It's rather easy the make a transformer yourself if you know how to do it
Well I guess anything is easy when you know how to do it. There is nothing trival about designing pulse tranformers for digital audio. Most of what have measured and listened to were not very good. I spent a year optmizing mine and read several books and much computer simulation and listening on mine. I was also tutored by John Marshall at Schott who designed the origional ones for Crystal Semiconductor. The Pulse Engineering unit is not even close to being one of the better ones available. Add the effects on TDR from leakage inductance and non linearities from ANY! DC currents. Oh ya..... forgot to mention hysteresis and Magnetostriction. This was one of the toughest audio design projects I have ever undertaken.
H.H.
P.S. Most optical interfaces (ST glass or Toslink) sound awful.
Well I guess anything is easy when you know how to do it. There is nothing trival about designing pulse tranformers for digital audio. Most of what have measured and listened to were not very good. I spent a year optmizing mine and read several books and much computer simulation and listening on mine. I was also tutored by John Marshall at Schott who designed the origional ones for Crystal Semiconductor. The Pulse Engineering unit is not even close to being one of the better ones available. Add the effects on TDR from leakage inductance and non linearities from ANY! DC currents. Oh ya..... forgot to mention hysteresis and Magnetostriction. This was one of the toughest audio design projects I have ever undertaken.
H.H.
P.S. Most optical interfaces (ST glass or Toslink) sound awful.
So....
...😱
Well it seems that it would take a lots of time, to learn the phenomenom behind those small transformers.
So, it leans more and more into buying a commercial one.
Any of those transformers you used that wasnt that bad as the rest and measured decent, which you could recommend?
/kasra
...😱
Well it seems that it would take a lots of time, to learn the phenomenom behind those small transformers.
So, it leans more and more into buying a commercial one.
Any of those transformers you used that wasnt that bad as the rest and measured decent, which you could recommend?
/kasra
DAC digital input
What you really want on the digital input to your DAC is.....
http://products.analog.com/products/info.asp?product=AD830
Well Jocko and Scott, the cat is out of the bag now.....
H.H.
What you really want on the digital input to your DAC is.....
http://products.analog.com/products/info.asp?product=AD830
Well Jocko and Scott, the cat is out of the bag now.....
H.H.
Steroids
I think your right. Alot of weight lifters use them but, someone should tell that girls mother. And she should also be using a support belt if she is going to lift that fat cat or any other heavy weights!
H.H. ( heavy hefter)
I think your right. Alot of weight lifters use them but, someone should tell that girls mother. And she should also be using a support belt if she is going to lift that fat cat or any other heavy weights!
H.H. ( heavy hefter)
Inpiration
kasra, check my DAC if you want some inspiration. I have pcb's left but I imagine that you don't have the necessary Crystal chips.
kasra, check my DAC if you want some inspiration. I have pcb's left but I imagine that you don't have the necessary Crystal chips.
Most optical outputs???
All sound bad. Because no one, repeat, no one, designing them knows squat about optical transmission. They could be made to sound good if they just got their head out of that bag.
Another of my claims to fame was the $2 coax (the cost being in the BNC connectors) that wiped out every optical cable a large US high-end company had in their arsenal. They won't admit it to you guys, but they know the truth.
They then started selling coax cables, quietly. Said it was in response to customer demand. But I knew better.
Jocko
All sound bad. Because no one, repeat, no one, designing them knows squat about optical transmission. They could be made to sound good if they just got their head out of that bag.
Another of my claims to fame was the $2 coax (the cost being in the BNC connectors) that wiped out every optical cable a large US high-end company had in their arsenal. They won't admit it to you guys, but they know the truth.
They then started selling coax cables, quietly. Said it was in response to customer demand. But I knew better.
Jocko
Many ways to skin a cat...
Digital transformers are supposed to provide good insulation to interference and to mangle the signal as little as possible.
I always thought a toroid had the lowest inductive losses of any type of transformer. What losses are you concerned with then? Radiated energy? One could always lock the transformer into a little iron enclosure.
Secondly, you would like to have as little capacitive coupling as possible. I guess the Elektor design, which has the primary and secondary windings separated by at least 15 mm, is superior to anything Harry can do with his cores. But then, I have not discerned exactely what separates the windings.
And finally, the "sound" of the digital link will be much less influenced by imperfections of the transformers if you use clock feedback. I use clock feedback between my DAC and player without any transformer (and without S/PDIF) and am quite happy for the moment.
Differential amps sound like a good idea. However, they can only handle a signal that does not exceed their common mode input range. While this should be the case in most instances, just to go sure, one would be tempted to either connect grounds directly or through a capacitor. This kind of defeats the purpose....
Also, interference will only be rejected up to some percentage of the bandwidth of the amp, unless you use passive filtering.
Greetings,
Eric
Digital transformers are supposed to provide good insulation to interference and to mangle the signal as little as possible.
I always thought a toroid had the lowest inductive losses of any type of transformer. What losses are you concerned with then? Radiated energy? One could always lock the transformer into a little iron enclosure.
Secondly, you would like to have as little capacitive coupling as possible. I guess the Elektor design, which has the primary and secondary windings separated by at least 15 mm, is superior to anything Harry can do with his cores. But then, I have not discerned exactely what separates the windings.
And finally, the "sound" of the digital link will be much less influenced by imperfections of the transformers if you use clock feedback. I use clock feedback between my DAC and player without any transformer (and without S/PDIF) and am quite happy for the moment.
Differential amps sound like a good idea. However, they can only handle a signal that does not exceed their common mode input range. While this should be the case in most instances, just to go sure, one would be tempted to either connect grounds directly or through a capacitor. This kind of defeats the purpose....
Also, interference will only be rejected up to some percentage of the bandwidth of the amp, unless you use passive filtering.
Greetings,
Eric
Transformer and digital interface design
The further you seperate the windings the more leakage inductance you have. Leakage inductance limits bandwidth and causes signal reflections that are easy to see with TDR measurements. Most design techniques that concentrate on lowering interwinding capacitance, do so by compromizing the factors that make for good bandwith and minimal relections. High permability cores minimize the number of windings and therefore the interwinding capacitance. BUT, the permability drops off with increasing frequency and reflections increase since the core losses at high frequencies no longer let the actual impedance of the secondary be coupled to the primary. A 75 ohm termination can look like a couple of hundred ohms at a few megahertz. Bifilar windings are the best winding technique for good bandwidth and low leakage inductance but also have the highest capacitance. Does this sound simple....? It is not! Also the core different materials sound different. I found two I liked out of the 2 dozen I listened to. The sound of coupling caps for this interface are also quite audible.
"Differential amps sound like a good idea. However, they can only handle a signal that does not exceed their common mode input range. While this should be the case in most instances, just to go sure, one would be tempted to either connect grounds directly or through a capacitor. This kind of defeats the purpose....
Also, interference will only be rejected up to some percentage of the bandwidth of the amp, unless you use passive filtering"
Differential video amps have large common mode range and good CMRR up to several megahertz. That is what they are designed for. Go read the data sheets! Grounds can be partially decoupled to take advatange of a differential vidio amp digtal input. This is not easy stuff like audio frequencies, we are talking R.F., magnetics design, and TDR here. Anybody wonder why the digital interface is seldom done right?
The further you seperate the windings the more leakage inductance you have. Leakage inductance limits bandwidth and causes signal reflections that are easy to see with TDR measurements. Most design techniques that concentrate on lowering interwinding capacitance, do so by compromizing the factors that make for good bandwith and minimal relections. High permability cores minimize the number of windings and therefore the interwinding capacitance. BUT, the permability drops off with increasing frequency and reflections increase since the core losses at high frequencies no longer let the actual impedance of the secondary be coupled to the primary. A 75 ohm termination can look like a couple of hundred ohms at a few megahertz. Bifilar windings are the best winding technique for good bandwidth and low leakage inductance but also have the highest capacitance. Does this sound simple....? It is not! Also the core different materials sound different. I found two I liked out of the 2 dozen I listened to. The sound of coupling caps for this interface are also quite audible.
"Differential amps sound like a good idea. However, they can only handle a signal that does not exceed their common mode input range. While this should be the case in most instances, just to go sure, one would be tempted to either connect grounds directly or through a capacitor. This kind of defeats the purpose....
Also, interference will only be rejected up to some percentage of the bandwidth of the amp, unless you use passive filtering"
Differential video amps have large common mode range and good CMRR up to several megahertz. That is what they are designed for. Go read the data sheets! Grounds can be partially decoupled to take advatange of a differential vidio amp digtal input. This is not easy stuff like audio frequencies, we are talking R.F., magnetics design, and TDR here. Anybody wonder why the digital interface is seldom done right?
Don't pick a fight with Harry and me on this one. I know who will win.
1.) He is right about low capacitance. Just because Crystal has it wrong in their data sheets...........come to think of it.......they have everything wrong about transmission lines.
2.) Diff amps are the only way to go.
3.) You are right about clock feedback. It solves amost all of the problems. But a D/A box is supposed to be able to hook up to any source without all the ancillary stuff. But since the three of us didn't invent the format, we have to work with what is provided.
The obvious solution is to use a one box unit, and get rid of all the refection, jitter, and isolation problems.
Hope we are all still pals. [joke] Remember, we have German ancestors.
Jocko
1.) He is right about low capacitance. Just because Crystal has it wrong in their data sheets...........come to think of it.......they have everything wrong about transmission lines.
2.) Diff amps are the only way to go.
3.) You are right about clock feedback. It solves amost all of the problems. But a D/A box is supposed to be able to hook up to any source without all the ancillary stuff. But since the three of us didn't invent the format, we have to work with what is provided.
The obvious solution is to use a one box unit, and get rid of all the refection, jitter, and isolation problems.
Hope we are all still pals. [joke] Remember, we have German ancestors.
Jocko
speed?
Go slow with the horses (German saying). I never mind a well-fired-up discussion.
If I remember correctly, the S/PDIF signal is supposed to be bandlimited at the source. On the other hand, this would probably increase the susceptibility to jitter, so few designers decide to limit in the first place.
What is partial grounding?
In theory, reflections are not so bad as long as they always stay the same and the Schmitt-Trigger in the receiver works perfectly.
Eric
Go slow with the horses (German saying). I never mind a well-fired-up discussion.
If I remember correctly, the S/PDIF signal is supposed to be bandlimited at the source. On the other hand, this would probably increase the susceptibility to jitter, so few designers decide to limit in the first place.
What is partial grounding?
In theory, reflections are not so bad as long as they always stay the same and the Schmitt-Trigger in the receiver works perfectly.
Eric
- Status
- Not open for further replies.