Here you have the essentials from people who knows how about transformers.
DPA Microphones :: Transformer vs. Transformerless Output
DPA Microphones :: Transformer vs. Transformerless Output
If you only look at numbers without understanding the background...
It is very clear that a transformer can only transform impedances while an amplifier can provide both voltage gain and low output impedance.
If I was a manufacturer of any kind of gear that could both use transformers and amplifiers, I probably would promote what is cheaper for me...
If an amplifier has lower distortion as a transformer, how is that achieved ?
Loads of NFB.
What will be sounding better ?
Last edited:
Resonant transformers are a matter of bad implementation.
If you know everything, why ask for input and not start with the design ?
Perhaps because you hack on everybody else's designs including nonos, tubes, transformers and everything else that does not fit your bill.
It's like "Hey people, look, we are doing this and that and everything else is bad, that's why we do things right."
The law of nature is that transistors are nonlinear and need NFB to correct that, which introduces new problems.
I think you should have a listen to this then:
http://www.dpamicrophones.com/en/Download/~/media/PDF/Download/grandpiano.pdf
I´m not sure they still have stocks of that CD/SACD, but I´ve got one of my own.
DPA microphones are probably "the microphone".
They did 8 identical recordings of Niels Thybo playing a grand piano, miced in 8 different ways with 8 different mics.
The result speaks for itself, TL mics rules, and everyone with experience with transformers knows that. Thus the TL alternatives excists.
NFB is not necessarily bad, as well as it is not necessarily needed, but anyways, the transformer still would be the last choice for me, due to ringing and resonating.
In home audio no one would choose more than 100m of cable between components, which makes transformers completely irrelevant.
On top of that, transformers are an additional component, making the product even more complex and expensive to the customer. It will also prolong the signal path as well as breaking it..
Trying to to transmit signals unharmed through a reactive winding and a piece of iron seems pretty hopeless to me.
Transformers can easily be heard, just as capacitors, but still in a different way.
Well have it your way.
We are trying to build a DAC without any of the components you´ve mentioned, and we are open to suggestions for improvements, which will then be tested.
So far we decided to try out ESS DAC chips and WM receivers, but we will not try to make any non transparent analog stage based on reactive components or tubes, which also automatically outrules NOS DACs.
Hi,
Hysteresis is a POTENTIAL problem in Transformers. It has solutions. The bottom line is that hysteresis would invariably drive measured THD up. If the measured THD of a Transformer is very low (say < -115dB/0.0002% for THD) we may conclude that the transformer is essentially free from hysteresis.
Of course, some pro-audio gear uses iron cored (or large iron content cored) transformers which have hysteresis as this form of distortion is felt by some engineers to be pleasing and to enhance sound quality. I disagree and avoid transformers that have such traits.
Of course, the cost of really good transformers is very high, so in most cases gear cannot be designed to incorporate them, not because they are bad, but because going transformerless has only drawbacks that are not easily measured (though easily heard) and the main advantage of being extremely cheap.
I don't know.
When I worked in the 80's for a small company that made Mixing Desks for the state radio & tv we used transformers on both inputs and outputs, as this was the only to get sufficient EMI Resilience and general noise immunity.
As for sound quality, we evaluated transformers by measurements but more crucially by connecting them "back to back" and simply looping them into the resident studio monitor system (Schulze TH315 Coaxials with matching amplifier, later also MEG RL900) and listened.
The transformers we routinely used for Microphone duties and outputs needed six PAIRS linked together to be audible and we actually figured that the audibility was due to the losses in level from the DCR losses, not due to bandpass behaviour or distortion, as they went away after matching levels.
And yet the most prized microphones among recording professionals (reflected by the price that they fetch) are tubed & transformer coupled Neumann Microphones and their replicas, just as the most prized microphone preamplifiers are the TAB and REDD Tubed and transformer coupled ones and their replicas and the most prized compressors are old Teleonix Tube/Optical ones.
Somehow it seems that not EVERYBODY has yet understood that "transformers are unwanted" and instead evidence tends to take the opposite position, namely that with the exception of a small minority EVERYBODY wants transformers, this includes all those too poor to afford the good (and very expensive) stuff, who may publically dump on anything they cannot afford, while secretely hankering after the lottery win to allow them to buy these exact toys themselves.
Of course. Magnetic tape used Iron Oxide and Iron is subject to a lot of hysteresis. If iron cored transformers are used they should have some bias (can be DC or HF) to overcome this. Good studio grade signal transformers do not use iron cores anyway.
As said, transformers (in this case a 1:1 line type) can be manufactured that have better than -115dB THD&N for 2V/+8dBu Input (THD falling further with lowering levels to at least better than an AP2 can measure) and can handle +26dBu with equally low distortion, this is still WAY BETTER than any op-amp or DAC out there, with a dynamic range of AT LEAST 133dB and likely better (we just cannot measure something with that low distortion with the usual industry standard audio measurement gear).
Well, as long as the best transformers have lower noise, THD, IMD, TIMD, SIMD than the best active circuits I think I prefer to use them instead.
Only by those who favour noise and distortion over the absence of noise and distortion. Honestly.
Of course, just like with active circuits, there are more bad ones than good ones. But if we judge everything by bad examples than we can use nothing and we must even disregard most musicians and orchestras as "not capable of high fidelity" (to the notes in the score).
Funny. The MEASUREMENTS for transformers from the makers of HIGH QUALITY transformers attest that in fact that transformers are MORE linear than active circuits. So I suspect that those who use THESE transformers do not know that Transformers are not linear, simply because it is patently untrue.
Because YOU personally dragged the matter into this thread, kicking and screaming, by simply dumping on everything with transformers, without qualification and in clear and obvious opposition to facts on the matter.
May I suggest that like many of us, myself included you tend towards easy generalisations.
For example I do tend to generalise Cirrus Logic products are overall poor and in my own testing this has almost always been born out (I cannot think of one instance where I ended up adopting a CS product when alternatives existed - they really where always that poor - even if they where cheapest).
However, what I do tend to do much is to qualify statements of sweeping dismissal with the term "in my experience".
For example:
In my experience the CS SPDIF receivers and the CS DAC's and ADC's I evaluated are poor performers sonically and often also in measured performance. This includes the CS4398.
This allows other experiences and tends to avoid excessive counter reactions.
So before dismissing anything you did not happen to include in your design as "useless' and "coloured" a minimal "IME" may be advised. Further, a pretty solid amount of testing may also be in order before dismissing anything, probably more though testing than you have done.
Ciao T
POST SCRIPTVM
Speaking of Microphones, I hope you realise that even the transformerless Microphones include FET's (that is condensor microphones) or tubes. So I sincerely hope you do not use any such condensor microphones with evil Fets that steal low level information for recording.
If you follow your reasoning correctly Ribbon Microphones are also out as they invariably need transformers and of course the iron in the magnet system is subject to hysteresis.
So this leaves only dynamic microphones like Sure SM58 fit for recording, ooops, darn, these have an iron core in the voice coil so there is hysteresis also and you cannot use these microphones as well.
So in the end you cannot use any microphone as they all invariably incorporate something that you disagree with. And of course speakers are also out for listening.
If you are really consequent with your prejudices you can neither record nor play back music, so I suspect you are actually making the DAC to look at sinewaves on a 'scope.
Anyway, this little POST SCRIPVM is intended not to insult you, but to use the method known as "surrealist investigation of claims of the normal" on your claims that Fets are bad, Tubes are bad, Transformers are bad and so on but that what you use is not "bad". If you may, I am practising guerilla ontology, to highlight the hilarity and untenability of your position.
In the buzzing spiky world of reality (such as we perceive and access it anyway), all things are bad in some ways and good in others. The wise man uses the good sides to his advantage and minimises the drawbacks of the bad sides. The fool condemns everything with a bad side and finds himself with nothing.
Hysteresis is a POTENTIAL problem in Transformers. It has solutions. The bottom line is that hysteresis would invariably drive measured THD up. If the measured THD of a Transformer is very low (say < -115dB/0.0002% for THD) we may conclude that the transformer is essentially free from hysteresis.
Of course, some pro-audio gear uses iron cored (or large iron content cored) transformers which have hysteresis as this form of distortion is felt by some engineers to be pleasing and to enhance sound quality. I disagree and avoid transformers that have such traits.
Of course, the cost of really good transformers is very high, so in most cases gear cannot be designed to incorporate them, not because they are bad, but because going transformerless has only drawbacks that are not easily measured (though easily heard) and the main advantage of being extremely cheap.
I don't know.
When I worked in the 80's for a small company that made Mixing Desks for the state radio & tv we used transformers on both inputs and outputs, as this was the only to get sufficient EMI Resilience and general noise immunity.
As for sound quality, we evaluated transformers by measurements but more crucially by connecting them "back to back" and simply looping them into the resident studio monitor system (Schulze TH315 Coaxials with matching amplifier, later also MEG RL900) and listened.
The transformers we routinely used for Microphone duties and outputs needed six PAIRS linked together to be audible and we actually figured that the audibility was due to the losses in level from the DCR losses, not due to bandpass behaviour or distortion, as they went away after matching levels.
And yet the most prized microphones among recording professionals (reflected by the price that they fetch) are tubed & transformer coupled Neumann Microphones and their replicas, just as the most prized microphone preamplifiers are the TAB and REDD Tubed and transformer coupled ones and their replicas and the most prized compressors are old Teleonix Tube/Optical ones.
Somehow it seems that not EVERYBODY has yet understood that "transformers are unwanted" and instead evidence tends to take the opposite position, namely that with the exception of a small minority EVERYBODY wants transformers, this includes all those too poor to afford the good (and very expensive) stuff, who may publically dump on anything they cannot afford, while secretely hankering after the lottery win to allow them to buy these exact toys themselves.
Of course. Magnetic tape used Iron Oxide and Iron is subject to a lot of hysteresis. If iron cored transformers are used they should have some bias (can be DC or HF) to overcome this. Good studio grade signal transformers do not use iron cores anyway.
As said, transformers (in this case a 1:1 line type) can be manufactured that have better than -115dB THD&N for 2V/+8dBu Input (THD falling further with lowering levels to at least better than an AP2 can measure) and can handle +26dBu with equally low distortion, this is still WAY BETTER than any op-amp or DAC out there, with a dynamic range of AT LEAST 133dB and likely better (we just cannot measure something with that low distortion with the usual industry standard audio measurement gear).
Well, as long as the best transformers have lower noise, THD, IMD, TIMD, SIMD than the best active circuits I think I prefer to use them instead.
Only by those who favour noise and distortion over the absence of noise and distortion. Honestly.
Of course, just like with active circuits, there are more bad ones than good ones. But if we judge everything by bad examples than we can use nothing and we must even disregard most musicians and orchestras as "not capable of high fidelity" (to the notes in the score).
Funny. The MEASUREMENTS for transformers from the makers of HIGH QUALITY transformers attest that in fact that transformers are MORE linear than active circuits. So I suspect that those who use THESE transformers do not know that Transformers are not linear, simply because it is patently untrue.
Because YOU personally dragged the matter into this thread, kicking and screaming, by simply dumping on everything with transformers, without qualification and in clear and obvious opposition to facts on the matter.
May I suggest that like many of us, myself included you tend towards easy generalisations.
For example I do tend to generalise Cirrus Logic products are overall poor and in my own testing this has almost always been born out (I cannot think of one instance where I ended up adopting a CS product when alternatives existed - they really where always that poor - even if they where cheapest).
However, what I do tend to do much is to qualify statements of sweeping dismissal with the term "in my experience".
For example:
In my experience the CS SPDIF receivers and the CS DAC's and ADC's I evaluated are poor performers sonically and often also in measured performance. This includes the CS4398.
This allows other experiences and tends to avoid excessive counter reactions.
So before dismissing anything you did not happen to include in your design as "useless' and "coloured" a minimal "IME" may be advised. Further, a pretty solid amount of testing may also be in order before dismissing anything, probably more though testing than you have done.
Ciao T
POST SCRIPTVM
Speaking of Microphones, I hope you realise that even the transformerless Microphones include FET's (that is condensor microphones) or tubes. So I sincerely hope you do not use any such condensor microphones with evil Fets that steal low level information for recording.
If you follow your reasoning correctly Ribbon Microphones are also out as they invariably need transformers and of course the iron in the magnet system is subject to hysteresis.
So this leaves only dynamic microphones like Sure SM58 fit for recording, ooops, darn, these have an iron core in the voice coil so there is hysteresis also and you cannot use these microphones as well.
So in the end you cannot use any microphone as they all invariably incorporate something that you disagree with. And of course speakers are also out for listening.
If you are really consequent with your prejudices you can neither record nor play back music, so I suspect you are actually making the DAC to look at sinewaves on a 'scope.
Anyway, this little POST SCRIPVM is intended not to insult you, but to use the method known as "surrealist investigation of claims of the normal" on your claims that Fets are bad, Tubes are bad, Transformers are bad and so on but that what you use is not "bad". If you may, I am practising guerilla ontology, to highlight the hilarity and untenability of your position.
In the buzzing spiky world of reality (such as we perceive and access it anyway), all things are bad in some ways and good in others. The wise man uses the good sides to his advantage and minimises the drawbacks of the bad sides. The fool condemns everything with a bad side and finds himself with nothing.
@ThorstenL:
Did you have the chance to evaluate the new ES9018 DAC??? We did test it some weeks ago, with the eval. board analog stage, based on AD797, which made it sound like any other op-amp based DAC (But I guess that THD and noise figures very great).
Today we will start evaluating ES9018 with our own No NFB discrete analog stage, and hope that this will show the real potential in ES9018.
Stil, I would like to know if you have any experiences with it.
The transformer discussion is not really of my interest. I have not yet had the pleasure of listening to any transformer coupled design, that performed the way I like.
I erspect the tests you did in the 80's, showing that you could not hear any change when using a transformer. But still this may not be the whole story.
Depending on the potential of the setup you are using to do these tests, some changes may not be audioble, even though they may be in a different setup.
Generally we experienced, that when using an high NFB op-amp design, most changes in the rest of the chain becomes in-audioble.
Bottom line is, that the gear using when testing often limits what is audioble. We have done lots of changes on our last DAC, that was not audioble on most setups. But when listening @ KvK they became crystal clear.
Supposed the rest of the gear is made from the same ideas like:
- You can not hear a transformer in the signal patch.
- Tubes are OK.
- High NFB is OK
and on and on....
Then I agree.... Adding a transformer in a DAC, may not be audioble. But that's because of the rest of the gear.
Did you have the chance to evaluate the new ES9018 DAC??? We did test it some weeks ago, with the eval. board analog stage, based on AD797, which made it sound like any other op-amp based DAC (But I guess that THD and noise figures very great).
Today we will start evaluating ES9018 with our own No NFB discrete analog stage, and hope that this will show the real potential in ES9018.
Stil, I would like to know if you have any experiences with it.
The transformer discussion is not really of my interest. I have not yet had the pleasure of listening to any transformer coupled design, that performed the way I like.
I erspect the tests you did in the 80's, showing that you could not hear any change when using a transformer. But still this may not be the whole story.
Depending on the potential of the setup you are using to do these tests, some changes may not be audioble, even though they may be in a different setup.
Generally we experienced, that when using an high NFB op-amp design, most changes in the rest of the chain becomes in-audioble.
Bottom line is, that the gear using when testing often limits what is audioble. We have done lots of changes on our last DAC, that was not audioble on most setups. But when listening @ KvK they became crystal clear.
Supposed the rest of the gear is made from the same ideas like:
- You can not hear a transformer in the signal patch.
- Tubes are OK.
- High NFB is OK
and on and on....
Then I agree.... Adding a transformer in a DAC, may not be audioble. But that's because of the rest of the gear.
Well I do not agree that CS products both sound and measure poorly, the proof is the Bryston, which measures very fine.
Regarding microphones then omnidirectional microphones for AB stereo is definately not Neumann tube mices.
Try to have a listen to this record TACET-Website - english
Andreas Spreer is recording the same jazz band with about 10 different mices, first omnis in AB, then cardioids in XY and I think figure 8 in Blumlein stereo.
This gives you the opportunity to evaluate the different makes and types, and guess what, The B&K by far is the better one, unfortunately the one used isn´t even transformerless, but I do have recordings to compare the 4006 and 4006TL´s as well as the tubes versions.
There is no doubt at all, TL´s are by far the most neutral and natural ones. Transformers can be necessary in studios, where the environment is noisy. But they are a least bad solution.
If they measure very well or not is of no concern for me, because they simply do not sound good, neither in MC step-up´s nor in mikes.
And they have problems with linearity @ low frequencies, which is also easily heard.
Some good producers and balance engineers.
My favourites are Steven Epstein (Sony), Peter Willemoes (Erato, Denon, DR, Andreas Spreer (Tacet), Kieth Johnson (RR), Dorian, Telarc ECM among others.
Regarding microphones then omnidirectional microphones for AB stereo is definately not Neumann tube mices.
Try to have a listen to this record TACET-Website - english
Andreas Spreer is recording the same jazz band with about 10 different mices, first omnis in AB, then cardioids in XY and I think figure 8 in Blumlein stereo.
This gives you the opportunity to evaluate the different makes and types, and guess what, The B&K by far is the better one, unfortunately the one used isn´t even transformerless, but I do have recordings to compare the 4006 and 4006TL´s as well as the tubes versions.
There is no doubt at all, TL´s are by far the most neutral and natural ones. Transformers can be necessary in studios, where the environment is noisy. But they are a least bad solution.
If they measure very well or not is of no concern for me, because they simply do not sound good, neither in MC step-up´s nor in mikes.
And they have problems with linearity @ low frequencies, which is also easily heard.
Some good producers and balance engineers.
My favourites are Steven Epstein (Sony), Peter Willemoes (Erato, Denon, DR, Andreas Spreer (Tacet), Kieth Johnson (RR), Dorian, Telarc ECM among others.
I think that the ESS chip costs just around the same as a reciever+ASRC+DAC, so it should not be any costlier, but we will have to check it out first.
The analog stage is completely described in the earlier project on page 20, so feel free to use it for what ever purpose, but it will need an I/V converter to function with the ESS chip.
That really depends on the price on the ES9018. It seems that ES9018 may be a little hard to source. And I really have no idea about the price.
Anyone who know the price???
Hi,
No, tested the prior one, in current output mode, passive I/U conversion and with Tube output. Also used differe nt analog stage regulation. Results where rather good.
As with any other DAC, you will hear the analogue supplies. I'd personally suggest discrete (C4S) current sources and discrete shunt regulators without LED's or Zenners (too noisy), menaing you need to trim voltages at operating temperature as you only have Vbe as reference and this is subject to a fair bit of drift.
I have designed a "post regulator" topology that eliminates these issues, but s this is IP owned by one of my employers I am not at liberty to disclose how it is done.
Nor of mine.
Our Mixing Desks used low NFB discrete circuits, I tried to re-engineer them for the then new Op-Amps and the sonic results where disastrous. So we stuck to discrete circuits. Of course, modern Op-Amp's of the highest performance classes do perform better, but non of these I see listed as "Audio" Op-Amp's.
As for the rest, I am first of all an engineer, not a scientist, so I am mostly interested in what works for my purpose. I find that nothing is perfect and different devices and arts need different approaches to minimise the limitations.
So for me, thankfully almost anything works.
Tubes work okay.
Transistors work okay.
Fet's work okay.
Op-Amp's work okay.
Transformers work okay.
Capacitors work okay.
Direct coupling works okay.
Capacitor microphones work okay.
Ribbon microphones work okay.
Dynamic microphones work okay.
Electrostatic speakers work okay.
Dynamic speakers work okay.
Horns work okay.
Dipoles work okay.
Cardiod speakers work okay.
I could go on.
But NON are universal panaceas.
Non work everywhere and all the time, they only "work" if you understand the benefits and limitations and know how to exploit them to the best effect, while avoiding the drawbacks.
Ciao T
No, tested the prior one, in current output mode, passive I/U conversion and with Tube output. Also used differe nt analog stage regulation. Results where rather good.
As with any other DAC, you will hear the analogue supplies. I'd personally suggest discrete (C4S) current sources and discrete shunt regulators without LED's or Zenners (too noisy), menaing you need to trim voltages at operating temperature as you only have Vbe as reference and this is subject to a fair bit of drift.
I have designed a "post regulator" topology that eliminates these issues, but s this is IP owned by one of my employers I am not at liberty to disclose how it is done.
Nor of mine.
Our Mixing Desks used low NFB discrete circuits, I tried to re-engineer them for the then new Op-Amps and the sonic results where disastrous. So we stuck to discrete circuits. Of course, modern Op-Amp's of the highest performance classes do perform better, but non of these I see listed as "Audio" Op-Amp's.
As for the rest, I am first of all an engineer, not a scientist, so I am mostly interested in what works for my purpose. I find that nothing is perfect and different devices and arts need different approaches to minimise the limitations.
So for me, thankfully almost anything works.
Tubes work okay.
Transistors work okay.
Fet's work okay.
Op-Amp's work okay.
Transformers work okay.
Capacitors work okay.
Direct coupling works okay.
Capacitor microphones work okay.
Ribbon microphones work okay.
Dynamic microphones work okay.
Electrostatic speakers work okay.
Dynamic speakers work okay.
Horns work okay.
Dipoles work okay.
Cardiod speakers work okay.
I could go on.
But NON are universal panaceas.
Non work everywhere and all the time, they only "work" if you understand the benefits and limitations and know how to exploit them to the best effect, while avoiding the drawbacks.
Ciao T
Well... At 50USD, it should be possible to make the complete DAC at 200USD, including PCB, onboard transformers and connectors. A complete DAC.
Actually I see no reason why anyone should pay more, for less than that.
Where do you source your ESS chip at TPA?
Well then go for it. The end-product will show if it is more for less, or less for less. Can the ES9018 be applied in hard-ware mode?
A suggestion: as you seem to have your ideas about the output stage pretty clear why don't you rapidly make a PCB/circuit available that we can plug onto the TPA Buffalo II for comparison with other output topologies? I guess that there are quite a few of us that would like to try different I/V stages with the ES9018. That should take you very short time, cost very little, have a market, and might even be interesting!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.