Sorry Jacco,
I was talking about figure 2.
I tihnk the real beauty about figure 1 (despite one of the current sources being drawn backwards) is in putting the shunt branch of the circuit on the local circuit, near the load, and keeping the current source close to the power supply, In this way, there is little or no radiation from the power distribution network.
Figure 2 may indeed show better transient rejection from the PSU though.
I was talking about figure 2.
I tihnk the real beauty about figure 1 (despite one of the current sources being drawn backwards) is in putting the shunt branch of the circuit on the local circuit, near the load, and keeping the current source close to the power supply, In this way, there is little or no radiation from the power distribution network.
Figure 2 may indeed show better transient rejection from the PSU though.
Richard, I like both 1 and 2. I like the way that you can think 'backwards'. 
I can't be sure which would be best, overall.
One factor that I am always concerned about is noise creation and amplification by current sources. I can't comment here too much, because I don't have specific values.
However, either of these current sources would probably be better than any IC over a large frequency range.
I can't be sure which would be best, overall. One factor that I am always concerned about is noise creation and amplification by current sources. I can't comment here too much, because I don't have specific values.
However, either of these current sources would probably be better than any IC over a large frequency range.
Here, more detailed schematics, with values (not exactly fixed, but around them), I have not tested or even listen to them but I think that I will go first for the fig2 schematic, for a better line rejection, in load transients they must be very near, your insight on all this is welcome as always. I replaced the small current sources to bias the zeners by resistors, very often when I use them, even if the measures are better, while listening I generally change them to resistors where I can.
An externally hosted image should be here but it was not working when we last tested it.
Other than the absence of a gate-stopper, I like the position of the cap where it is. With the series resistor, it forms a first-order low-pass filter for the Zener noise, which I think is more effective than bypassing the Zener directly.
No doubt John will rag on me if I'm wrong about that.
No doubt John will rag on me if I'm wrong about that.
Recently, I tried to use TL431 for my headphone amp . .
Just similar as Justcallmedad's . .
I arrnaged the cap down to the ground . .
And, the noise was incredible . . too loud . . ear pain . .
Couldn t find any solution . .
I finally gave up TL431 . .
Probably, I don t really understand how to use TL431 correctly . .
I personally decided no more to try to use it.
jh
Just similar as Justcallmedad's . .
I arrnaged the cap down to the ground . .
And, the noise was incredible . . too loud . . ear pain . .
Couldn t find any solution . .
I finally gave up TL431 . .
Probably, I don t really understand how to use TL431 correctly . .
I personally decided no more to try to use it.
jh
I've been following DIYAudio.com for a few months and only a few days ago zero'd in on this topic. Having slowly read through the 60 pages, I'm left with the thought that this site should give John a forum of his own. The Blowtorch Preamplifier thread name gave no indication of his involvement.
After reading through the threads and appreciating his professional approach to his input (when I'm sure he would have liked to have occasionally muttered a couple of sage observations) I'm struck with the the depth of knowledge and the hands on experience his observations are backed by. I chuckle at the attempts to bleed knowledge, knowing full well that, given a schematic of the blowtorch, few would even get in the same ballpark (sonically) that he builds into the preamp as second nature. I have never heard the preamp, but am a student of the art and his reputation surely precedes him.
I feel that he has the ability to challenge you to think rather than just spouting dogma, those trying to educate themselves might benefit. My motivation is that I might learn a thing or two, but others might realize he's involved if his name is attached to it.
Just a thought.
After reading through the threads and appreciating his professional approach to his input (when I'm sure he would have liked to have occasionally muttered a couple of sage observations) I'm struck with the the depth of knowledge and the hands on experience his observations are backed by. I chuckle at the attempts to bleed knowledge, knowing full well that, given a schematic of the blowtorch, few would even get in the same ballpark (sonically) that he builds into the preamp as second nature. I have never heard the preamp, but am a student of the art and his reputation surely precedes him.
I feel that he has the ability to challenge you to think rather than just spouting dogma, those trying to educate themselves might benefit. My motivation is that I might learn a thing or two, but others might realize he's involved if his name is attached to it.
Just a thought.
I report here again the two current sources schematics for better clarity and to avoid changing from one page to another.
Jam, if you speak about the zener to set the current trough the BJT (not the TL431) I think that the noise contribution is not so important for a low voltage zener (just a few volts, it could be also 2 led’s), this noise is not “amplified” by the circuit. I fear that a cap in // with the zener worsens things in matter of line rejection and transient load regulation, moreover I think of replacing it by a resistor…
SY, yes a gate stopper resistor might be necessary, I would see with a real circuit, however on the very first circuit I tested (Post #259) it's not necessary.
jh6you, do you speak about the 10uF cap between the TL431 adjust and output pins ?
“And, the noise was incredible . . too loud . . ear pain . .”
Did you tested it between adj and ground ???
Hi EUVL, Happy to hear from you
Yes I try to learn a little more about audio circuits.
For a 5V version I will do it the same way, I think. Tell me more, is 50mA the current drawn by your audio circuit, and the 8.4V the unreg voltage at the input to the shunt regulation compound?
An externally hosted image should be here but it was not working when we last tested it.
Jam, if you speak about the zener to set the current trough the BJT (not the TL431) I think that the noise contribution is not so important for a low voltage zener (just a few volts, it could be also 2 led’s), this noise is not “amplified” by the circuit. I fear that a cap in // with the zener worsens things in matter of line rejection and transient load regulation, moreover I think of replacing it by a resistor…
SY, yes a gate stopper resistor might be necessary, I would see with a real circuit, however on the very first circuit I tested (Post #259) it's not necessary.
jh6you, do you speak about the 10uF cap between the TL431 adjust and output pins ?
“And, the noise was incredible . . too loud . . ear pain . .”
Did you tested it between adj and ground ???
Hi EUVL, Happy to hear from you
Yes I try to learn a little more about audio circuits.
For a 5V version I will do it the same way, I think. Tell me more, is 50mA the current drawn by your audio circuit, and the 8.4V the unreg voltage at the input to the shunt regulation compound?
Poobah, the goal is to carry out a good power supply for audio , so which are the objectives, IMMO:
1) Very good rejection of the power line garbage (high bandwidth noise).
2) Stable Vout with a very good transient response to line and load variations.
The impedance in itself isn’t a goal (I think) provided as I said that Vout is constant, without ringing to load or line transients and of course as noise free as possible.
, so which are the objectives, IMMO:1) Very good rejection of the power line garbage (high bandwidth noise).
2) Stable Vout with a very good transient response to line and load variations.
The impedance in itself isn’t a goal (I think) provided as I said that Vout is constant, without ringing to load or line transients and of course as noise free as possible.
Justcallmedad,
Mr. Curl tends to agree with you and also the cap to ground might not be necessary............bset to try it to see if there are any sonic advantages or disatvantages.
An LED might also make a good alternative to a zener but the zener you have chosen should also be low in noise.
Regards,
Jam
Mr. Curl tends to agree with you and also the cap to ground might not be necessary............bset to try it to see if there are any sonic advantages or disatvantages.
An LED might also make a good alternative to a zener but the zener you have chosen should also be low in noise.
Regards,
Jam
Richard, I like this approach that you have taken. I personally don't know which one to pick, but I see potential pitfalls for the non-experienced in this circuit.
Actually the way you have made it is OK. However, let me point out where some things could go wrong:
First, zeners are generally very noisy devices. Where 4nV/rt Hz might be considered a 1K noise source, a 15V zener diode might have 2000nV/rt Hz noise. However, a 3V zener is pretty darn quiet, because it relies on the actual 'Zener effect' rather than an avalanche effect. Over 5V or so, zener diodes get really noisy! Of course, with this current source, you used only 3V, so you are in good shape. That is why I had to ask for specific values, in order to see if there was any problem.
It is also important to see this circuit as a potential amplifier of its own self noise. Without the 10 ohm resistor in the emitter, you would have a noisy current source. This is a frequent oversight made by many designers of active loads, etc. However, with the 10 ohm resistor you have dropped your potential noise contribution by the current source about 10/.125 or about 80 times. That 10 ohm resistor pads the transconductance of the transistor that much! Perhaps, you knew that, Richard, but I hope that I can point this out to others, who might be looking at these circuits.
Finally, I would have used a jfet as a constant current source instead of the 1 K resistor. The actual current is not too important , 2-10ma would probably do, so long as you have enough voltage across the fets to keep it acting like a current source.
Actually the way you have made it is OK. However, let me point out where some things could go wrong:
First, zeners are generally very noisy devices. Where 4nV/rt Hz might be considered a 1K noise source, a 15V zener diode might have 2000nV/rt Hz noise. However, a 3V zener is pretty darn quiet, because it relies on the actual 'Zener effect' rather than an avalanche effect. Over 5V or so, zener diodes get really noisy! Of course, with this current source, you used only 3V, so you are in good shape. That is why I had to ask for specific values, in order to see if there was any problem.
It is also important to see this circuit as a potential amplifier of its own self noise. Without the 10 ohm resistor in the emitter, you would have a noisy current source. This is a frequent oversight made by many designers of active loads, etc. However, with the 10 ohm resistor you have dropped your potential noise contribution by the current source about 10/.125 or about 80 times. That 10 ohm resistor pads the transconductance of the transistor that much! Perhaps, you knew that, Richard, but I hope that I can point this out to others, who might be looking at these circuits.
Finally, I would have used a jfet as a constant current source instead of the 1 K resistor. The actual current is not too important , 2-10ma would probably do, so long as you have enough voltage across the fets to keep it acting like a current source.
I only ask because in both circuits your Vout is very dependant on the Vgsth & the gm of the FET.
The Tl431 is not controlling the output voltage but rather the gate... I realize this may be exactly your intent. But this circuit will not have the lowest output impedance as compared to other topologies.
Are you trying to avoid the TL431 from exerting direct control over the output voltage in an effort to reduce noise?
The Tl431 is not controlling the output voltage but rather the gate... I realize this may be exactly your intent. But this circuit will not have the lowest output impedance as compared to other topologies.
Are you trying to avoid the TL431 from exerting direct control over the output voltage in an effort to reduce noise?
Justcallmedad said:
hi,Justcallmedad
I had tried this topo of reg ps for my pre amp a few months ago.I use LM317 for the the 1st stage,and I found the sound of the LM317 use as standard reg for the 1st stage much better than CCS.
http://www.hifidiy.net/dispbbs.asp?BoardID=2&id=19991&replyID=205137&star=4&skin=0
An externally hosted image should be here but it was not working when we last tested it.
Justcallmedad said:
Yes.. But, I had it across Cathode and Anode of TL431..
I got all voltages as I wanted..
But, through my headphone, I also got very loud "bi--------" sound..
Couldn't understand..
Soon I gave up TL431 and replaced it with zener diodes..
The problem immediately gone..
I think that the problem was my ignorance about how to use TL431 correctly..
Regards
jh
john curl said:
Hi,
I can't say that I share your opinion.
In the vishay data sheet of the bzx55c3v0 they choose an operating point of 5 mA, probably close to optimum, and the differential resistance there is 85 ohms.
In fig. 1, the 20V source sees a load of 85 Ohm + 220u (plus the current that flows through the transistor + payload).
Since at AC frequencies the impedance of the capacitor is small, all ripple and noise of the 20V source appears across the 85 Ohms of the Zener diode.
But then, all ripple and noise appear across the BE junction of the transistor. Now imagine the effect of 100 mV ripple on the impedance of the current source!
In fact, with a load of a few KOhms it might make a nice ripple & noise amplifier.
In fig. 2, the source impedance of the current source is limited by the 85 Ohms + 220uF. In my book, that's not a current source at AC.
I think that in both figs the role of the 220u cap has not been engineered in but comes from gut feeling. "We'll decouple everyting to GND, that can't be bad." But in Fig.1 we open the doors to ripple & noise, and in Fig. 2 we spoil the high impedance of the current source.
Now it's 5:30 AM here and I'm too tired to check the par regulator, but on first sight I think that the second 220u should go from gate to source and not from gate to drain, because usually the gate/source voltage is asumed to control the current in a FET.
But then, filtering with 100K * 220uF makes the FET so slow that
we can't expect any AC suppression.
And then, the FET's CutOff voltage enters the equation for the output voltage. The CutOff voltage of a FET resembles a random number generator. I see trimpots and/or handselected FETs.
good night, Gerhard
- Status
- Not open for further replies.