Jeff Young
pfm Member
Yep. The HackerCAP is CRCRC.
[Or at least it looks like it from the board. I don't see a schematic for it.]
[Or at least it looks like it from the board. I don't see a schematic for it.]
Starfish revival
- Thread starter jpk
- Start date
Yank
Bulbous Also Tapered
It can also be CLCLC.
PCB layout:
I added bypass caps and tried to avoid crossings. But I am not sure about the best way to connect the power supplies to the audio circuits, see the following diagram:
A) standard Naim way with 27R and 47uF
B) recommended by Hacker in this post
C) recommended by Andy Weekes, see here
D) NAC102 mod by Neil Jadman
E) the arrangment I now have in my layout
I want to use my regulator boards (as shown at the beginning of this thread) and later might upgrade to super regs. What is the most flexible and best sounding solution? Also is it a good idea to add 33nF films to the signal decoupling caps?
I added bypass caps and tried to avoid crossings. But I am not sure about the best way to connect the power supplies to the audio circuits, see the following diagram:
A) standard Naim way with 27R and 47uF
B) recommended by Hacker in this post
C) recommended by Andy Weekes, see here
D) NAC102 mod by Neil Jadman
E) the arrangment I now have in my layout
I want to use my regulator boards (as shown at the beginning of this thread) and later might upgrade to super regs. What is the most flexible and best sounding solution? Also is it a good idea to add 33nF films to the signal decoupling caps?
Last edited:
Andrew L Weekes
Reverse Engineer
I have sometimes used air-cored chokes in place of the 27R in the Naim circuit, damped with a parallel resistor to avoid any high-Q effects. That, in conjunction with the scheme shown in C above *using a film capacitor*, is very effective. The series 0.5R is to negate any high-Q effects of having a film capacitor in there.
Any reactive components (e.g. capacitors) on the output of the regulator have a direct effect on its stability, especially with capacitors that can form a high Q resonant circuit with the various stray reactance in the circuit and its layout. By using a film cap and a small resistor, you can achieve a nice wide bandwidth, low impedance, decoupling effect, but without risk of oscillation and resonance at certain frequencies. Standard electrolytics achieve a similar effect via their ESR, but since this internal resistance is non-linear and prone to variation with age and time, a film cap and resistor provides a closer to perfect solution. The film cap won't age, will work better at high frequencies, and the resistor is as close to ideally linear as you can get.
Polyester film capacitors are available in small sizes with values of several uF and whilst not optimal for signal paths, are way better than any electrolytic for PSU decoupling.
The local filtering provides two functions - it compensates for both the regulator's reduction in performance as it's open loop gain drops off with increasing frequency and also the series inductance of the wiring or PCB traces, which do the same. It also keeps any high frequency currents from the circuit being powered in a tight local loop, rather than radiating from the whole PSU wiring. The latter is more prevalent with high speed digital circuits, but it's good practice anyway to decouple locally.
Andy.
Any reactive components (e.g. capacitors) on the output of the regulator have a direct effect on its stability, especially with capacitors that can form a high Q resonant circuit with the various stray reactance in the circuit and its layout. By using a film cap and a small resistor, you can achieve a nice wide bandwidth, low impedance, decoupling effect, but without risk of oscillation and resonance at certain frequencies. Standard electrolytics achieve a similar effect via their ESR, but since this internal resistance is non-linear and prone to variation with age and time, a film cap and resistor provides a closer to perfect solution. The film cap won't age, will work better at high frequencies, and the resistor is as close to ideally linear as you can get.
Polyester film capacitors are available in small sizes with values of several uF and whilst not optimal for signal paths, are way better than any electrolytic for PSU decoupling.
The local filtering provides two functions - it compensates for both the regulator's reduction in performance as it's open loop gain drops off with increasing frequency and also the series inductance of the wiring or PCB traces, which do the same. It also keeps any high frequency currents from the circuit being powered in a tight local loop, rather than radiating from the whole PSU wiring. The latter is more prevalent with high speed digital circuits, but it's good practice anyway to decouple locally.
Andy.
Andy, thanks a lot! So I will try the following:
I wonder if the solution in this diagram would suit only if the preamp (or DAC or whatever) is powered from the ALWSR, or would that also be good for powering from a LM317 based supply with the usual 22~47uF at the output?
In this post you suggest using 3.9uH with 1 Ohm for the choke and mention that winding some enamelled copper wire around a drill bit would suffice. Does the diameter of the wire matter?
I wonder if the solution in this diagram would suit only if the preamp (or DAC or whatever) is powered from the ALWSR, or would that also be good for powering from a LM317 based supply with the usual 22~47uF at the output?
In this post you suggest using 3.9uH with 1 Ohm for the choke and mention that winding some enamelled copper wire around a drill bit would suffice. Does the diameter of the wire matter?
Andrew L Weekes
Reverse Engineer
Andy, thanks a lot! So I will try the following:
Unless I've misread that you want the choke and the resistor between the regulator output and the 2u2 / 0R5, assumign the latter is taking the place of the 47u used conventionally on Naim circuits. The choke and damping resistor is functionally equivalent to the 27R.
It's worth having a look at learning some of the SPICE simulation tools (and looking at the coil32 tool mentioned above) so you can model this and play with values to see the effects. Wire diameter largely determines current carrying capability and series DC resistance, which raises the PSU impedance at low frequencies. I like Simetrix as a simple easy to use SPICE simulator, with reasonably generous node limits, but there's lots of full featured free versions out there, like LT spice or even Micro-Cap, which used to retail for $4500 and is now free.
Thanks! Thanks also for your pre-built starfish which I managed to take some pictures of:
Seems it's a full build with symmetric supplies and a small yellow "nert" PCB (not shown on these photos):
As stated in BOM v1.2 C112 and C114 (input and feedback caps of the gain circuit) are wire links...
Just did some tests, everything seems to work fine. I used my old Philips PE1542 bench power supply, a Rigol scope, and a diy signal generator made from left overs (the black blox with the red LED). The signal generator uses a ATTINY84 controlling a AD9850 DDS chip amplified by a TBA820M opamp, powered by a 9V battery:
DC offset at the output of the Starfish was <1mV, power consumption around 300mA. The small yellow PCB between the Rigol and the Starfish is the NERT circuit, it was hooked up and in use. The Traco got warm but not hot. Here is a screenshot from the scope at full volume:
Last edited:
Thanks for the input! I also thought to take out the Traco: can I just desolder it and connect the super regs instead? I don't have the full schematic, only the one shown in the build manual: do you have the separate full schematic of the Starfish?
I see. But what happens to the NERT in this case?
OK, got it, many thanks!
