Amplifier Power Supply

[modulardesign]

Interesting models/explanations here. Might you consider when the switch to the voltage rail is closed as high impedance and when its partially or fully open low impedance

 

[modulardesign]

Do you want a white wine or balsamic vinaigrette with your salad?

Any of them will do a great job in keeping the salad fresh for longer and also add that extra zing

 

[modulardesign]

https://www.vishay.com/docs/70598/70598.pdf

 

[modulardesign]

Where automatic voltage detection may be preferred to manual voltage selection. Use a smaller transformer that can operate at 120 or 220 volts to get a regulated voltage to derive a voltage for the comparator and to drive relays. Use the regulators resistors and compare to a fixed voltage such as a Zener. Switch the amplifiers main power supply to 120 volts if AC voltage is less than 150 volts else switch to 220 volts using a suitable DPDT relay. You could consider adding additional circuitry to provide under voltage or over voltage protection.

Alternative DPDT wiring

 

[egellings]

Since voltage selection on a device tends to be a one-time set it and forget it function, I can't see where automating that is worth the effort. Flip the voltage selector switch once and be done with it. For a device that regularly travels internationally, then it makes sense to automate that.

 

[modulardesign]

Proper earthing/grounding/shielding of electronics plays two roles

1. Noise reduction and better signal integrity
2. Electrical safety

Starting with the analogy of the coaxial cable that requires a 50 Ohm or 75 Ohm terminator at one end then we can start to understand the problem.

For single rail electronics, the negative rail plays the role of earth. For dual rail electronics, the zero volt rail serves as earth.
If the potential difference between the power supply rails of an electronic circuit are within safe levels for touch and also the power is from a battery or an isolation transformer where the primary coil will never cross with the secondary coil even under fault, then such a circuit may use a two pin A/C connector and have a floating earth. In such devices the ground plane as well as the chassis are connected together.

Where the chassis is exposed to possible electrical connection to other devices either through input earth,output earth or the chassis itself to an equipment rack. With our initial coaxial cable analogy the problems encountered with 10base2 networks come into play. Only one end of a 10base2 network was terminated with a 50 Ohm resistor.

In our audio example we can manage our ground loops by increasing resistance through paths which we dont want currents to prefer. To better manage ground loops and other problems, we terminate at only one point that being between the chassis and the power supply ground terminal. Chassis consists of the enclosure, heatsinks,and the transformer iron core. The resitor used at this termination point can be anything between 100 to 10 Ohms depending on the currents at play. Where we want to create an AC short for certain high frequencies we bypass this resistor with a suitable capacitor between 10nF to 100nF.For electrical safety we also bypass this parallel resistor and capacitor with a pair of parallel diodes each oriented in the opposite direction. These are high current diodes capable of remaining intact while the AC fuse blows incase of a fault condition. These diodes also feature a high forward voltage for higher perceived resistance.

Two pin equipment does not protect the user from external electrical fault that comes into contact with the chassis. For three pin equipment the earth pin is connected directly to chassis. Any dangerous fault will blow the fuse disconnecting the dangerous voltage.

At the circuit level the electronics are mounted with isolating standoffs if using plated holes or holes that do not completely clear the circuit ground plane. Brass or non isolated standoffs maybe used where the mounting holes have no electrical connection with the PCB ground plane. For further noise and interference management the circuit should feature power supply high frequency decoupling capacitors especially across opamp IC power pins. These capacitors work hand in hand with input high frequency filter capacitors. High frequency noise management for PCBs having external inputs maybe further improved with termination at the circuit level between circuit ground and chassis using a capacitor atleast 10x smaller than our main terminator bypass capacitor and a bleeder resistor atleast 10x greater than our main terminator resistor.

 

[modulardesign]

In signal transmission between audio circuits, the transmitting end shields the signal with its earth, some manufacturers terminate high frequency noise at this end but it all depends on whether the transmitting circuit is using negative feedback or not. With negative feedback it maybe unnecessary. On the receiving end it matters whether or not the receiver has high currents on the circuit that are a magnitude higher than the transmitting equipment? If so the receiving circuit will try to isolate these currents with a small resistor value separating the grounds. It is also at the receiver where we terminate and filter high frequency noise.
There are other factors to consider for every circuit including the topology. For example the 1DIFFQC and the SYMEF are different puppies in some respects. However both have their passionate audiences, there are are extra bells and whistles happening for the SYMEF. Whether or not these contribute to the overall listening experience is for the listener to decide.

 

[Speedskater]

Starting with the analogy of the coaxial cable that requires a 50 Ohm or 75 Ohm terminator at one end then we can start to understand the problem.

That is about 'transmission line theory', it has nothing to do with power supply, grounding or Earthing.

With our initial coaxial cable analogy the problems encountered with 10base2 networks come into play. Only one end of a 10base2 network was terminated with a 50 Ohm resistor.

again, 'transmission line theory',

In our audio example we can manage our ground loops by increasing resistance through paths which we dont want currents to prefer.

It's better to reduce the resistance and area of a 'ground loop'.

 

[modulardesign]

Internal voltage select switch bracket

 

[modulardesign]

Alternative perspective on linear power supply design
Assumptions
1. Load is 2 Ohm
2. Transformer has a secondary of 28-0-28
3. Current requirement 28/2 =14Amps
4. Transformer can provide 20% more so we can skimp and choose 10Amps*56 ~=500VA-600VA (you will need tighter tolerances if powering a power regenerator)
5. Capacitors can provide a ripple current of 14Amps
6. Transformers secondary winding are less than or equal to load resistance/10 = 0.2 Ohms (otherwise efficiencies start being lower)
7. Capacitors have an ESR that is equal to load resistance/100 = 0.02 Ohms

 

[modulardesign]

Assuming you decided to upgrade your current good amplifier devices that the market offers that may meet this need

 

[restorer-john]

The worst troll. Ever.

 

[modulardesign]

Lets assume we charged the capacitor at 120Hz and at the same time drew a pure sine wave at 120Hz for our regenerator. Lets start with the assumption where the capacitor is fully charged and its capacity is ten times the cut off frequency capacitance for a 2 Ohm load at 6630uF.
The load will be powered by both the transformer and capacitor on lucky swings, although for a regenerator you can align these swings periodically we can use this extra capability to ignore the fact that the load is not really a resistor..
If the transformer is at a valley while were draining the capacitor with a peak voltage, the capacitor will start discharging. We aren't worried much as we chose the capacitors ESR and capacity such that its voltage drop wont be much before the transformer starts to kick in.
Since some of these capacitors are rated at +/- 20% we can multiply 6630*1.2 ~=8200uF
We still have a ripple of 0.1*40 = 4volts, we might want to lower this to 2volts and do a capacity of 8200uf * 2 for a total of 16,400/=
Should we add the capacity or should we start looking at diminishing returns ?

 

[modulardesign]

Switchmode offline/isolated flyback power supplies
Usually these have more stages but smaller sized components due to the higher switching frequency. A frequency of 200kHz results in smaller sized components
Topological overview
1. Rectifying 325Volts at a lower current means a smaller bridge rectifier, however the capacitor has to be rated at 400Volts to 600Volts. This stage alone is equivalent to our linear power supplies rectification stage. However a balance has to be struck between this capacitors storage capacity and its high frequency capability
2. Depending on type of SMPS a flyback isolation transformer for 200KHz operation is used, feedback maybe via a third secondary winding or via an optocoupler depending on the flyback chip used
3. The final rectification and storage stage creates a balance between the charging frequency , the storage capacitors high frequency behavior, feedback or no feedback affects design choices on capacitor value, this stage requires high speed rectifiers otherwise efficiencies will be affected

 

[modulardesign]

Following from our linear power supply example, at the first rectifier pulling 14 Amps across 120 Volts AC would require approximately 3900uF.We need to keep the traces around the isolation transformer (irrespective of how your switching the DC) and its drive circuitry very short. At the second rectifier a 100uF capacitor would suffice however we would end up using a 4700uF capacitor due to real world device current limitations. Switch mode power supplies and linear power supplies may have the same total costs of production and may even occupy the same volume area, however, the switch mode power supply starts winning the shipping weight cost. In terms of perceived audio differences when powering the same amplifier one would have to back that up with some measurements. For an SMPS if well designed from a theoretical perspective at the first rectifier the load seems a DC load to the rectifier. At the second rectifier, the power supply seems like A DC supply to the amplifier. However the SMPS seems to have a lot of moving parts that need to be evaluated for fitness of application as such one might shy from using it in high gain small signal sections or not. Would a class D amplifier sound better with a linear power supply or an SMPS? on the flip side would a class A amplifier sound better with an SMPS or a linear power supply ? We are currently spoilt for choices. Example SMPS caps

 

[modulardesign]

Vinyl and class D amplifiers have something in common. The class D output stage eyeballs the voltage, if we made the class D output stage eyeball the voltage based on the output of a tube amplifier input stage, things get pretty interesting quickly, coupled with a linear power supply one might be reminded of Vinyl

 

[modulardesign]

Following from above post here's the idea

 

[modulardesign]

Thank you all and happy new year. Hope the power supplies met your needs.
There are updates.
Following requests that some of you prefer PCBs

PCBs

Please find attached version 3
with dedicated chassis connection