Except........... 15 x 1,414 = 21 volts......Loss approx. 2 volt........ Which means feeding +/- 19 volt to Pearls 7815/7915.
Which gives excellent result 😉
Sure.. my results are purely virtual, yours are real world and much more valid... It is just to show differences..
Here same scenario with indicators instead, 20mh 1.5 ohm dc resistance..
Replace the 1 Ω-per-block resistor string with a single 20 mH choke (DCR ≈ 1.5 Ω) and you get a classic C-L-C filter:
• Ripple kill: At 45 kHz the choke’s reactance is ~5.6 kΩ, so the 200 mV noise is divided down by about a million (-120 dB). Practically silent.
• DC loss: Only the copper matters. 0.2 A × 1.5 Ω ≈ 0.30 V drop per rail. Feed the board ±16 V and you still land a clean ±15 V.
• In-rush: Inductor throttles surge current naturally—no “arc-welder” start-up like the pure cap bank.
• Low--freq quirk: 20 mH with 8 mF resonates near 12 Hz; damping ζ ≈ 0.5. It rings a touch, but a tiny 0.1 Ω series resistor or an RC snubber will hush it.
Bottom line: same ripple-obliteration as the big resistor ladder, one-fifth of the voltage loss, and almost no heat. If the coil fits and doesn’t hum, it’s the sweet spot.
Here same scenario with indicators instead, 20mh 1.5 ohm dc resistance..
Replace the 1 Ω-per-block resistor string with a single 20 mH choke (DCR ≈ 1.5 Ω) and you get a classic C-L-C filter:
• Ripple kill: At 45 kHz the choke’s reactance is ~5.6 kΩ, so the 200 mV noise is divided down by about a million (-120 dB). Practically silent.
• DC loss: Only the copper matters. 0.2 A × 1.5 Ω ≈ 0.30 V drop per rail. Feed the board ±16 V and you still land a clean ±15 V.
• In-rush: Inductor throttles surge current naturally—no “arc-welder” start-up like the pure cap bank.
• Low--freq quirk: 20 mH with 8 mF resonates near 12 Hz; damping ζ ≈ 0.5. It rings a touch, but a tiny 0.1 Ω series resistor or an RC snubber will hush it.
Bottom line: same ripple-obliteration as the big resistor ladder, one-fifth of the voltage loss, and almost no heat. If the coil fits and doesn’t hum, it’s the sweet spot.
Hi, just a short demo because I like very much how the Pearl sounds:
Except........ 78xx/79xx don´t work (supression wise) with 1 volt difference.
My experience is, that 2,5 volt is sufficient 90% of the times. This depends on the manufacturer of the regulator..
3 volts more on input, and you´re home free
Sad to say it , but i dont have a nice stereo like that... Half finished amps, cables its a mess
Speaking of ac suppression... I just bought theese two bad boys..
50mh..3A.. roughly 20 kilo each.
50mh..3A.. roughly 20 kilo each.
So this is a Luxsonix TR-009 discrete dual opamp (~$42 in Japan).
Which other opamps did you try?
Not impressive compared to lm49720.
Luxsonix TROP-009 Discrete Audio Opamp
Key Features:
Discrete operational amplifier optimized for audio applications.
FET input stage ensuring high input impedance.
Stable at unity gain (0 dB).
Specifications:
Operating Voltage: ±5V to ±15V
Recommended Gain: 0 dB (unity) to +26 dB (non-inverting)
THD+N: ≤ 0.03% (at clipping power, 20Hz–20kHz, 6 dB gain, 600Ω load)
Residual Noise: ≤ 1.5 µV (A-weighted, input shorted, unity gain)
Gain-Bandwidth Product (fT): 2 MHz
Slew Rate: ±10 V/µs (at ±12V supply)
Output DC Offset: ≤ ±50 mV (at unity gain)
Current Consumption:
11 mA typ. (±12V su
pply)
8 mA typ. (±5V supply)
Luxsonix TROP-009 Discrete Audio Opamp
Key Features:
Discrete operational amplifier optimized for audio applications.
FET input stage ensuring high input impedance.
Stable at unity gain (0 dB).
Specifications:
Operating Voltage: ±5V to ±15V
Recommended Gain: 0 dB (unity) to +26 dB (non-inverting)
THD+N: ≤ 0.03% (at clipping power, 20Hz–20kHz, 6 dB gain, 600Ω load)
Residual Noise: ≤ 1.5 µV (A-weighted, input shorted, unity gain)
Gain-Bandwidth Product (fT): 2 MHz
Slew Rate: ±10 V/µs (at ±12V supply)
Output DC Offset: ≤ ±50 mV (at unity gain)
Current Consumption:
11 mA typ. (±12V su
pply)
8 mA typ. (±5V supply)
I tried muses05 x2, muses01, muses02,njm5532,Burson V6classic
trop009 has best bass,3D expression,natural sound and thicker and better colouration in sound.
I surprised difference compared to other opamps.
Price is around 35usd each.
It's worth to add collection.
trop009 has best bass,3D expression,natural sound and thicker and better colouration in sound.
I surprised difference compared to other opamps.
Price is around 35usd each.
It's worth to add collection.
Currently I'm using the "standard" CRCRC filtering with 4700uF/4 ohms. If I want to expand this with two more RC elements, I think I'll have to reduce the R - so instead of the current 2x4 ohms, 4x2 ohms? Or is that calculation too simplistic?
Keep 4 Ω per section: Ripple attenuation doubles but you sacrifice another 1.6 V and burn 0.32 W. It only makes sense if you have plenty of raw DC headroom
Switch to 2 Ω per section: The total voltage drop stays the same as now, while attenuation rises by “only” ~19 dB. Still an improvement, and each resistor runs cooler (0.08 W versus 0.16 W).
Or use a indicator 10 or 40mh with dc resistance of 1 to 3 ohm, this will get you even lower ripple rejection, and your voltage doesent drop as much... Choices choices.. it all depends on your dc voltage headroom, your 7815 7915 work best with an overvoltage of at least 2v. Got any to spare ?
Physics dont bend its all a matter of compromise.. for what it worth, theese improvements might only show up on measurements, your perception might dictate otherwise..
Switch to 2 Ω per section: The total voltage drop stays the same as now, while attenuation rises by “only” ~19 dB. Still an improvement, and each resistor runs cooler (0.08 W versus 0.16 W).
Or use a indicator 10 or 40mh with dc resistance of 1 to 3 ohm, this will get you even lower ripple rejection, and your voltage doesent drop as much... Choices choices.. it all depends on your dc voltage headroom, your 7815 7915 work best with an overvoltage of at least 2v. Got any to spare ?
Physics dont bend its all a matter of compromise.. for what it worth, theese improvements might only show up on measurements, your perception might dictate otherwise..
In my humble opinion, you need to seek very low esr and impedance close to your opamps rail pins , 1uf and 100nf, or higher, c0g are probably the best right at the opamp pin.
I used 22uf oscons, they are polymer and low esr, i had a bunch laying around.
You have your bulk supply, and that connects to pearl board via cable, cable and traces have inductances.
Use heavy gauge cabeling, short as possible, avoid screw terminals.
In chassis have the pearl boards far away as possible from any ac such as transformer, of you have a oscilloscope you can move transformer around and observe noise pickup.
If you use film capacitors, they are not polarized, but the do have a optimal orientation, if its an axial , its inner Windings must go to highest impedance om pcb, outer winding to lowest , you can see the effect on oscilloscope, the can and will act as antennas, notable in high gain circuits.
Did i mention twisted cabeling, ferrite beads etc.
I used 22uf oscons, they are polymer and low esr, i had a bunch laying around.
You have your bulk supply, and that connects to pearl board via cable, cable and traces have inductances.
Use heavy gauge cabeling, short as possible, avoid screw terminals.
In chassis have the pearl boards far away as possible from any ac such as transformer, of you have a oscilloscope you can move transformer around and observe noise pickup.
If you use film capacitors, they are not polarized, but the do have a optimal orientation, if its an axial , its inner Windings must go to highest impedance om pcb, outer winding to lowest , you can see the effect on oscilloscope, the can and will act as antennas, notable in high gain circuits.
Did i mention twisted cabeling, ferrite beads etc.
Last edited:
Has anybody, questioned the ancient 7815-7915 ? ... Many of us have spend serious cash on low noise psu, thats feeds theese dinosaurs,
I would look into LT3045/LT3094 .
7815 / 7915: about 60 µV RMS over 10 Hz–100 kHz, noise density ≈ 25 nV/√Hz at 10 kHz.
LT3045 / LT3094: about 0.8 µV RMS, noise density ≈ 2 nV/√Hz at 10 kHz.
Impedance is also much improved,
The old 7815/7915 pair present roughly twenty-milliohm output impedance at 1 kHz and climb fast with frequency: a few dozen milliohms by 10 kHz and well into the half-ohm-to-one-ohm range around 100 kHz.
The LT3045 and its negative twin LT3094 stay in the low-milliohm bracket almost all the way up to RF: about one milliohm through the entire audio band and only a few milliohms at 100 kHz.
Looks like a nobrainer, and they come ready made on discrete to220 package, drop in replacement for 7815-7915.
Today om 49 years old, that means i just dropped my diapers when 7815 7915 was introduced.
I would look into LT3045/LT3094 .
7815 / 7915: about 60 µV RMS over 10 Hz–100 kHz, noise density ≈ 25 nV/√Hz at 10 kHz.
LT3045 / LT3094: about 0.8 µV RMS, noise density ≈ 2 nV/√Hz at 10 kHz.
Impedance is also much improved,
The old 7815/7915 pair present roughly twenty-milliohm output impedance at 1 kHz and climb fast with frequency: a few dozen milliohms by 10 kHz and well into the half-ohm-to-one-ohm range around 100 kHz.
The LT3045 and its negative twin LT3094 stay in the low-milliohm bracket almost all the way up to RF: about one milliohm through the entire audio band and only a few milliohms at 100 kHz.
Looks like a nobrainer, and they come ready made on discrete to220 package, drop in replacement for 7815-7915.
Today om 49 years old, that means i just dropped my diapers when 7815 7915 was introduced.
Last edited:
Thanks for the confirmation, I will go with 2.x ohms.
Regarding LT3045, technically it is great, but I read some reviews and they state that the sound can be influenced negatively