Pros and Cons of ... no, not Hitchhiking ... SUTs vs. headamps!

[nigel pearson]

If the PU maker had a dedicated transformer this idea might go somewhere. If the final voltage was about 10 mV I dare say dynamic range would be optimum. There is also a chance an ideal supersonic output could be arrived at. Otherwise this looks doomed. I tried to make the 1960's Tobey and Dinsdale phono stage. It looked so interesting until I realised in no way could it be a universal design. Denon DL110 was the star device with it.

The thing to realise is the cartridge maker could make the SUT device work. We only have luck as our engineering input.

I have always been tempted to unwind a AT95 to see if it could make a fake MC. It should do 90% the same things, about 500 uV output. Stereo separation I suspect would be worse. B&O MMC 20 was based on that idea although retained good output. B&O thought MC was about people liking distortion. After research they found people prefer perfection! AT95 has low distortion and was Linn K5/9/18, the latter using the Linn Troika diamond.

 

[andyr]

Let's be clear:
- SUT's increase C-loading, however not the way you stated initially;
- SUT's decrease R-loading, giving the opportunity to load MC carts correctly.
I am having no problems with promoting headamps over SUT's; it becomes nasty when false statements come into play.

Absolutely, Pieter - let's be clear:
- SUTs increase C-loading. I'm afraid I'm not with you when you say I stated it differently, initially? All I have been saying is that a SUT multiplies the capacitance present at the input of a phono stage.

- SUTs decrease R-loading - resulting in some carts being fed a sub-optimal R-load (ie. far too low). This can be alleviated (still increasing the cart's output) by using a headamp.


Regards,
Andy

 

[abraxalito]

It so happens I have been going through various posts and articles that I have collected over the past 30 years and I found an email from Jim Hagerman from 2003, which talked about the effect of a SUT:
* not only does it reduce the value of the phono stage's input R-load by the square of the turns ratio but
* it also increases the phono stage's input C-load by that same factor!

So a phono stage with a 100pF input capacitance presents itself as a 0.01uF capacitance load, in the case of a 1:10 SUT. Not all carts would like this.

What Jim Hagerman says is quite correct, the capacitance goes up as the square of the turns ratio. In practice the capacitance seen by the cart is going to be higher because you've neglected the self-capacitance of the SUT's secondary winding, which for an X32 step-up ratio is quite likely to exceed the 100pF input capacitance of the phono stage.

Then there's the issue of the SUT's leakage inductance which has a low-pass effect, the magnitude of which depends on the loading on the secondary.

 

[andyr]

What Jim Hagerman says is quite correct, the capacitance goes up as the square of the turns ratio. In practice the capacitance seen by the cart is going to be higher because you've neglected the self-capacitance of the SUT's secondary winding, which for an X32 step-up ratio is quite likely to exceed the 100pF input capacitance of the phono stage.

Thank you for your post, abraxalito. So I am not "imagining things"!

Then there's the issue of the SUT's leakage inductance which has a low-pass effect, the magnitude of which depends on the loading on the secondary.

OK ... so what is the effect of this "leakage inductance"? It's not something I've come across before.

Andy

 

[martyh]

What Jim Hagerman says is quite correct, the capacitance goes up as the square of the turns ratio...

I could not find anything in RDH so I set up a sim in LT spice and proved it to myself. Still don't think it matters much though, looks to me like any effect with the exception of minor HF phase shift in extreme cases would be well above the audio band.

 

[pieter t]

a quality SUT performs flawlessly!!

What Jim Hagerman says is quite correct, the capacitance goes up as the square of the turns ratio. In practice the capacitance seen by the cart is going to be higher because you've neglected the self-capacitance of the SUT's secondary winding, which for an X32 step-up ratio is quite likely to exceed the 100pF input capacitance of the phono stage.

Then there's the issue of the SUT's leakage inductance which has a low-pass effect, the magnitude of which depends on the loading on the secondary.

I have to stand corrected.....
When using a SUT, the source impedance for the preamp input is the combined impedance of the source impedance of the cart, DC resistances of windings/cables, and step up ratio.
The input capacitance of the preamp is fixed.
Tonight I did some measurements with one of my 1:10 SUT's.
DCR of the primary is 2 ohm; DCR of the secondary is 124 ohm.
I created a 19.4 ohm "cart impedance" by shunting the output of the 50 ohm signal generator with a 31.6 ohm resistor.
The combined source impedance for the preamp therefore is 10² x (19.4+2.2) + 124 = 2284 ohm.
This 2284 ohm source impedance will "see" the input impedance/capacitance of the preamp.
I aimed for - 3dB at 150 kHz to see how much capacitance is allowed to reach this bandwidth.
Input of the oscilloscope is done with 47k resistor to mimic the phono preamp; normal cable (80 pF capacitance) is used, which would give some 100 pF of total input capacitance.
To maintain the -3 dB bandwidth at 150 kHz, I could add another 330 pF of capacitance.
The total of capacitive load therefore was some 430 pF.
Calculation of load capacitance with a 2284 source impedance is 465 pF for 150 kHz.
The remaining capacitance will be caused by the SUT.
430 pF of input capacitance and a 150 kHz bandwidth is a safe margin to guarantee flawless operation of this SUT in normal conditions.

We may disagree on the necessity of higher than "normal" load resistance for MC carts.
A quality SUT however is uncompromized wrt bandwidth and load capacitance.
Absence of noise, and the option to connect the cart in balanced mode without the necessity of balanced preamps, are other PRO's.
Discussion of sound quality of SS vs tube and headamp vs SUT does not make much sense as personal preferences prevail.

 

[pieter t]

All I have been saying is that a SUT multiplies the capacitance present at the input of a phono stage.
Regards,
Andy

A SUT does not "multiply capacitance"; rather it multiplies source impedance so that less capacitance is allowed for a certain bandwidth.
At the input of a phono stage the source impedance of the MC cart is magnified by the SUT (function of step up ratio and DC resistances - see above).
So, in my measurement, a 20 ohm source impedance of the cart turns into some 2280 impedance by using the SUT. That magnified impedance will allow a certain amount of load capacitance to make the setup function properly.
Normal phono preamps present capacities which can be perfectly dealt with using quality SUTs.
When using headamps the load capacitance can be much higher, but it is not an issue.

 

[martyh]

Pieter, I'm curious now... I set up an ideal transformer in ltspice and it simmed like I replaced the primary with a capacitor scaled by the inductance ratio. Does that jive with your conclusions above?

 

[abraxalito]

Thank you for your post, abraxalito. So I am not "imagining things"!

Thanks to you too for a stimulating thread

OK ... so what is the effect of this "leakage inductance"? It's not something I've come across before.

The leakage inductance is the source of the bandwidth limitation of any transformer. If a transformer had no leakage L its bandwidth would be effectively infinite. On LTSpice the leakage inductance can be modelled simply with the parameter 'K' for describing the coupling coefficient of coupled inductors. In conjunction with the the load resistance the leakage inductance forms a first order LR low pass filter.

In practice leakage inductance is minimized by keeping primary and secondary windings as closely entwined as possible. To maintain a 150kHz -3dB point into a 47kohm load (with zero capacitance) the leakage L would need to be kept under 150mH, a not very difficult target to meet. Which could be why you've not heard of this However the leakage inductance does form a resonant circuit in conjunction with the load capacitance and may well need some damping (RC or zobel network) to keep under control.

 

[nigel pearson]

It might be that some loss at 150 kHz is no bad thing. Many MC's peak somewhere up that end of things, can be a large peak. Far better the transformer is doing it rather than the 2uS passive filter some add to all active circuits.

MC PHONO CARTRIDGE SUT TRANSFORMERS

 

[pieter t]

Pieter, I'm curious now... I set up an ideal transformer in ltspice and it simmed like I replaced the primary with a capacitor scaled by the inductance ratio. Does that jive with your conclusions above?

To check my measurement I removed the SUT.
With the same source impedance and 47k load I could add some 45 nF of load capacitance to get the same -3 dB @ 150 kHz bandwidth.
That confirmed that the measurement was correct (load capacitance an order of 10² higher without the SUT).
When simulating be sure that the inductance ratio for a 1:10 SUT is 1:100.
My SUT has 3,6 H of primary inductance, and therefore 360 H of secondary inductance.

 

[pieter t]

The leakage inductance is the source of the bandwidth limitation of any transformer. If a transformer had no leakage L its bandwidth would be effectively infinite. On LTSpice the leakage inductance can be modelled simply with the parameter 'K' for describing the coupling coefficient of coupled inductors. In conjunction with the the load resistance the leakage inductance forms a first order LR low pass filter.

In practice leakage inductance is minimized by keeping primary and secondary windings as closely entwined as possible. To maintain a 150kHz -3dB point into a 47kohm load (with zero capacitance) the leakage L would need to be kept under 150mH, a not very difficult target to meet. Which could be why you've not heard of this However the leakage inductance does form a resonant circuit in conjunction with the load capacitance and may well need some damping (RC or zobel network) to keep under control.

Leakage inductance is just one source of bandwidth limitation in real world transformers.
When primary and secondary windings are closely coupled leakage is minimized, but capacitance maximized.
Only with 1:1 transformers (interstage transformers for example) the close coupling comes without penalty; actually the bandwidth is very good because of the capacitive coupling.
Closely coupled transformers with other winding ratios will show HF loss because of capacitance; in that case the challenge is to find the best compromise in how closely primary and secondary are coupled. Output transformers are an obvious example.
Next to capacitance, source impedance will greatly influence bandwidth of any transformer (the lower the better).
I measured leakage inductance of the 1:10 SUT: when shorting the secondary and measuring the primary, leakage inductance is some 10 uH.
Shorting the primary and measuring the secondary some 20 mH.
Pretty good results . This SUT is as good as resonance free.
Enough about transformers as this is not the thread topic

 

[andyr]

Enough about transformers as this is not the thread topic

I assure you, Pieter, as the guy who started this thread I have absolutely no problems with detailed discussions of transformers being included in the discussion.

But so far all it has done is confirm my belief that a headamp is a better way (than a SUT) of increasing the signal voltage of a LOMC to match the signal required by an MM phono stage.

I am very interested in your comment "Only with 1:1 transformers (interstage transformers for example) the close coupling comes without penalty".

It is very tempting to use a 1:1 transformer in place of a coupling cap to, for example, stop DC offset passing from one gain stage to another in a phono stage. Caps all have their own sound and can be amazingly expensive (Jupiter copper, anyone! ) and I think an interstage transformer should:
* be cheaper, and
* not have any intrinsic 'sound'?

I have read, though, that to be able to use a transformer, you need to satisfy some rigorous conditions, in terms the Zout of the preceeding gain stage and the Zin of the following gain stage ... which are difficult to achieve. Am I correct on this?


Thanks,
Andy

 

[abraxalito]

I am very interested in your comment "Only with 1:1 transformers (interstage transformers for example) the close coupling comes without penalty".

Me too, I couldn't understand this either.

It is very tempting to use a 1:1 transformer in place of a coupling cap to, for example, stop DC offset passing from one gain stage to another in a phono stage.

I've not studied it but I believe Lynn Olson designed an amp which used transformers for interstage coupling rather than caps. The other advantage of an interstage transformer is the re-referencing of the ground - the input side can use a totally different ground from the output. Try doing that with a cap You really don't want to put a DC offset into a transformer so either you'll still use the cap or gap your transformer so it doesn't saturate or devise some other clever ruse to get around this.

Caps all have their own sound and can be amazingly expensive (Jupiter copper, anyone! ) and I think an interstage transformer should:
* be cheaper, and
* not have any intrinsic 'sound'?

Transformer specialists do talk about the sound of different core materials so I dunno about the latter point. The former is true if you DIY your own transformer, if you buy off the shelf then trafos can be quite expensive too. Depending on core material and if its shielded or not.

I have read, though, that to be able to use a transformer, you need to satisfy some rigorous conditions, in terms the Zout of the preceeding gain stage and the Zin of the following gain stage ... which are difficult to achieve. Am I correct on this?

You really want a low Zout (not hard to achieve with transistors) to drive the primary. If the Zin is high you may need a damping network on the secondary.

 

[andyr]

Thanks for your comments, abraxalito.

You really don't want to put a DC offset into a transformer

Aah, OK - well, that kills that idea!

Transformer specialists do talk about the sound of different core materials so I dunno about the latter point. The former is true if you DIY your own transformer, if you buy off the shelf then trafos can be quite expensive too. Depending on core material and if its shielded or not.

Mmmm, OK. So there's not really any 'advantage' in using a trafo instead of a cap!


Thanks,
Andy

 

[martyh]

To check my measurement I removed the SUT.
With the same source impedance and 47k load I could add some 45 nF of load capacitance to get the same -3 dB @ 150 kHz bandwidth.
That confirmed that the measurement was correct (load capacitance an order of 10² higher without the SUT).
When simulating be sure that the inductance ratio for a 1:10 SUT is 1:100.
My SUT has 3,6 H of primary inductance, and therefore 360 H of secondary inductance.

I think perhaps there was not enough control of parameters for you reach a conclusive answer re transformer on capacitive load with the experiment you set up. Ideal transformers are only available in simulation. Either way though the practical answer is the same as I see it, non issue with any practical MC cart.

 

[pieter t]

Me too, I couldn't understand this either.

Think of capacitors...

 

[pieter t]

Aah, OK - well, that kills that idea!

No, it does not kill that idea.
When there is DC the transformer must be made to cope with it.

 

[pieter t]

Mmmm, OK. So there's not really any 'advantage' in using a trafo instead of a cap!

Oh that depends on the application!
Transformers can do things that capacitors can't do.

 

[andyr]

When there is DC the transformer must be made to cope with it.

I'm all ears, Pieter ... how does one do this?. I'm sure without a coupling cap, DC offset will be small ... but it will be present.

Andy