Hey Anyone,
I recently learned from another thread that the power supply is actually part of the audio circuit. I thought about and realized that while the DC current is grounded at the supply the AC audio signal is not. We ground the tubes and the output transformer to the power supply so there is a path for AC through the power supply to the transformer. As I understand it then the power supply is actually in parallel with the audio circuit.
This leads me to a couple questions.
First, is it desirable to isolate the power supply from the audio output? And how could you do that? A pair of parallel reversed diodes between the transformer and ground or or a cap to earth ground?
If it is neither desirable nor possible do we then need to use filter caps that have better audio properties?
Thanks, hoping this isn't a stupid question?
I recently learned from another thread that the power supply is actually part of the audio circuit. I thought about and realized that while the DC current is grounded at the supply the AC audio signal is not. We ground the tubes and the output transformer to the power supply so there is a path for AC through the power supply to the transformer. As I understand it then the power supply is actually in parallel with the audio circuit.
This leads me to a couple questions.
First, is it desirable to isolate the power supply from the audio output? And how could you do that? A pair of parallel reversed diodes between the transformer and ground or or a cap to earth ground?
If it is neither desirable nor possible do we then need to use filter caps that have better audio properties?
Thanks, hoping this isn't a stupid question?
You can isolate the power supply from the amp, but this generally requires a high impedance CCS and shunt regulator. For single ended amplifiers, parallel feed output stages offer some isolation between the power supply and signal handling components.
The better option is to use caps that you wouldn't mind listening to.
The better option is to use caps that you wouldn't mind listening to.
Not exactly in parallel, actually the usual sort of passive power supply is in series with the signal current.
Trace the path of audio load current in the output stage of a power amplifier and see.
The load current must complete a loop. In most cases the load current must flow through a large capacitor.
Or two capacitors, in the case of a push-pull circuit.
Last edited:
I think I've interpreted the query right - the power supply output filter capacitor(s) is normally the interface with the audio amplifier output stage circuitry (B+ or supply rails and 0V) - the 'ground' connection is arbitrary .
That interface capacitor handles the audio output signal current (as seen at the primary side of a valve amps output transformer, or the speaker/output stage ss). That capacitor also handles the charging current from the power supply.
Typically that interface capacitor has the lowest impedance path for all circulating currents, but that isn't to say that some audio side currents don't circulate further in to the power supply circuitry (depending on the filtering arrangement), and that some charging power supply ripple current doesn't circulate further in to the amplifier B+/supply rail distribution (depending on that circuit arrangement).
The larger the interface capacitance, the smaller the voltage variations due to charging and audio currents. Yes the 'capacitance' of an electrolytic cap will fall with signal frequency, but the reduction is typically minor for frequencies out to above the audio band (where circulating currents are normally reducing due to feedback or speaker transducer). The 'impedance' of an electrolytic cap will normally remain low and constant out to 100's of kHz, especially for a radial. So in general, there is only aesthetic 'under the bonnet' appeal to replacing or bypassing an electrolytic cap for that high value interface function.
That interface capacitor handles the audio output signal current (as seen at the primary side of a valve amps output transformer, or the speaker/output stage ss). That capacitor also handles the charging current from the power supply.
Typically that interface capacitor has the lowest impedance path for all circulating currents, but that isn't to say that some audio side currents don't circulate further in to the power supply circuitry (depending on the filtering arrangement), and that some charging power supply ripple current doesn't circulate further in to the amplifier B+/supply rail distribution (depending on that circuit arrangement).
The larger the interface capacitance, the smaller the voltage variations due to charging and audio currents. Yes the 'capacitance' of an electrolytic cap will fall with signal frequency, but the reduction is typically minor for frequencies out to above the audio band (where circulating currents are normally reducing due to feedback or speaker transducer). The 'impedance' of an electrolytic cap will normally remain low and constant out to 100's of kHz, especially for a radial. So in general, there is only aesthetic 'under the bonnet' appeal to replacing or bypassing an electrolytic cap for that high value interface function.
Last edited:
The impedance of the supply vs frequency, where it should be acting as a voltage source, there can be resonance in the supply itself or the components.
There can be noise earlier in the supply which could intermodulate.
As the source, it is in series with the amp (there is a current loop). The current will at least pass through the final capacitor(s). Your question is usually asked either in regard to whether signal current makes it back further in the supply, or maybe regarding the character of the final cap.
There can be noise earlier in the supply which could intermodulate.
As the source, it is in series with the amp (there is a current loop). The current will at least pass through the final capacitor(s). Your question is usually asked either in regard to whether signal current makes it back further in the supply, or maybe regarding the character of the final cap.
That may or may not be true. It is true to varying extents for different circuits. It is true for a SET, but almost untrue for push-pull. Don't be tempted to gain understanding via simplistic recipes.mr2racer said:
Meaningless question. As the PSU supplies power to the output it cannot be isolated.
No. Diodes which are reverse biased might as well be removed from the circuit, as they do nothing. Diodes which are forward biased might as well be replaced by a short circuit. Some people have the crazy idea that a diode can somehow pass DC and block AC; not true.
No, we need to design our circuits - including the power supply.
Yes. Note that frequencies include not just the audio but also the audio envelope. At the top end this overlaps with the LF end of the audio band; at the bottom end it goes down to maybe 0.1Hz (strictly speaking it goes lower than this - a symphony lasting 30 mins would have envelope frequencies down to 0.55mHz but we can consider that to be almost DC). People often forget about the envelope, and so design a PSU resonance right in the middle of the envelope region.TheGimp said:
I use JJ PSU caps and Triad chokes. I found a calculation that gives the internal resistance of the power supply based on no load and loaded voltages. I found .766 ohms with a GZ34 and .679 ohms for a full wave bridge. I don't at this time have a way to calculate the actual output impedance over frequencies. Can anyone guess at the PSU output impedance from this? I'm assuming that internal PSU resistance is related to output impedance.
Currently, I'm designing a push pull amp that is differential. It uses an interstage transformer instead of coupling caps. Fixed bias on both the voltage amps and the outputs. I intended the design to limit the effect of capacitors on the audio signal.
Currently, I'm designing a push pull amp that is differential. It uses an interstage transformer instead of coupling caps. Fixed bias on both the voltage amps and the outputs. I intended the design to limit the effect of capacitors on the audio signal.
> 563.37 - 447.26/.1516 = 765.89 ohms
Write "(563-447)/0.15" to show the subtraction has to be done first, and we are not really getting a 5-place exact answer. (And I hate naked leading decimal points because the BluePrint machine smudges them.)
This is the Steady DC impedance. What you get if you connect different loads and wait for the meter to settle.
We do not hear DC. It's boring.
> the power supply is actually part of the audio circuit
As "part of the audio circuit", *most* simple supplies act-like their capacitor as far as the "audio path" needs to know. Say 40uFd? Then 210 Ohms at 20Hz, 21r at 200Hz, 2r at 2KHz, and probably 2r all the way up due to internal resistance. Pretty much "under 100 Ohms" for all except the bottom octave where speakers do not go (affordably).
445V 0.15A suggests the amplifier acts-like about a 3K resistor. It might be one big tube with a 3K-5K load, or two tubes with 6K load each, or one push-pull stage with 6Kpp load. "Under 100r" is "in the signal path" but very small compared to total circuit impedance. And some of these cases, these days, we might be leaning from 40uFd old-school to 400uFd big-cans. Now we have less than 10 Ohms FPP. A very small and fairly linear "step" in the path.
Amplifiers work by taking a small precious signal and working it onto a large boring electric source. A sculptor works a mental image onto a chunk of marble. The raw source IS the sculpture, minus the bits chipped off. The amplifier output IS the DC power supply, minus the parts shaved-away to get bi-directional varying size music same-shape bigger-size as the precious signal in the groove.
Write "(563-447)/0.15" to show the subtraction has to be done first, and we are not really getting a 5-place exact answer. (And I hate naked leading decimal points because the BluePrint machine smudges them.)
This is the Steady DC impedance. What you get if you connect different loads and wait for the meter to settle.
We do not hear DC. It's boring.
> the power supply is actually part of the audio circuit
As "part of the audio circuit", *most* simple supplies act-like their capacitor as far as the "audio path" needs to know. Say 40uFd? Then 210 Ohms at 20Hz, 21r at 200Hz, 2r at 2KHz, and probably 2r all the way up due to internal resistance. Pretty much "under 100 Ohms" for all except the bottom octave where speakers do not go (affordably).
445V 0.15A suggests the amplifier acts-like about a 3K resistor. It might be one big tube with a 3K-5K load, or two tubes with 6K load each, or one push-pull stage with 6Kpp load. "Under 100r" is "in the signal path" but very small compared to total circuit impedance. And some of these cases, these days, we might be leaning from 40uFd old-school to 400uFd big-cans. Now we have less than 10 Ohms FPP. A very small and fairly linear "step" in the path.
Amplifiers work by taking a small precious signal and working it onto a large boring electric source. A sculptor works a mental image onto a chunk of marble. The raw source IS the sculpture, minus the bits chipped off. The amplifier output IS the DC power supply, minus the parts shaved-away to get bi-directional varying size music same-shape bigger-size as the precious signal in the groove.
Yes, you can have a constant current source that feeds the current that your circuit needs, plus some headroom to operate a shunt voltage regulator.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.