nuryev, here's a picture.
It's all point to point wired, except for the super regulator boards. It uses a differential input with LME49990s, into an OPA627 and then into a Burson discrete using active filtering. It has 62dB of gain for my Ortofon Vivo Blue cartridge.
The patchcords shown are for burn in purposes. I built inverse RIAA networks into them which also reduce the amplitude of my Duotech down to about 4.5mv. I burned in the unit for 10 days before I started listening.
As you can see, i use the super regulator boards for each channel. The transformer is oversized because it was recycled from a 1993 project. I grounded the chassis through a resistor and .1uf cap, which improved the noise and hum slightly. I thought I might have to go to a balanced cable with XLR connection because of the differential input, but hum has not been a problem unless I use a shielded (unbalanced) cable, putting all the noise onto one side of the differential input.
It's sounding quite nice at the moment and is the first opamp stage I've built which beats what I was using.
It's all point to point wired, except for the super regulator boards. It uses a differential input with LME49990s, into an OPA627 and then into a Burson discrete using active filtering. It has 62dB of gain for my Ortofon Vivo Blue cartridge.
The patchcords shown are for burn in purposes. I built inverse RIAA networks into them which also reduce the amplitude of my Duotech down to about 4.5mv. I burned in the unit for 10 days before I started listening.
As you can see, i use the super regulator boards for each channel. The transformer is oversized because it was recycled from a 1993 project. I grounded the chassis through a resistor and .1uf cap, which improved the noise and hum slightly. I thought I might have to go to a balanced cable with XLR connection because of the differential input, but hum has not been a problem unless I use a shielded (unbalanced) cable, putting all the noise onto one side of the differential input.
It's sounding quite nice at the moment and is the first opamp stage I've built which beats what I was using.
An externally hosted image should be here but it was not working when we last tested it.
Just now I re-derived it by hand, and fresh circuit analysis agrees with foggy old memory
Transconductance gm gets bigger and bigger, the more you increase transistor current. In BJTs, gm is proportional to (Ice raised to the power 1.0), in JFETs gm is proportional to (Ids raised to the power 0.5), and in MOSFETs gm is proportional to (Ids raised to the power ~ 0.4). So in every case we can increase gm and decrease {open loop output impedance} simply by increasing the current that flows thru the output transistor. And as I derived just this morning, {closed loop output impedance} is linearly proportional to {open loop output impedance}. If you decrease open loop output impedance by 75%, you also decrease closed loop output impedance by the same 75%.
Amplifier gain "Av" is just as important. {closed loop output impedance} is linearly proportional to (1/Av). Increasing Av by a factor of three, decreases {closed loop output impedance} by the same factor of three. So if your goal is low output impedance, what you want is an opamp with truly enormous gain-bandwidth product; this gives high Av at high frequencies. You also want an opamp with truly enormous DC gain; this gives high Av at low frequencies. In other words, you want the AD797.
However it is not universally agreed upon that lower and lower output impedance gives greater and greater listening pleasure, without limit. Correlation between (listening pleasure) and (output impedance) is not strong.
{closed loop output impedance} = {open loop output impedance} / (1 + LoopGain)
If the output transistor is operating as a source follower or emitter follower, as in the Jung/Didden Super Regulator, {open loop output impedance) equals (1/gm).Transconductance gm gets bigger and bigger, the more you increase transistor current. In BJTs, gm is proportional to (Ice raised to the power 1.0), in JFETs gm is proportional to (Ids raised to the power 0.5), and in MOSFETs gm is proportional to (Ids raised to the power ~ 0.4). So in every case we can increase gm and decrease {open loop output impedance} simply by increasing the current that flows thru the output transistor. And as I derived just this morning, {closed loop output impedance} is linearly proportional to {open loop output impedance}. If you decrease open loop output impedance by 75%, you also decrease closed loop output impedance by the same 75%.
Amplifier gain "Av" is just as important. {closed loop output impedance} is linearly proportional to (1/Av). Increasing Av by a factor of three, decreases {closed loop output impedance} by the same factor of three. So if your goal is low output impedance, what you want is an opamp with truly enormous gain-bandwidth product; this gives high Av at high frequencies. You also want an opamp with truly enormous DC gain; this gives high Av at low frequencies. In other words, you want the AD797.
However it is not universally agreed upon that lower and lower output impedance gives greater and greater listening pleasure, without limit. Correlation between (listening pleasure) and (output impedance) is not strong.
Good post Mark.
A few comments if I may: indeed the Gm of the pass device (and with it changing, the closed loop characteristics also change) increase with current load. That is also the reason that sometimes a borderline stable reg is stable at low current but not with high load current.
I agree that audibility is only weakly related to Zout, but is is stronger related to the (pre)amp you drive from it. A simple preamp with very little or no power supply rejection (tech term PSRR) is more sensitive to what happens at the supply lines, so will have more audible effects from changes in the reg.
An opamp-based (pre)amp has almost by definition extremely large supply noise and ripple rejection and in such a case, if changing the reg opamp gives audible changes, something is broken or oscillates.
Jan
A few comments if I may: indeed the Gm of the pass device (and with it changing, the closed loop characteristics also change) increase with current load. That is also the reason that sometimes a borderline stable reg is stable at low current but not with high load current.
I agree that audibility is only weakly related to Zout, but is is stronger related to the (pre)amp you drive from it. A simple preamp with very little or no power supply rejection (tech term PSRR) is more sensitive to what happens at the supply lines, so will have more audible effects from changes in the reg.
An opamp-based (pre)amp has almost by definition extremely large supply noise and ripple rejection and in such a case, if changing the reg opamp gives audible changes, something is broken or oscillates.
Jan
I couldn't get the LT1028 to operate properly with the SR. Mbe it's just me.
The first sentence in the datasheet "Description" is weasel-ey. They are trying to tell you NOT to use the LT1028 in noninverting amplifiers with low gain; sadly the Jung/Didden Super Regulator is a noninverting amplifier with low gain. Linear Technology are trying to tell you to select the LT1128 instead; it is stable at gain=+1. But they don't want to print the word "unstable" on the datasheet, so the weasel phraseologists come to the rescue.
If you want to experiment with this opamp family in the Super Reg, I suggest either
- Use the LT1128, or
- Change the VREF diode to a 1.2V shunt reference IC like the LM285-1.2, which forces you to run the opamp + feedback network at much higher gain, which gives you plenty of phase margin even with the LT1028, or
- Connect a 220 pF capacitor from pin 5 to pin 6, as per the "Over Compensation" figure in the datasheet,
- or a combination of the above
Attachments
I tried the LME 49990s in the super regulators, since their specs are similar to the AD797 (0.9nV noise, 135dB gain, 22V/uS slew, 110MHz GBW). They oscillate badly!
I'm impressed with the performance using the AD825s. My phonostage sounds great. I listened to several more opamp combinations today and was amazed that they sounded so close. Using the LME 49990s for all the signal path the performance was excellent. Swapping in the Burson discretes and the OPA627 each brought a little more life to the music and was more enjoyable, though is close on an audiophile checklist. The LME49990s are excellent and have great bass and soundstaging. The OPA627s have less bass emphasis and more midrange emphasis and warmth (in my circuit).
I noticed that the Parts Connection has Dexa special edition discretes on sale for half price. I have one of the DEXA dual opamps and was impressed (better than any monolithic opamps I have).
Nice job on that super regulator, because my project sounds great!
I'm impressed with the performance using the AD825s. My phonostage sounds great. I listened to several more opamp combinations today and was amazed that they sounded so close. Using the LME 49990s for all the signal path the performance was excellent. Swapping in the Burson discretes and the OPA627 each brought a little more life to the music and was more enjoyable, though is close on an audiophile checklist. The LME49990s are excellent and have great bass and soundstaging. The OPA627s have less bass emphasis and more midrange emphasis and warmth (in my circuit).
I noticed that the Parts Connection has Dexa special edition discretes on sale for half price. I have one of the DEXA dual opamps and was impressed (better than any monolithic opamps I have).
Nice job on that super regulator, because my project sounds great!
Tried a couple other opamps
I just tried swapping out the AD825's for LME49710s and the AD8065. They both worked fine.
The LME49710s were maybe slightly more quiet, but I preferred the AD825s which sounded more lifelike, dynamic and enjoyable.
The AD8065 were the interesting one. Their specs looked promising for the task:
7nV noise
180 V/uS slew
120MHz GBW
55nS settling time
They have a maximum power supply of +/- 12V (which is fine for the circuit, since 15V is +/- 7.5V).
They sounded great! They resulted in music from my circuit which sounded more quiet, more natural, smoother and more detailed. My results were even more mesmerizing. If you're looking for an alternate opamp, these are worth a try. I did not put them on the scope, but there was no apparent oscillation.
I just tried swapping out the AD825's for LME49710s and the AD8065. They both worked fine.
The LME49710s were maybe slightly more quiet, but I preferred the AD825s which sounded more lifelike, dynamic and enjoyable.
The AD8065 were the interesting one. Their specs looked promising for the task:
7nV noise
180 V/uS slew
120MHz GBW
55nS settling time
They have a maximum power supply of +/- 12V (which is fine for the circuit, since 15V is +/- 7.5V).
They sounded great! They resulted in music from my circuit which sounded more quiet, more natural, smoother and more detailed. My results were even more mesmerizing. If you're looking for an alternate opamp, these are worth a try. I did not put them on the scope, but there was no apparent oscillation.
I couldn't even get the SR to "start" with the LT1028.
As they say, and as you aptly point out: "Read the prospectus carefully before you invest..."
+/- 17v out
Since the SilentSwitcher is not available for now, I am continuing my Superreg build. I need to get +/- 17v out. Please confirm if this is correct:
17+4=21. so I need to have at least 21v input on the Superreg
17V - 6.9V (this is the LM329 value) = "10.1"
10.1/6.9 = "Z"= 1.46
"Z" multiplied by R7 or R14 value = R6/R13 value
for
R7/R14 = 1K
R6/R13 = 1.46K
Since the SilentSwitcher is not available for now, I am continuing my Superreg build. I need to get +/- 17v out. Please confirm if this is correct:
17+4=21. so I need to have at least 21v input on the Superreg
17V - 6.9V (this is the LM329 value) = "10.1"
10.1/6.9 = "Z"= 1.46
"Z" multiplied by R7 or R14 value = R6/R13 value
for
R7/R14 = 1K
R6/R13 = 1.46K
a bit of smoke and confusion...
Some smoke appeared so I assume -- problem! Anything wrong with this picture?
I checked my pre-regulator and it's got +/-21 Vdc out, as needed. Hooked up the Superreg to it according to instructions: Ground is on the inside Vin J1 and J5.
To me this seems WRONG because looking at the traces Ground seems to be on the outside Vin of J1 and J5. Anyway...
So I followed the instructions, with Ground on inside pins of J1 and J5. Turned on and a little smoke came up, the LEDs did not go on. So I turned off immediately.
No load was connected to superreg.
Please give me an idea of where to start troubleshooting. Thanks you!
Herman
Some smoke appeared so I assume -- problem! Anything wrong with this picture?
I checked my pre-regulator and it's got +/-21 Vdc out, as needed. Hooked up the Superreg to it according to instructions: Ground is on the inside Vin J1 and J5.
To me this seems WRONG because looking at the traces Ground seems to be on the outside Vin of J1 and J5. Anyway...
So I followed the instructions, with Ground on inside pins of J1 and J5. Turned on and a little smoke came up, the LEDs did not go on. So I turned off immediately.
No load was connected to superreg.
Please give me an idea of where to start troubleshooting. Thanks you!
Herman