the TPS7A4001 regulator output noise of 4uV. No idea how much noise Salas shunt regulator delivers?
Im thinking to look at shunt regulators but was quite surprised its following.
Using salas surpasses the TPS7A4001 Reg?
Im thinking to look at shunt regulators but was quite surprised its following.
Using salas surpasses the TPS7A4001 Reg?
@rhythmsandy: interesting question! And I wonder why there is no post so far!
I am affected by the search for a "ripple-free" PSU for quite a long time, since I have built the "CALVIN - LineBuffer".
I have given Calvin's buffer the best +/-18VDC PSU that I could think of so far: a LM3x7 pre-regulator design (see image 1).
My HP/Agilent DMM is showing an AC-ripple of about 450µV.
How do I measure the quality of a PSU?
But there is still a lot of hum on its output.
A friend of mine told me that there is no hum using SALAS' shunt.
I then decided to find out by myself (we are DIY - aren't we?), which PSU will give the best results!
I did an EAGLE-layout for 3 different PSU-designs (see image 2) and ordered the PCBs yesterday:
1st: Jan Didden's SuperRegulator (giving it an additional LM317 line Pre-regulator)
2nd: Aleš Mravlak permitted me to do a TPS7A4701 layout on a 50x50mm PCB. You do need to have eyes like an eagle to solder the TPS7A4701! I do not have!
3rd: a very low-cost LT1763 regulator
I wonder, how these 3 different PSUs will perform with about 100mA on their outputs!
I need to be patient for about 4 weeks, until I will receive the etched PCBs.
Kind regards - Rudi_Ratlos
I am affected by the search for a "ripple-free" PSU for quite a long time, since I have built the "CALVIN - LineBuffer".
I have given Calvin's buffer the best +/-18VDC PSU that I could think of so far: a LM3x7 pre-regulator design (see image 1).
My HP/Agilent DMM is showing an AC-ripple of about 450µV.
How do I measure the quality of a PSU?
But there is still a lot of hum on its output.
A friend of mine told me that there is no hum using SALAS' shunt.
I then decided to find out by myself (we are DIY - aren't we?), which PSU will give the best results!
I did an EAGLE-layout for 3 different PSU-designs (see image 2) and ordered the PCBs yesterday:
1st: Jan Didden's SuperRegulator (giving it an additional LM317 line Pre-regulator)
2nd: Aleš Mravlak permitted me to do a TPS7A4701 layout on a 50x50mm PCB. You do need to have eyes like an eagle to solder the TPS7A4701! I do not have!
3rd: a very low-cost LT1763 regulator
I wonder, how these 3 different PSUs will perform with about 100mA on their outputs!
I need to be patient for about 4 weeks, until I will receive the etched PCBs.
Kind regards - Rudi_Ratlos
Attachments
Its said to be constant Impedance is the king but im more worried about the noise levels need it to handle very sensitive currents so it has to be very very quite and hence got attracted to the TPS7A but overall its said shunt regs work in class A either you pull 50ma or 1A they are meant to be work in class A though. This angle was not thought earlier.
If salas replies on this It would be very helpful for the salas fans...
If salas replies on this It would be very helpful for the salas fans...
@rhythmsandy:
So: you are a SALAS - fan!
I am not (!) and will therefore withdraw from this thread.
Best regards - Rudi
So: you are a SALAS - fan!
I am not (!) and will therefore withdraw from this thread.
Best regards - Rudi
Hi Rudi, Maybe a bit off topic (sorry) but what about the FC100 FE shunt? It works very well in the FC100. Maybe you could substitute 18v zener and remodel for 18v? You already have the artwork and maybe even some spare boards!
There is many ways to skin a cat...
The Salas Reflektor D is a shunt reg dedicated to digital circuits only.
It has been built hundreds of times (read the thread) and has returned very favourable results for many. Still it is not the only way to power your DAM R2R 😉 (you would have to use the Salas BiB to get 10V out).
Juggling with sheer numbers it pretty pointless anyway... at what load and what voltage did they arrive at the number of 4uV? Will you be able to hear a difference? There is dozens of threads about power supplies.
As Rudi has pointed out you can use a Jung based design (the Diyaudio Super Regulator in the store), a Sjostrom Audio reg, a DIY LTxxx or many others. I am currently using a simple DC supply (LM317/337 based) to power the DAM... it is as deadsilent as it can be. Will be replaced with the Jung style reg just for peace of mind in the final case.
The Salas Reflektor D is a shunt reg dedicated to digital circuits only.
It has been built hundreds of times (read the thread) and has returned very favourable results for many. Still it is not the only way to power your DAM R2R 😉 (you would have to use the Salas BiB to get 10V out).
Juggling with sheer numbers it pretty pointless anyway... at what load and what voltage did they arrive at the number of 4uV? Will you be able to hear a difference? There is dozens of threads about power supplies.
As Rudi has pointed out you can use a Jung based design (the Diyaudio Super Regulator in the store), a Sjostrom Audio reg, a DIY LTxxx or many others. I am currently using a simple DC supply (LM317/337 based) to power the DAM... it is as deadsilent as it can be. Will be replaced with the Jung style reg just for peace of mind in the final case.
Alas.
Rudi what you're planning -- the test of 3 PSUs -- is most interesting. I hope you'll change your mind about withdrawing from this thread & post your data when the boards come. I'm amusing myself designing a shunt in LTSpice at the moment, and am very keen to see your results.
If you put them in another thread please could you post a link to it here.
P.S. I'm not anyone here's enemy or fan.
P.P.S. Belleson have compared thier regulator to some others, and published oscillographs here:
Superpower Super Regulator by Belleson
Before anyone start building this (you never now) a little adjustement.
Just to prevent that he (or she) is getting Ratlos too 😀
Mona
Attachments
Well then, Gentlemen: I received a couple of the PCBs shown in my post #2 several days ago, cut them and started with the assembly of the LT1763 - PCB.
Find its schematics attached in image 1, its layout in image 2 and the soldered PCB in image 3.
I did a very "quick and dirty" layout, used a potentiometer to adjust the output-voltage (10VDC for use with the Nelson Pass B1 - Buffer - the adjustment should be done with a fixed resistor! - I know)
and soldered the PCB within 20 minutes.
I connected the PCB to a 9VAC transformer and just wanted to know, if it is adjustable, ...
It is!
The performance of the LT1763 LDO regulator is shown in image 4.
I did not yet insert a load (for example: 100 Ohm resistor) on the output.
One reason for this: I do not know, how to measure the performance of the LT1763.
My old Tektronix 365B oscilloscope has a minimum scale of 5mV per division.
So: how will I be able to measure some µV ripple?
I have told you in my post #5 that I am not a SALAS - shunt fan.
I appreciate SALAS very much for all of his distributions to this forum.
But: using a SALAS - shunt for the B1-buffer as published in his B1-Hypnotize thread - the B1-buffer only needs a couple of mA - means to me:
to "shoot with canonballs on sparrows (German manner of speaking, meaning: to take a sledgehammer to crack a nut )".
Best regards - Rudi_Ratlos
P.S.: the LT1763 seems so easily to be used: I wonder, if I will go on and find a "better, small current voltage regulator".
Find its schematics attached in image 1, its layout in image 2 and the soldered PCB in image 3.
I did a very "quick and dirty" layout, used a potentiometer to adjust the output-voltage (10VDC for use with the Nelson Pass B1 - Buffer - the adjustment should be done with a fixed resistor! - I know)
and soldered the PCB within 20 minutes.
I connected the PCB to a 9VAC transformer and just wanted to know, if it is adjustable, ...
It is!
The performance of the LT1763 LDO regulator is shown in image 4.
I did not yet insert a load (for example: 100 Ohm resistor) on the output.
One reason for this: I do not know, how to measure the performance of the LT1763.
My old Tektronix 365B oscilloscope has a minimum scale of 5mV per division.
So: how will I be able to measure some µV ripple?
I have told you in my post #5 that I am not a SALAS - shunt fan.
I appreciate SALAS very much for all of his distributions to this forum.
But: using a SALAS - shunt for the B1-buffer as published in his B1-Hypnotize thread - the B1-buffer only needs a couple of mA - means to me:
to "shoot with canonballs on sparrows (German manner of speaking, meaning: to take a sledgehammer to crack a nut )".
Best regards - Rudi_Ratlos
P.S.: the LT1763 seems so easily to be used: I wonder, if I will go on and find a "better, small current voltage regulator".
Attachments
Last edited:
Happy to see you back Rudi_Ratlos. Thanks for sharing the first results of your work.
5mV/div, yes. My Rigol does a little better at 2mV/div (in 32 A to D steps/div). But neither is anywhere near good enough for meaningful noise comparisons.
To get a noise measurement in micro-volts per root Hertz a (low noise) pre-amp is needed between the regulator and the scope. This will, of course, limit the bandwidth to that of the pre-amp, which could be less than the “normal” 20MHz power supply noise measurement bandwidth.
However, to compare the noise levels of 3 regulators may be easier. Why not just AC couple the regulator output to an input on your hi-fi, turn up the volume, and listen to the noise. This test won't yield any numerical output, but could be quite valid as a way of comparing the noise levels of different regulators.
Another common test of regulator performance is to load the regulator with a square wave current load, while viewing the regulator output voltage on a scope. Regulator designs vary greatly in their ability to hold the output voltage steady during rapid changes of load current. Seeing the extent to which the output voltage does (or doesn't) wander around as the load current changes can be very telling. This is the type of test that Belleson has published for quite a few regulator designs (not just their own) here:
Superpower Super Regulator by Belleson
I'm guessing that (at least for line level use) all 3 regulators will be more than adequately quiet, making the regulators response to changes in load current more sonically significant than the noise. But it is just a guess, I've been wrong plenty of times before.
P.S. The usual translation of the German expression is “quail shooting with cannon”.
P.P.S. Hope you did Ketje's fix to the super regulator circuit.
Last edited:
Hi Rudi,
Last year, I've built an LT1963 power supply for a CALVIN buffer. To stay below the maximum 20V input voltage of this regulator, I used a "cap multiplier" front end with a protective zener in it's base.
All I can say, it's noise free and makes the buffer sound very good.
Curious you got hum. The 450µV you measure is wide band noise that is to be expected at 18V output. Given the TI datasheet, it's in the range of 0.003% of Vout.
The way you built your reg/pre-reg is fine. At 100Hz, even with an LM317/337, the rejection is high enough and you should not hear any hum.
Given the poor PSRR of this kind of buffer, I've also tried Jung's super_reg but with no audible difference to my ears. If you decide to go this route, like Ketje, I will point to some further adjustments or you will end up with high noise at the output.
SV1-1 must be connected to the positive output so as to keep AC gain unity.
SV1-2 must be connected to the negative output.
The lower leg of C312 must be connected to SV1-2 or it will filter nothing.
My scope is not better at estimating output noise but for sure, it was clean with both regs.
Thanks Sandy to make me just discovering TPS7A33 wich has a much higher max input voltage and is TO220.
Francis
Last year, I've built an LT1963 power supply for a CALVIN buffer. To stay below the maximum 20V input voltage of this regulator, I used a "cap multiplier" front end with a protective zener in it's base.
All I can say, it's noise free and makes the buffer sound very good.
Curious you got hum. The 450µV you measure is wide band noise that is to be expected at 18V output. Given the TI datasheet, it's in the range of 0.003% of Vout.
The way you built your reg/pre-reg is fine. At 100Hz, even with an LM317/337, the rejection is high enough and you should not hear any hum.
Given the poor PSRR of this kind of buffer, I've also tried Jung's super_reg but with no audible difference to my ears. If you decide to go this route, like Ketje, I will point to some further adjustments or you will end up with high noise at the output.
SV1-1 must be connected to the positive output so as to keep AC gain unity.
SV1-2 must be connected to the negative output.
The lower leg of C312 must be connected to SV1-2 or it will filter nothing.
My scope is not better at estimating output noise but for sure, it was clean with both regs.
Thanks Sandy to make me just discovering TPS7A33 wich has a much higher max input voltage and is TO220.
Francis
Hi Francis,
TransmissionAuqitain advised me to:
1) either "listen" to the hum that a PSU is producing.
But I do not dare to do this! The inputs of my 2 DIY - AMPs (SYMASYM, FC-100) are AC-coupled, but I am scared feeding one of my AMPs with a 10DC voltage and a small AC ripple on it.
2) build a circuit that switches on / off a PSU's load and measure the resulting waveform (that is, how I understood his proposal).
So I built a circuit consisting of a NE555 timer that produces a 50Hz square wave and fed it o the base of a MJE15034, whose collector is connected to the PSU's output and whose emitter is connected to a 100 Ohm resistor.
I am seeing a square-wave on the MJE's collector, no ripple, ...
Thank you for indicating another error of my Jung-Didden regulator PCB-design. This has been a very "quick and dirty" layout, I admit.
I did not yet solder it.
I am currently waiting for a TPS7A4700 - PCB (layout done be me, see attached image 1) and will first go on and evaluate this one.
Some weeks ago I have sent my Calvin-buffer PCB (image 2), which is driven by a LM3x7 pre-regulator to a friend of mine, who owns a "modern, up-to-date", cannot be compared with my 30 year old Tektronix oscilloscope)
and very powerful measurement equipment.
My PCB causes an awful hum, when it is used as a "line-buffer" in one of my friend's audio-setup.
I am quite sure that my used LM3x7 - pre-regulator is very well designed, but I have some doubts concerning Calvin's buffer.
His buffer is undoubtedly a gem when it comes to buffering a very low-level phono-signal, but I am afraid that it does not behave as designed when it comes to buffering a 100mV CDP - attenuated signal (being used as a "line-buffer").
I hope that I am right.
Else: I apologize for this assumption very much, Calvin!
I need to wait for the results of my friend.
Best regards - Rudi_Ratlos
TransmissionAuqitain advised me to:
1) either "listen" to the hum that a PSU is producing.
But I do not dare to do this! The inputs of my 2 DIY - AMPs (SYMASYM, FC-100) are AC-coupled, but I am scared feeding one of my AMPs with a 10DC voltage and a small AC ripple on it.
2) build a circuit that switches on / off a PSU's load and measure the resulting waveform (that is, how I understood his proposal).
So I built a circuit consisting of a NE555 timer that produces a 50Hz square wave and fed it o the base of a MJE15034, whose collector is connected to the PSU's output and whose emitter is connected to a 100 Ohm resistor.
I am seeing a square-wave on the MJE's collector, no ripple, ...
Thank you for indicating another error of my Jung-Didden regulator PCB-design. This has been a very "quick and dirty" layout, I admit.
I did not yet solder it.
I am currently waiting for a TPS7A4700 - PCB (layout done be me, see attached image 1) and will first go on and evaluate this one.
Some weeks ago I have sent my Calvin-buffer PCB (image 2), which is driven by a LM3x7 pre-regulator to a friend of mine, who owns a "modern, up-to-date", cannot be compared with my 30 year old Tektronix oscilloscope)
and very powerful measurement equipment.
My PCB causes an awful hum, when it is used as a "line-buffer" in one of my friend's audio-setup.
I am quite sure that my used LM3x7 - pre-regulator is very well designed, but I have some doubts concerning Calvin's buffer.
His buffer is undoubtedly a gem when it comes to buffering a very low-level phono-signal, but I am afraid that it does not behave as designed when it comes to buffering a 100mV CDP - attenuated signal (being used as a "line-buffer").
I hope that I am right.
Else: I apologize for this assumption very much, Calvin!
I need to wait for the results of my friend.
Best regards - Rudi_Ratlos
Attachments
Last edited:
Rudi,
The sense node of your regulators are connected at the wrong place.
You've put them where high charging current pulses occur and they are too far away from each other.
Cut two traces, add two straps and it should be way better.
Good luck,
Francis
The sense node of your regulators are connected at the wrong place.
You've put them where high charging current pulses occur and they are too far away from each other.
Cut two traces, add two straps and it should be way better.
Good luck,
Francis
Surely you don't expect your scope alone to show you the relative performance of very good regulators. At a minimum you will need an instrument amplifier. You might want to take a look at the work Preamp did in this thread.
I just ordered an AC millivoltmeter for $100 + $40 shipping from ebay
New Dual 2 Channel AC Millivoltmeter Milli Voltmeter Audio Stereo Hi Fi Tuning | eBay
It's a 2 channel unit with a bandwidth of 10Hz to 1MHz, and 0.3mv FSD at maximum sensitivity. It works by using a pre-amp to drive the meter, and makes each channel's pre-amp output available on BNCs at 0.1v FSD. That equates to a gain of 333 at maximum sensitivity.
As well as being a meter, I'm expecting to use its pre-amp as an instrument amplifier for my scope (albeit with a bandwidth of only 1MHz). With a gain of 333, 5mv per div becomes 15uv per div, 2mv per div becomes 6uv per div, 1mv per div becomes 3uv per div, etc. Making regulator performance measurement much easier.
It should arrive by mid August. When it comes I'll test it with a signal generator & a scope, and report on how well it works as an instrument amplifier.
New Dual 2 Channel AC Millivoltmeter Milli Voltmeter Audio Stereo Hi Fi Tuning | eBay
It's a 2 channel unit with a bandwidth of 10Hz to 1MHz, and 0.3mv FSD at maximum sensitivity. It works by using a pre-amp to drive the meter, and makes each channel's pre-amp output available on BNCs at 0.1v FSD. That equates to a gain of 333 at maximum sensitivity.
As well as being a meter, I'm expecting to use its pre-amp as an instrument amplifier for my scope (albeit with a bandwidth of only 1MHz). With a gain of 333, 5mv per div becomes 15uv per div, 2mv per div becomes 6uv per div, 1mv per div becomes 3uv per div, etc. Making regulator performance measurement much easier.
It should arrive by mid August. When it comes I'll test it with a signal generator & a scope, and report on how well it works as an instrument amplifier.
Thanks for publishing the new work. No ripple is good.
The voltage across the 100R resistor in the MJE's emitter divided by its value gives the regulator load current. It will, of course, be a square wave varying between 2 different load currents as the 555's output goes up and down.
The experiment I had in mind goes something like this. View the square wave voltage across the resistor in the MJE's emitter on 1 scope channel, and trigger from this channel. Put the regulator output voltage on the scope's other channel. Look to see how much the regulator's output voltage's disturbed by the change in load current at the rising and falling edges in the square wave voltage across the resistor in the MJE's emitter.
It sounds like the square wave you're seeing is just such a disturbance of the regulator's output, which should, of course, remain at a constant voltage. Will the other regulator's show a larger or smaller disturbance? Will they have ripple? A different disturbance waveshape? These are the kinds regulator comparison questions the experiment intends to answer. I bet they don't all perform the same.
Last edited:
What should be the problem with 10V DC when you AC couple?
My PCB causes an awful hum, when it is used as a "line-buffer" in one of my friend's audio-setup.
I am quite sure that my used LM3x7 - pre-regulator is very well designed, but I have some doubts concerning Calvin's buffer.
His buffer is undoubtedly a gem when it comes to buffering a very low-level phono-signal, but I am afraid that it does not behave as designed when it comes to buffering a 100mV CDP - attenuated signal (being used as a "line-buffer").
I hope that I am right.
Else: I apologize for this assumption very much, Calvin!
Guess there is something wrong with your setup. I build a linestage with my own "all SMD" Version (as per Calvin's original Design) and it is dead quiet with low THD up to several Vrms into high loads. The buffer works extremely well for line level applications!
kind regards, Daniel
I build 3 power supplies based on the TPS7A4700.
1) 5V LiPo powered version
2) 230V<>5V version
3) 230V<>12V version
Here are the average measurement results, measured with a HP3561A.
5V LiPo powered version
Voltage ripple: 460 nVrms
Noisefloor: -137 dBV
230V<>5V version
Voltage ripple: ~760nVrms
Noisefloor: -122 dBV
230V<>12V version
Voltage ripple: ~3μVrms
Noisefloor: -130 dBV
http://www.diyaudio.com/forums/vend...w-noise-ldo-regulator-pcb-12.html#post4388982
1) 5V LiPo powered version
2) 230V<>5V version
3) 230V<>12V version
Here are the average measurement results, measured with a HP3561A.
5V LiPo powered version
Voltage ripple: 460 nVrms
Noisefloor: -137 dBV
230V<>5V version
Voltage ripple: ~760nVrms
Noisefloor: -122 dBV
230V<>12V version
Voltage ripple: ~3μVrms
Noisefloor: -130 dBV
http://www.diyaudio.com/forums/vend...w-noise-ldo-regulator-pcb-12.html#post4388982
- Status
- Not open for further replies.