I am building a power amp based on Aussieamplifiers new module NXV200, with a PSU based on a 650VA 31VAC toroidal to get good 4ohm performance. I have a couple of questions regarding optimizing the PSU.
1. Having great respect for N Pass amplifiers I checked out his PSUs, and they are either CLC (Aleph) or CRC (FirstWatt). Would there be any benefit to use a CRC design for a class AB design, as compared to Pass designs that are class A ?
2. Some people bypass the last electrolytic cap with a film cap to get better high frequency performance or subjective improvement in "smoothness" of high frequencies. PSU designers claim that using a series resistor with the bypass cap to lower the Q and thus the risk for resonance is wise. Others call this to "snubberize" the PSU, and they use a fairly large resistor around 1R. Is this resistor really needed/beneficial with low capacitanc film caps with low inductance design ? Has anyone got better performance with film cap bypassing ?
I have read some threads regarding these topics, but I have not found any clear answers. Perhaps you have to try different combinations meassure performance..
1. Having great respect for N Pass amplifiers I checked out his PSUs, and they are either CLC (Aleph) or CRC (FirstWatt). Would there be any benefit to use a CRC design for a class AB design, as compared to Pass designs that are class A ?
2. Some people bypass the last electrolytic cap with a film cap to get better high frequency performance or subjective improvement in "smoothness" of high frequencies. PSU designers claim that using a series resistor with the bypass cap to lower the Q and thus the risk for resonance is wise. Others call this to "snubberize" the PSU, and they use a fairly large resistor around 1R. Is this resistor really needed/beneficial with low capacitanc film caps with low inductance design ? Has anyone got better performance with film cap bypassing ?
I have read some threads regarding these topics, but I have not found any clear answers. Perhaps you have to try different combinations meassure performance..
Hi!
First of all the Alephs did not use CLC, but AFAIK only C filtered PSUs. The CLC you've seen are probably from DIYer pcbs and not in the service manual.
CRC is a cheap way to get better filtering, at the cost of higher psu output impedance. For ClassA that's not a real issue, since they rather statically draw current, so voltage drop over the resistor stays about the same. Changes in total supply current are rather small as well (max 50% for (most) stock Alephs).
For ClassAB or just ClassB you get a modulated voltage rail depending on the load (speaker impedance) and dynamic range of the music you're listening to. That's plain distortion.
ClassAB does not need much filtering (due to comparable small currents), so go for pure C filtering, or CLC filtering if you can afford (and want to work with a large magnetic stray field), or voltage regulation.
Have fun, Hannes
First of all the Alephs did not use CLC, but AFAIK only C filtered PSUs. The CLC you've seen are probably from DIYer pcbs and not in the service manual.
CRC is a cheap way to get better filtering, at the cost of higher psu output impedance. For ClassA that's not a real issue, since they rather statically draw current, so voltage drop over the resistor stays about the same. Changes in total supply current are rather small as well (max 50% for (most) stock Alephs).
For ClassAB or just ClassB you get a modulated voltage rail depending on the load (speaker impedance) and dynamic range of the music you're listening to. That's plain distortion.
ClassAB does not need much filtering (due to comparable small currents), so go for pure C filtering, or CLC filtering if you can afford (and want to work with a large magnetic stray field), or voltage regulation.
Have fun, Hannes
Thank's for your replies. I assumed that CRC wasn't really a good option for class AB and you gave a good explanation. As you said CLC might get you into trouble because of magnetic fields as well as quite high cost for low R high current coils, so I should probably follow the main stream and get pure C filtering.
The question of film cap bypass is more complicated though. I don't have enough resources (or know how) to come up with a good design myself. DIY people sometimes use wound film caps for bypass (as used in speaker filters) but don't they have too much inductance (even though for instance Mundorf claim that their caps are wound in a special low inductance way)? I wonder if the often used series R with the film cap really is of benefit to lower the risk of resonance? Doesn't it just raise the impedance ? I have seen professional PSUs that use a small electrolytic bypass cap with relatively high esr, followed by film cap bypass without any additional series R. I am not a pro,so I am trying to understand this. Any comments, or advice ?
The question of film cap bypass is more complicated though. I don't have enough resources (or know how) to come up with a good design myself. DIY people sometimes use wound film caps for bypass (as used in speaker filters) but don't they have too much inductance (even though for instance Mundorf claim that their caps are wound in a special low inductance way)? I wonder if the often used series R with the film cap really is of benefit to lower the risk of resonance? Doesn't it just raise the impedance ? I have seen professional PSUs that use a small electrolytic bypass cap with relatively high esr, followed by film cap bypass without any additional series R. I am not a pro,so I am trying to understand this. Any comments, or advice ?
Yes, it is a little complicated. I don't know how familiar you are with resonance and the maths of resonant circuits so I'll start with a qualitative explanation. Apologies if I am going over stuff you already know.
Think of a mechanical system. Say a car suspension. There are three main ingredients to think about: the mass of the car, the springiness of the suspension and the damping of the suspension. I once drove a Citroen that I had borrowed from a friend and it's rear shocks weren't working. The back end bounced up and down in an alarming way and I had to keep adjusting my speed, especially on bumpy roads, to keep the thing under control. The problem was that there was mass and spring but not enough damping, so the car was prone to resonance at certain frequencies.
Any system that contains mass and spring will resonate at certain frequencies if there is insufficient damping. What is actually happening is that energy is being stored by the mass and spring and the energy exists in the continuous exchange between potential and kinetic energy contained in the velocity of the mass and the compression of the spring. With no damping, there will be a frequency, called the resonant frequency, where the mass and spring will absorb energy indefinitely and as they do so the size of the oscillations increases. This causes many problems, as you know, in things like bridges and can lead to destruction of the system. The idea of the damper is to get rid of excess energy by turning it into benign heat. In a mechanical system it is some sort of frictious thing, typically an air or oil piston. There are forumlas that relate the mass, spring and damper friction to the size of oscillations at the resonant frequency.
Electrical systems are analogous. If you consider voltage to be velocity then capacitance is mass and inductance is springiness and resistance is damping. If you connect a pure capacitance and a pure inductance in parallel then the two will absorb any energy that reaches them at their resonant frequency. There will be an oscillation where the energy is exchanged between the electrostatic charge in the C and the magnetic field in the L. If energy keeps being absorbed the voltage swing across the C and the current swing in the L will become huge and even dangerous. Many ham radio enthusiasts have received fatal shocks from transmission cables where impedance mismatches existed (a coax cable is a resonant circuit).
Every capacitor has all three elements. In modern electrolytics the resistance is pretty darned small...tenths of ohms or less, which means not much built-in damping. The inductance is primarily related to the physical size of the cap and the length of its leads and pcb traces. When you put a big electrolytic in parallel with a small film cap you have to take into account that the two caps are connected by the equivalent of a large inductor in series with a very small resistance. Usually, this creates an undesirable resonance at some frequency. To reduce the size of this resonance requires extra series resistance...ie: extra damping.
This is a trade-off: you are trading the output resistance of your psu filter against its tendency to resonate at a particular frequency. Sometimes, designers go for lowest resistance and try to reduce the inductance such that the resonant frequency becomes so high that it doesn't interfere and doesn't get excited. Others accept the extra resistance if they want a psu that works well across a wide frequency band. The choice completely depends upon what the nature is of the circuit the psu is supplying and what bandwidth demands it will place on the psu. Be aware that many "professionals" don't understand this stuff properly so don't take their circuits for granted.
As a simple rule of thumb, to avoid excessive oscillation at the resonant frequency the damping resistance should be no smaller than the reactance of the cap or inductor at the resonant frequency (they are the same at the resonant f).
This may help with the calculations:
Wikipedia on RLC circuits
Thank's for your elaborate reply traderbam. I spent two hours yesterday reading the thread regarging Carlos snubberized PSU, and related documents, and I feel that I am now "confused but at a higher level". It is not at all clear that PSU bypassing when it comes to power amp unregulated PSUs is really beneficial. I can see benefits of bypassing after a regulated PSU designed for a low current amp like an outputstage for a CD Player though. Perhaps selecting low inductance large reservoir caps like for instance EPCOS Sikorel and leave the bypassing is best after all. But then again, modding and experimenting is the prime reason for going the DIY route in the first place. Do I compromise the impedance at low frequencies by using a bypass cap combination with a series R for the high frequencies ?
"confused but at a higher level" 
That is some progress, I suppose! It is really hard to be sure of anything you read in this forum because there are so many different levels of expertise and different assumptions (or omissions) made. There is also a regular trap that people fall into when they think of one part of a circuit in isolation without considering the impact on the whole system.
The same applies here. I am not familiar with the Aussieamplifiers NXV200 design. It is therefore impossible for me to say what psu characteristics are optimum for it. If you email me the schematic I might be able to be more specific.
Let me generalize. Resonances are best avoided. So I would not bypass the electrolytics with small caps unless you have a quantifiable justification to do so that you understand yourself.
You might also get the impression from the posts here that the pursuit of zero psu output impedance is a high priority. That impression can lead to lower impedance at the expense of stability. Stability is really important. Impedance is easier to understand and measure than stability so it gets more attention.
All capacitors will look capacitive at low frequencies, then as the f increases they look resistive for a short spell and then look inductive. The usual rational for bypassing a big cap with a small cap is to try to reduce the inductance at high f (a small cap has lower L because it is physically smaller). But unless you know what you are doing you end up introducing resonances and you may not have any significant impact on the inductance your amp sees anyhow because the inductance will be dominated by the pcb tracks and power leads between the amp and the psu.
The ideal psu output looks like a resistance at all frequencies. This is impossible to implement. The question is what realistic psu impedance characteristic does the NXV200 need for optimal sonic performance. Out of curiosity you might email Mr. Holton and ask him; I'd be interested to see what sort of answer he offers.
If I were in your situation, I would not use bypass caps. I would use good quality electrolytics and I would try to minimize the lead and pcb trace inductances between the cap terminals and the amplifiers. You can always add the film caps afterwards and see if you can hear any difference. I'd be sure to have some series R to avoid any parasitic oscillation problems...no less than 0.5 ohms.
Brian
That is some progress, I suppose! It is really hard to be sure of anything you read in this forum because there are so many different levels of expertise and different assumptions (or omissions) made. There is also a regular trap that people fall into when they think of one part of a circuit in isolation without considering the impact on the whole system.
The same applies here. I am not familiar with the Aussieamplifiers NXV200 design. It is therefore impossible for me to say what psu characteristics are optimum for it. If you email me the schematic I might be able to be more specific.
Let me generalize. Resonances are best avoided. So I would not bypass the electrolytics with small caps unless you have a quantifiable justification to do so that you understand yourself.
You might also get the impression from the posts here that the pursuit of zero psu output impedance is a high priority. That impression can lead to lower impedance at the expense of stability. Stability is really important. Impedance is easier to understand and measure than stability so it gets more attention.
All capacitors will look capacitive at low frequencies, then as the f increases they look resistive for a short spell and then look inductive. The usual rational for bypassing a big cap with a small cap is to try to reduce the inductance at high f (a small cap has lower L because it is physically smaller). But unless you know what you are doing you end up introducing resonances and you may not have any significant impact on the inductance your amp sees anyhow because the inductance will be dominated by the pcb tracks and power leads between the amp and the psu.
The ideal psu output looks like a resistance at all frequencies. This is impossible to implement. The question is what realistic psu impedance characteristic does the NXV200 need for optimal sonic performance. Out of curiosity you might email Mr. Holton and ask him; I'd be interested to see what sort of answer he offers.
If I were in your situation, I would not use bypass caps. I would use good quality electrolytics and I would try to minimize the lead and pcb trace inductances between the cap terminals and the amplifiers. You can always add the film caps afterwards and see if you can hear any difference. I'd be sure to have some series R to avoid any parasitic oscillation problems...no less than 0.5 ohms.
Brian
MarcH said:
Here's a link from Marshall Leach's site that discusses some practical considerations for the choice of local bypass capacitors for the output stage collectors, based on some measurements he did, and the real-world experience of others. The capacitors he's referring to (C21, C24 etc.) are in the schematic of the amp, which is here.
I'll leave the bypass caps out, and try some with a series resisitor at a later stage. Cable and trace inductance issues should be first priority.
The choice of 100uF in combination with 100nF film cap might be a good choice (like in the Leach amp) if you take advantage of a fairly high esr in the electrolyitic 100uF (high voltage low inductance version). Anoher option would be to use for instance a 2,2uF film cap in combination with a 10-100nF cap, both through a 1R series resisitor.
I will try to get specific information about the NXV200 requirements for a PSU, but I am not that hopeful since optimizing a PSU does not seam to be what amp designers take most interest in..
The choice of 100uF in combination with 100nF film cap might be a good choice (like in the Leach amp) if you take advantage of a fairly high esr in the electrolyitic 100uF (high voltage low inductance version). Anoher option would be to use for instance a 2,2uF film cap in combination with a 10-100nF cap, both through a 1R series resisitor.
I will try to get specific information about the NXV200 requirements for a PSU, but I am not that hopeful since optimizing a PSU does not seam to be what amp designers take most interest in..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.