janneman said:
Well, but my point earlier in this thread was that the 108X don't
really seem much better, judging from the data sheets, but as
i also said, I haven't read them carefully for the 108X regulators,
so maybe there are things that are not immediately obvious
from the data. So I think the question remains, are there any
advantages that outweigh the higher cost and potential
stability problems? And what about newer/other version of
317, how do they stand up in the competition?
Re: Regulators and capacitors
LT1762 at <20uV noise, a 17dB improvement over LT1085. 150mA.
LT1964 for negative. <30uV, 200mA.
These are so cheap, small and require so little in the way of support components that you can place a lot of them throughout a board, close to the chips you are powering.
Look through the rest of Linear's catalog; there are some higher power chips that still provide better noise performance than LT1085.
in this thread.
Last may I saidFred Dieckmann said:
LT1762 at <20uV noise, a 17dB improvement over LT1085. 150mA.
LT1964 for negative. <30uV, 200mA.
These are so cheap, small and require so little in the way of support components that you can place a lot of them throughout a board, close to the chips you are powering.
Look through the rest of Linear's catalog; there are some higher power chips that still provide better noise performance than LT1085.
in this thread.
Re: Regulators and capacitors
You forget one thing: the Q value! If this isn't high enough you won't get any resonance (at all).
Fred Dieckmann said:
You forget one thing: the Q value! If this isn't high enough you won't get any resonance (at all).
You forget one thing: the Q value!
No, I didn't.......... What do you think the suggestion about the series damping resistor was all about.
ALL regulators exhibit rising output imp with freq
No........ If you design a discreet transistor regulator with conservative amount of feedback and wideband techniques like cascode circuits and resistive collector/drain loads you can get a circuit with resistive output impedance to the point where the capacitive output impedance of the output cap dominates and the output impedance becomes capacitive.
I have also seen regulator circuits that use no output caps but I would rather rely on a capacitor for it's low impedance at high frequencies.
No, I didn't.......... What do you think the suggestion about the series damping resistor was all about.
ALL regulators exhibit rising output imp with freq
No........ If you design a discreet transistor regulator with conservative amount of feedback and wideband techniques like cascode circuits and resistive collector/drain loads you can get a circuit with resistive output impedance to the point where the capacitive output impedance of the output cap dominates and the output impedance becomes capacitive.
I have also seen regulator circuits that use no output caps but I would rather rely on a capacitor for it's low impedance at high frequencies.
Quite right Fred. If you do it right the output cap takes over at just that frequency the regulators output impedance starts to rise. That way you can have a power supply that has neither overshoot nor ringing. Overall output impedance can be pretty low then. I’ve build such simple low-noise regulator circuits with an output impedance at the low end of app. 75 mOhm and 20 mOms at the high end.
All integrated 3-pin regulators I’ve seen show ringing/overshoot with fast load transients.
Concerning the open loop x-lingtons: The dynamic output resistance of such a set-up is simply the output resistance of the last transistor: 25/I_load, where I_load is in mA.
Since the reference voltage of most Audio-DACs is taken from the power supply, any supply modulation due to noise or the load of the DAC itself will result in modulation of the output signal. So a high output resistance of the power supply is likely to cause (unnecessary) IM distortion.
All integrated 3-pin regulators I’ve seen show ringing/overshoot with fast load transients.
Concerning the open loop x-lingtons: The dynamic output resistance of such a set-up is simply the output resistance of the last transistor: 25/I_load, where I_load is in mA.
Since the reference voltage of most Audio-DACs is taken from the power supply, any supply modulation due to noise or the load of the DAC itself will result in modulation of the output signal. So a high output resistance of the power supply is likely to cause (unnecessary) IM distortion.
Further Thoughts on Power Supply
Hi,
Switching back and forth between the circuit with the LM431 and the LM329 I found that the sound with the LM329 is more smooth, almost less distorted. This is especially heard on voices that get a slight raspy quality with the LM431. When I replaced the 30 Ohm DC resistance NEOSID 2.2mH coil by a much more heftier one with a few tenths Ohm of reistance bass bloomed more.😕 😕
I keep you posted of further developments😉
Hi,
Switching back and forth between the circuit with the LM431 and the LM329 I found that the sound with the LM329 is more smooth, almost less distorted. This is especially heard on voices that get a slight raspy quality with the LM431. When I replaced the 30 Ohm DC resistance NEOSID 2.2mH coil by a much more heftier one with a few tenths Ohm of reistance bass bloomed more.😕 😕
I keep you posted of further developments😉
Re: Further Thoughts on Power Supply
can't imagine, it was an outside your head event
must be Imagine /John Lennon
is that so??????Elso Kwak said:
can't imagine, it was an outside your head event
must be Imagine /John Lennon
Pjotr said:
Almost any regulator (even the 431) has overshoots/ringing when they are connected with "normal" parts around.
halojoy said:what does the boy elso say?
😀😕 😕 😀
halojoy, you seems to be both 😕 and
but this is a normal feeling when someboby talks about audiophilic matters. This is feelings which is hard to argue about, not impossible though.I am no energy waisterperanders said:
halojoy, you seems to be both 😕 and
Are you?
peranders 😉 😉
Congratulations - first Swedish on the Moon
>999 posts
Ghost out of the Bottle
Well the ghost
is definitely out of the bottle 
Time will learn whether this brings good
or evil 
Hi Halojoy,halojoy said:what does the boy elso say?
😀
Well the ghost
is definitely out of the bottle Time will learn whether this brings good
or evil 
Pjotr said:
Pjotr, Fred,
I don't agree. However you design it, there will always be a freq where the inductance value of the output impedance and the cap of the bypass cap are equal and in opposite phase. Indeed, the Q, together with the available loop gain at that freq will determine whether you will get oscillations or just overshoot and ringing. You can design your regulator with a high resistive output imp component to lower the Q, but that makes you regulator less "good", if you know what I mean.
So, I gree that you can make a lot of design choises to position the oscillatory point, but it will always be there.
Jan Didden
Jan, that is true concerning the Q of the circuit: Finally you have mainly to do with a second order system if there is overall feedback. But if you keep the Q-factor of the closed loop gain below 0.6 (Bessel response) then there is no overshoot or ringing.
It was already stated earlier in this thread that one of the causes a 78xx do sound better because of the low bandwidth of the device. I think there is some truth in it, although it stays a noisy device.
You can use a 431 device as a voltage reference, that is no problem. But you need to avoid to expose it to load transients and you need to filter the remaining noise down with a low pass filter. Attached a circuit that uses this. R3 and C3 filter noise from the TL431 and prevent load transients entering the TL431. In stead of R4, R5 and IC1 a 6.8V low-noise zener diode can be used as well. R2 is somewhat arbitrary and depends on the input voltage. A current of app. 2 mA to 3 mA trough it is a good choice. You can use a 7812 as a pre-regulator, C1 can be much lower then (app. 100 uF). R1 and C1 will keep load transients away from the 7812.
Note that I did not test this circuit and it is only a proposal. I am myself not a fan of such open loop circuits due to the inherent high output impedance at low frequencies. Further such a circuit is not really “open loop”. There are many hidden feedback mechanisms involved.
The trick here is to build a regulator with discrete parts. Then you have complete control over the phase behaviour of the loop gain. When the phase of the loop gain is no more the 120 degrees at the 0dB crossing point there is no overshoot. With normal of-the-shelf components the loop bandwidth can’t be very high then and the output cap need to take over then for overall low output impedance.peranders said:
It was already stated earlier in this thread that one of the causes a 78xx do sound better because of the low bandwidth of the device. I think there is some truth in it, although it stays a noisy device.
You can use a 431 device as a voltage reference, that is no problem. But you need to avoid to expose it to load transients and you need to filter the remaining noise down with a low pass filter. Attached a circuit that uses this. R3 and C3 filter noise from the TL431 and prevent load transients entering the TL431. In stead of R4, R5 and IC1 a 6.8V low-noise zener diode can be used as well. R2 is somewhat arbitrary and depends on the input voltage. A current of app. 2 mA to 3 mA trough it is a good choice. You can use a 7812 as a pre-regulator, C1 can be much lower then (app. 100 uF). R1 and C1 will keep load transients away from the 7812.
Note that I did not test this circuit and it is only a proposal. I am myself not a fan of such open loop circuits due to the inherent high output impedance at low frequencies. Further such a circuit is not really “open loop”. There are many hidden feedback mechanisms involved.
Attachments
Re: Improved (?) Circuit
It is getting better and better!
I added two 4700µF/35V caps parallel to C10 in my latest circuit with the LM329.
Imaging is better and the soundfield is more coherent.
I measured the ripple voltage at the input of the regulator : It is 350mV pp (triangular shaped). At C10 it is 50 mV pp (sinusidal shaped). A sevenfold reduction!
🙂
Hi ,Elso Kwak said:
It is getting better and better!
I added two 4700µF/35V caps parallel to C10 in my latest circuit with the LM329.
Imaging is better and the soundfield is more coherent.
I measured the ripple voltage at the input of the regulator : It is 350mV pp (triangular shaped). At C10 it is 50 mV pp (sinusidal shaped). A sevenfold reduction!
🙂
Pjotr said:
OK Pjotr, I see your point on the critical damping of the loop. Your proposed "open loop" regulator being not really a regulator but a power emitter follower with a DC input. But I seem to remember that sometimes even emitter followers can oscillate. Do you know anything about that?
Jan Didden
317 / 1086
The main difference between 1086 and 317 is output noise, although the specs do not show this.
The LT1086 is considerably and consistently quieter, but I've not tried non-LT parts. Impedance performance is not dramatically different though, and this is usually more important than steady state noise, for most situations.
The LT1085's and higher current variants do have lower o/p impedance, but at a cost that makes discrete designs worth pursuing, if space is available.
I've never had any of these low-dropout reg's oscillate when used in accordance with the data sheet guidelines, and I've used loads.
Andy.
The main difference between 1086 and 317 is output noise, although the specs do not show this.
The LT1086 is considerably and consistently quieter, but I've not tried non-LT parts. Impedance performance is not dramatically different though, and this is usually more important than steady state noise, for most situations.
The LT1085's and higher current variants do have lower o/p impedance, but at a cost that makes discrete designs worth pursuing, if space is available.
I've never had any of these low-dropout reg's oscillate when used in accordance with the data sheet guidelines, and I've used loads.
Andy.
Thanks Andy,
This is the type of answer I had expected to get, but didn't get
until now. Just as you say there is no difference in datasheet
figures for noise or line/load regulation between the 1086 and
the 317 (or 317T as the 1.5A version is sometimes called). Yet,
people have been trying to tell me that there are such differences.
Maybe they have compared other regulators in the families, but
comparing these two seems fair, since they are both
1.5A regulators.
I assume you have measured a (large?) number of devices of
both types to come to the conclusion you stated? You wouldn't
have any figures to give an idea of how big the difference is,
would you?
BTW I understand you used LT317 only, which is interesting
since these seem usually considered better than other brands.
This is the type of answer I had expected to get, but didn't get
until now. Just as you say there is no difference in datasheet
figures for noise or line/load regulation between the 1086 and
the 317 (or 317T as the 1.5A version is sometimes called). Yet,
people have been trying to tell me that there are such differences.
Maybe they have compared other regulators in the families, but
comparing these two seems fair, since they are both
1.5A regulators.
I assume you have measured a (large?) number of devices of
both types to come to the conclusion you stated? You wouldn't
have any figures to give an idea of how big the difference is,
would you?
BTW I understand you used LT317 only, which is interesting
since these seem usually considered better than other brands.
- Status
- Not open for further replies.