SUT Loading and Phase

the road not taken.
Post Reply
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

SUT Loading and Phase

Post by dave slagle »

A few years back I was playing with the ideal transformer model in spice trying to match the simmed response to the measured response and I always turned off the phase plots to keep things simple. Recently I revisited the spice file and this time the phase behavior hit me over the head like a ton of bricks and made me wonder:

Could it be possible that I don't like the sound of loaded transformers because the loading shifts phase?

This is more of a generic question that applies to all voltage transformers but I'll stick with SUT's for the moment.

This is the simmed Frequency Response behavior of a 1:10 SUT with varying Secondary Loads. The cartridge is assumed to be 2 ohms and the load is swept to look at the effects on the transformer ringing. ("U" means unloaded)
Image

In looking at this picture the average person would think that using a 1K5 loading resistor would damp the resonance and give a -1dB point of 35Khz. Furthermore given the 1:10 ratio, It would reflect back 15 ohms to the cartridge. When I was originally doing these sims, the cartridge being designed for was the 2 ohm Lyra Olympos and the Owners always seemed to end up with a 10-15 ohm load for optimal sonics.

Things get really interesting when you look at the effects of the loading on the phase behavior of the SUT's Below are the Phase Response plots for the above Frequency Plots.
Image

suddenly if you care about Phase accuracy at high frequency, the 1K5 load isn't quite as appealing with 15 degrees of shift at 10Khz. I have been doing a bunch of actual frequency sweep measurements lately and along with the amplitude comes the phase response so I though it would be interesting to see if the numbers in the sim world can be made to match the real world. Since the top end limit of my sound card is 90K, I needed a SUT that rang below that so I quickly wound something that I knew would fit the bill. I did fairly well (or should I say bad) on my first attempt. (FWIW, I would never let a SUT that behaves like this make it to a customer but i needed something with improper behavior well below my 90Khz max of my soundcard)

I'll come clean here and mention that I did wind and measure the transformer first and then adjusted the virtual parameters in spice to match. (2/U/10K = 2R source/ no primary load / 10K load)
Image

The overall magnitudes are a bit off, (probably the lack of core losses in the sim) but the pattern looks to be just about right. The big question is what haapens when we look at the phase and.....
Image

I'm chalking up the occasional hiccups in the plots to lack of samples but the overall picture sure looks close to what the sims suggest would happen so I think there is at least a valid observation that secondary loading introduces premature phase shift. The question becomes could this be at least partially responsible for the reports by some (myself included) that loading of a voltage stage robs the life from the music.

dave
Last edited by dave slagle on Sun Feb 07, 2010 10:12 pm, edited 1 time in total.
shinebox
Posts: 88
Joined: Fri Jan 28, 2005 9:44 am

Post by shinebox »

(slaps head)
Many many years ago I sim'd an SE output stage deliberately playing with the cathode bypass cap (to introduce an LF resonance) and variable damping thereof to "extend" the LF response. Whilst I could get it reasonably flat, I didn't like what the phase was doing. Similar arguments to parafeeders bass boost?

I guess similar considerations apply to reflex bass tuning and the resulting group delay...

Should have remembered about the phase! It all seems so obvious now you've figured this aspect out... sounds very plausible to me as an explanation of the sonic effects.

cheers
jlsemrad
Posts: 15
Joined: Fri Apr 17, 2009 10:56 pm
Location: Dallas, TX

Post by jlsemrad »

Dave, what is the nature of the effect of the primary DC resistance and inductance of an unloaded-secondary SUT? In other words what kind of load to the cartridge does the primary winding present?

TIA,
John
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

It provides mostly an inductive load. With the secondary unloaded higher DCR's don't fare as bad as they do when the transformer is loaded. (by higher DCR's I mean those that approach, exceed or in many cases double the cartridge impedance.) There is a certain trend for using high ratio vintage mic transformers with High DCRs' for phono SUT's. A simple look at some of the numbers involved quickly shows that the theoretical 30dB of gain from a 1:30 only nets you 24dB when loaded.

dave
jlsemrad
Posts: 15
Joined: Fri Apr 17, 2009 10:56 pm
Location: Dallas, TX

Post by jlsemrad »

Yes, the HA100X has a primary DC resistance of about 55 ohms. That seems like plenty for a low impedance cartridge like an Airtight or Haniwa.

John
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

55 ohms seems really high. I like my DCR's to be a fraction of the cartridge impedance.

dave
corbingravely
Posts: 1
Joined: Wed Jan 29, 2014 6:12 am

Post by corbingravely »

Have you actually taken the harmonics into account? Because some harmonic movements can shift the wave out of phase by a sheer 30 degree and that is a lot when considered in the case of a transformer.








---------------------------------
CORBIN GRAVELY
walge
Posts: 36
Joined: Tue Nov 24, 2020 8:42 pm
Location: roma

Post by walge »

I am reading this old thread.
On nr. 400 of Audioreview we have spoke about the stp up ) in that case the 999 Tango) with different load on secondary.
Very interesting resiults, lot of test lab.
Then I buil the one in photo with a custo trafo, 12.20 and 40 ratiod, made for me by Sowter
There is a Elma swith, 24 position and 48 Holco H8 resistor.
The range of load is wide and on listening test the resuslts are easy to find.


Walter
Attachments
back of Triple
back of Triple
Triple_Retro_1.jpg (517.23 KiB) Viewed 11225 times
Triple front
Triple front
Triple_Fronte_1.jpg (510.83 KiB) Viewed 11226 times
1:10 with Zsource 40 ohm; secondary close to 47k
1:10 with Zsource 40 ohm; secondary close to 47k
1_10_40ohm.jpg (51.5 KiB) Viewed 11226 times
same before, ratio 20, source 10 ohm
same before, ratio 20, source 10 ohm
2_20_10ohm.jpg (48.36 KiB) Viewed 11226 times
ratio 40, source 2,5 ohm
ratio 40, source 2,5 ohm
3_40_2_5ohm.jpg (52.12 KiB) Viewed 11226 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Your plots do not show the phase and what would be interesting is to show the plots with phase of the same setup at various load values. In general the sonic results of secondary loading are incorrectly attributed to the load presented to the cartridge when the measured frequency is altered just above the audio band and the phase is altered well into the audio band.

Put simply, when you adjust the load on the secondary of the transformer you often alter the sound of the transformer and not that of the cartridge.

dave
Get Your Fix
www.hifiheroin.com
walge
Posts: 36
Joined: Tue Nov 24, 2020 8:42 pm
Location: roma

Post by walge »

About phase is not so important to know, for me .
But I check on my tests. Knowing the stuff from Sowter I am not worried.
In my opinion much better the square wave response.
Regarding the sound we have to consider that changin th load seen by cartridge we also change the damping
The sound you can ear changin the load , in my opinion, is not related mainly with the trafo.
Are a little "nuance" and when you get a gold spot moving the switch of the load the difference with other position are little.
We made some tests with my friend who has 105 pcs of MC heads.

In attach some square test at 1:20
This is mainly a joke but interesting.

Walter
Attachments
1:20 square 1khZ
1:20 square 1khZ
03_20_square_1K.jpg (38.56 KiB) Viewed 11210 times
1:20 square 10 kHz
1:20 square 10 kHz
03_20_square_10Khz.jpg (40.8 KiB) Viewed 11210 times
1:20 square 100hz
1:20 square 100hz
03_20_square_100Hz.jpg (38.84 KiB) Viewed 11210 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

10K square waves can be made to look pretty by network loading but rarely have I found that the overall sound is improved. I would call it more of a lateral move. The rise time of the 10K simply shows the limited high frequency extension that can be seen in your frequency plots.

Back in the day WE did a lot of experimenting into voice intelligibility and found that the phase linearity in the "voice region" 600Hz-4kHz was more important that amplitude linearity. I have found that this concept directly relates to the "sound" of transformers. The Peerless 4722 when used as a SUT driven from a low Z cart into a high impedance (200kΩ) has a bite to it. This can be attributed to the 6dB peak @ 20kHz due to the lack of damping. This is typically fixed by applying a secondary load which will give a response similar to your plots in the amplitude realm and also start to introduce substantial phase shift in the "voice range" which is what I see missing from your plots. The loaded or critically damped 4722 starts to get murky and starts to get thick sounding. In the case of the 4722 you are simply choosing a load for the best sonic compromise for a given system. In order to get a feel for the big picture of SUT behavior, I find frequency and phase response plots both loaded and unloaded are quite telling. The unloaded frequency and Q of the HF resonance being hidden by the load is quite telling about the ultimate sonic signature of a transformer. We all know any transformer can be made to behave... What I find more telling is how it misbehaves. I do not find an unloaded high Q resonance within an octave of the audio range acceptable but a low Q fairly benign resonance 2-3 octaves out is a completely different situation. The engineer feels the need to treat both situations the same but my ears tell me to not bother with the former and not to worry about the later.

dave
Attachments
Screen Shot 2021-02-01 at 2.12.12 PM.png
Screen Shot 2021-02-01 at 2.12.12 PM.png (238.06 KiB) Viewed 11205 times
Get Your Fix
www.hifiheroin.com
walge
Posts: 36
Joined: Tue Nov 24, 2020 8:42 pm
Location: roma

Post by walge »

This is the reason I look mainly on square wave test.
On 10 kHz test the wave is quite perfect without any ringin.

Each time you see a square with some ring at the end the sound seems to be althered.
Also it is very difficult to get the linearity as shown on diagram of three ratios
The response of the Triple is perfectly comparable with Tango 999 in freq. response and square ( I have done a lot of tests) and the sounds is quite similar .
The beautiful aspect of 999 and Triple is to have three ratios with high quality so you can choose the best for your cartridge.
In attach some test done with 999
I have tested lot of trafos, many of them are very fine but in most case you can't play.

The test on Altec are done by yourself?

Walter
Attachments
Inside the test box
Inside the test box
img_4612_173.jpg (131.11 KiB) Viewed 11192 times
The test box with 999 and switches
The test box with 999 and switches
img_4590_919.jpg (200.79 KiB) Viewed 11192 times
1_20_143.jpg
1_20_143.jpg (77.44 KiB) Viewed 11195 times
Tango 999, 1:20 square at 10kHz
Tango 999, 1:20 square at 10kHz
Figura 5C_20_10k_square.jpg (61.34 KiB) Viewed 11203 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I see multiple ratio transformers as a fools errand. It is like a swiss army knife that is great to get you out of a bind but not something you would use for fine woodwork.

In audio test measurements are a double edged sword. In the LCR thread you commented that people only publish macro level details and suggest that the truth may lie in the micro detail measurements of the inductors. Now you seem to be advocating for just the opposite and saying the perfectly loaded 10K square wave tells you everything.

From a marketing and sales perspective the one size fits all at an inexpensive price is the holy grail. Years ago I realized that the shiny grail was made of gold plated lead.

dave
Get Your Fix
www.hifiheroin.com
walge
Posts: 36
Joined: Tue Nov 24, 2020 8:42 pm
Location: roma

Post by walge »

Each of us has the proper considerations.
My idea is that the test lab on stuff are the only way to understand the way to follow.
The multitap trafos are the best way to reach the results for who has different MC, in general.
But, when you get the results as Tango or Sowter you can be safe about sonic performance.
After this, the multitap trafo with high performances are difficult to build.
Regarding the square response in my opnion the result I sent are fine and difficult to find.
And I said "mainly" not "everything"
Regarding LCR soon I will complete a test set on lab around this method and around some circuit

Walter
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

The multitap trafos are the best way to reach the results for who has different MC, in general.
My belief is it is the cheapest way not the best way. The best way is to build a SUT to match a specific cart to a specific situation and this transformer can work for cartridges with similar specifications. If a second cartridge with different specifications is in your arsenal then a different transformer should be used.
But, when you get the results as Tango or Sowter you can be safe about sonic performance.
Without seeing the unloaded behavior and the phase behavior I do not consider the information presented sufficient enough to make any judgements. For the record I do not consider response that is -10dB @ 100kHz acceptable and this can be seen in both the frequency and square wave plots you have presented. I suspect that if measured the phase response will show substantial deviation of phase well into the audio band and the result will be a device with a thick midrange. Playing with the secondary loading allows for the sound of the transformer to be altered and a compromise can be chosen between a thick dull and a bright forward sound. I consider this method of tuning a band aid for using an improperly designed device for the situation.

I made the observation of the phase difference with loading more than a decade ago and the measured and sonic results have remained correlated. More recently I became aware of the Bell Labs work in this area and it does serve as the most plausible explanation I have found to date.

dave
Get Your Fix
www.hifiheroin.com
walge
Posts: 36
Joined: Tue Nov 24, 2020 8:42 pm
Location: roma

Post by walge »

Hi

the facts are simply.
Do you have the lab test that can confirm your point of view?
If yes, very fine for me to learn something new.
And I will do the best to get a better results.
And I am ready to buy the stuff you show me better than Tango and Sowter ( at the same ratio)

In addition all the stuff ( Tango and Sowter) were tested live with some high level configuration very difficult to find ; when the best 10 MC head available are connected to them the sonic results were very fine.
Of course other step up were used.

Your consideration about sonic performances seems to be mainly theoretical.
If you want I can send you a Triple for tests (lab and listening). Let me know.
In every case the Triple is not on market at the moment and the type is made for me by Sowter.
I am waiting to complete one other phono project then, maybe, I will start with market.
Just after the articles on Audioreview magazine.

Walter
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

My design choices are driven first and foremost by what sounds best which is a very subjective thing. Once a pattern of good sound emerges, I then look for measured explanations for the "Why". I do find measurements valuable to show minimally acceptable function but aside from phase I have yet to see them as a good predictor of sound.

My general philosophy including measurements are in the link in the message at the top of the forum.

http://www.intactaudio.com/forum/viewto ... =7629#7629
Get Your Fix
www.hifiheroin.com
walge
Posts: 36
Joined: Tue Nov 24, 2020 8:42 pm
Location: roma

Post by walge »

Hi
in attach the test of Triple at 1:10 ratio with 100k, 47k and 27k load.
There are three lines that are perfectly equal. It is difficult to see!!
I will try to arrange a setting for higher load for trafo
The reponse is fine, in absolute.
Compared with Altec in much better

At 20 hz the shift is +2,2°
At 100Hz is +0,43°
At 10kHz -7,39
At 20 kHz -15,9
The reference is 1 kHz

(The graph on pdf has 5 k/division but on screen of AP are 2k/division, mistery! probably a setting, I have a look)

let me know if the test set is correct in your opinion


Walter
Attachments
Triple_10_in_out_phase_100k_47k_27k.pdf
phase response
(28.22 KiB) Downloaded 165 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

That graph does not look correct.

attached below is a simulated response of what the loading does. All three of your lines overlapping suggests some sort of error. Also having frequency and phase on the same plots and showing behavior down below 1kHz also helps. Seeing the unloaded behavior and the severely loaded behavior also is important. In the very first post in this thread I show the simmed and measured response for an actual device.

I must say I am beginning to wonder why we are having this discussion since I am a manufacturer and you are looking to sell a product with one of my competitors devices. I think you should be getting the support you need from your winder.

Dave
Attachments
Screen Shot 2021-02-03 at 11.42.54 AM.png
Screen Shot 2021-02-03 at 11.42.54 AM.png (31.64 KiB) Viewed 11160 times
Get Your Fix
www.hifiheroin.com
walge
Posts: 36
Joined: Tue Nov 24, 2020 8:42 pm
Location: roma

Post by walge »

As I wrote the Triple is not on the market and is not possible to buy.
Then the test lab is fine, this is a real test not simulation.
In addition soon I will send a beautiful diagram done by a particular software (not commercial) that plot the real parameters of coils, resistor and caps.

In every case is not a problem for me to end my post around trafos.

Walter
Post Reply