptable results and that is the way the industry has gone.  

From a measurement perspective it is interesting to look at the HF phase response of an unloaded transformer that rings and the same transformer loaded down for critically damped response.  The link below is for the sweeps of the peerless 4722 driven from a 4Ω source with varying secondary loads.

http://www.intactaudio.com/images/SUT%20white%20paper/4722%20load.png

It is interesting to note that as the frequency response approaches flat at 20kHz the phase shift increases further down into the audio band.  In this case it becomes the choice of the lesser of two evils. Since no two people hear the same, it is no wonder loading of cartridges is such a hotly contested subject when it comes to transformers.  It is my belief that often times when you adjust cartridge loading via the secondary of a SUT, the sound (and measured response) of the transformer is what is heard.  This change in sound is then falsely attributed to the cartridge seeing a different load.

dave

Hello Raul,

I actually posted that "story"on this site back in 2008.
 https://forum.audiogon.com/discussions/thoughts-on-cartridge-loading-with-a-sut

I have some more recent thoughts on the process and measured results at the link below. 

http://www.intactaudio.com/SUT%20design.html

dave

Raul,

A number of years back I always found it odd that it seemed that there were two camps on loading of MC cartridges.  One camp insisted that 47KΩ was the proper number and changing it made little difference.  The other camp was fanatical down to the single Ω precision and it was always a fraction of the 47kΩ number.  It coincided that the 47kΩ crowd all used active gain stages and the fraction faction used SUT's.  If I assumed the 47k-ers to be correct and loading had little sonic effect, it seemed logical that the loaders might be effecting some other change than the sound of the cartridge and I started looking more closely at SUT behavior.  This has lead me to my current belief that when you load the secondary of the transformer you  change the sound of the transformer more than that of the cartridge.  

The wrinkle to both of the situations above shows up when you consider current injection behavior for MC cartridges where the load tries to become a dead short.  This one really bothered me until I read an article by Peter Moncrief in IAR#5 where he makes a convincing argument that loading a MC cartridge does not appreciably change the measured frequency response but it does have a marked effect on how much IM distortion is created.  What I found interesting about this concept involves the general sonic terms used for the for the reduction of IM distortion and two new warring factions were created in my head. 

The "Dampers" use the loading to explain the taming of a rising top end inherent to the MC topology.   A light load causes a rising response, and a heavy load causes the top end to roll off turning things into mud.  Somewhere in between the two one finds a safe middle ground and can live in peace.  A number of years back I split from this faction since the easily measured behavior of the SUT showed this behavior to a far greater extent than the measured cartridge behavior.

The "Loaders" lead by Moncrief simply state that the etched detail of an unloaded cartridge is an excess of IM distortion artifacts and when those artifacts are reduced, the seemingly dull sound is actually correct and the result of a series of system wide decisions previously made to offset the overly forward sound of the unloaded cartridge.

Everyone considers this all to be a black art and in reality it is simply a puzzle where all of the pieces fit together.  When you find two pieces that  seem to join but the pictures do not match you need to keep trying to see if you find a better fit.  Taking parts of truth from all of the factions above I am slowly coming up with a picture in my mind of how this all fits together in a predictable and repeatable manner.  Obviously I make SUT's so I'm going to explore those options.  You prefer an active stage so that is where your biases are.  Tubed vs. solid state are a couple of more of the warring factions.  This is where the subjective results come into play and ultimately we choose and follow the path that gives the most musical enjoyment.  That doesn't mean we shouldn't occasionally check our GPS when we start seeing the occasional polar bear.

dave

In order to add a bit to the road Mike went down his first audition was with two pair of identical 1:20's, one in silver and the other in copper.  Initially the silver was preferred and after a few weeks he found that the internal SUT's to the CS Port offered some benefits over the 1:20's he had in system.  A little bit of sleuthing turned up the distortion spec of 0.1% @ 0.76Vrms output.  Working this number backwards from the 40dB of gain this translates to 7.6mV of input.  The Etsuro Gold has 0.56mV of output and fed to a 1:20 that will be slightly above the 7.6mV level.  It turns out that the CS Ports internal SUT had 3dB less gain which using simple math put the output right at that published distortion number which is why I suggested that mike try a 1:10.  I sent him a copper 1:10 for his 4Ω Etsuro and things improved over the internal SUT.  Replacing the copper with silver was the icing on the cake that he is currently consuming.  

I think the important thing to note about this is the high frequency overload characteristics of phono playback is typically not pretty.  At high frequencies many phono stages seem to have a more abrupt transition into distortion where the higher order artifacts show up and quickly dominate.  This high order high frequency distortion only happens at dynamic peaks nearing the highest possible groove velocities.  Holman notes a worst case 105cm/sec peak @ 7Khz on  Woody Herman Verve V-5885 and that represents 26dB above the standard 5cm/sec velocity referenced to 1kHz.  Granted the 7kHz signal receives 10.7dB less gain but that still leaves a 15dB dynamic peak above the 1kHz baseline.  It is these periodic events tickling the abrupt onset of distortion that I find gives a "something is not right but I'm not sure what it is" type of feeling.  It is the periodic occurrence at dynamic peaks that make it so elusive.   The trend as of late is for MC carts with 6dB or more output than their predecessors which simply translates to a 6dB loss of high frequency dynamic headroom when considering a MC stage built with a 0.2mV SPU in mind.  10 years ago a 1:10 was an anomaly and today with MC outputs in the 0.5mV to 1mV range it is slowly becoming the norm.

dave 

Dover,

This simply means that the Zesto added the network in order to deal with the transformer behavior. That is very different than having a phono with that network at the input and then designing the transformer to work into it. This is an easily solvable chicken or the egg paradox.  My response asking about this was in relation to the quoted exchange below

I do not consider the use of the rather heavy secondary load in addition to a Zobel as a sonically acceptable solution.  

FWIW they *are* designed for that.

I took this to mean that the transformers were designed to work into that specific network which made me question where that network actually exists to require that transformer design to naturally work into it.


You said above:

The reality is that transformers are non linear in both amplitude and more importantly phase. It is the phase anomalies that kill the music - musical timing and natural harmonics are destroyed by phase anomalies.

If the time domain is more important than the amplitude domain then what is the logic of using a network on the secondary that corrects for the amplitude domain at the cost of additional phase shift in the time domain? The link below shows the simulation of a hypothetical transformer with and without a Zobel network to damp ringing at high frequencies.

http://www.intactaudio.com/forum/files/screen_shot_2020_08_02_at_81425_am_139.png

If one choses to use a loading network and likes the sonic results, that is fine. I have always found that sonically, loading is the worst thing you can do to a transformer and using anything more than the absolute minimum required is relying on a band-aid for a preventable injury.

dave

I do, it retails for $3500 and has had stellar reviews.

Which phono stage is this?  

I agree with you on the phase issues and at some low and high frequency a transformer is going to shift phase and I think that the shift at high frequencies is much more of a concern than at  low.

 musical timing and natural harmonics are destroyed by phase anomalies.

yes....  bell labs found a strong correlation between phase shifts and voice intelligibility.  Higher order odd harmonics are dissonant to begin with, throw in some phase shift and i suspect they get downright ugly.

 Look at the bottom end in your graphs - its there for all to see.

I don't follow.  transformers like all systems have a finite bandwidth and with that they have finite linear phase.  It is much more of a struggle to  get a full 10 octaves of flat phase response out of a transformer than an active circuit so it becomes increasingly important to place the plateau of linear phase in the proper spot.

dave

Ralph,

Cartridge loading is another hotly contested subject that is best to save for a new discussion.  It seems we both agree that applying a load through the secondary of a SUT has a much greater chance of changing the SUT behavior in an audible way than the cartridge behavior.

The inductance question can be answered by looking at SUT behavior when driven from various cartridge impedances.  The plot linked below is the 4722 wired 1:18 driven from 2Ω to 100Ω and at 100Ω the -3dB point is ~10Hz which translates to 1.6Hy.

http://www.intactaudio.com/images/SUT%20white%20paper/4722%20cart%20z.png

I chose the 4722 because it is a known device that represents the typical behavior of a microphone transformer pressed into SUT usage.  I agree that the top end behavior of this device leave a lot to be desired. When paired with an EMT or a Denon 103 into a 47K load it is -1dB @ 10Hz and 20kHz and has what many call a "vintage" sound.  As an aside, Bell Labs did a lot of work on voice transmission and intelligibility and found that linearity in the phase domain was far more important than linearity in the amplitude domain.  They considered the range from 600hz to 4Khz the critical range and found that phase shifts creeping down below 4kHz tended to muddle the voice.  I agree that the audio bandwidth is much larger than that required for voice but it is interesting to note that the phase shift of the 4722 does creep down into that "sacred area" that Bell Labs found had an impact on intelligibility.   I just measured some 900:900 Western Electric repeat transformer that had a specified frequency range of "Voice" and 425Hz-1615Hz was mentioned.  When fed from a 50Ω source and left unloaded the -1dB points were 18Hz and 2.4Mhz.  I don't think the crazy high top end bandwidth was an accident and the phase stays flat to 100Khz.

http://www.intactaudio.com/forum/files/phase_385.png 

18:1 is a bit on the high side for a 103 but not terrible.  Problem with the higher step up and the 40Ω cart is any capacitance on the secondary claims the 20kHz+ info quickly.  The 103 is a 0.3mV cart and into a 1:18 loaded with 47K you get really close to the 5mV "Norm" for most mm inputs.  Now 1:70.... that is high... but also a perfect fit for a 0.05mv cart into the right cable / pre combo.

I don't see where the inductance comes into play here.  as long as there is sufficient inductance to cover the low end without going too low it falls out of the picture.  

An interesting anecdote about the 4722 and its ilk which can be wired 1:18 / 1:36 is you get just about the same output due to the 1:36 loading the cartridge below its internal impedance and the frequency / phase plots are quite similar with the 1:36 losing a little more than 1/2 a dB @ 20Hz & 20kHz.  I know of a few people who simply prefer the 1:36 hookup and given the similarity of the measured response of each it makes me wonder if any of this preference is due to the drastic load improving traceability ala Moncrief.  I know this seems in contradiction to what I said above but in this case the loading in both cases is right where the transformer was specified (ie 50K load) and the measured responses are similar so I see a window of possibility.  An unfair test would be to load the 1:18 setting with an additional 16K on the secondary and compare it to the 1:36 with a typical 47K.  The proper way to test it would be to simply parallel 55Ω with the primary of the 1:18.  The gain would be down around 4-5dB but the frequency response would minimally change.

http://www.intactaudio.com/images/SUT%20white%20paper/4722.png'

dave

I specifically chose the 4722 to avoid the appearance of throwing another manufacturer under the bus so to speak.  The only place I have seen signal level have an effect on frequency response in SUT's was with units wound on cores with poor low level permeability in which case the bass disappears. 

I'm not sure of the origins of the cult of the 4722 but I do think you can do much better.  The jensens are nice but I do not consider the use of the rather heavy secondary load in addition to a Zobel as a sonically acceptable solution.  

dave