For amplifier power supplies use IT = CV where I is current draw, T is time (i.e. 8.3 milliseconds for 60 Hz full wave) equals C for capacitance in farads and V for voltage drop from peak allowed. Solve for the variable you want.
Very good demonstration indeed. Thanks alot. I can only say that my humble experience confirms the importance of power supply caps quality and quantity for the sound. I increased the uF in a commercial amp and the result was impressive. I remember i put two big very high quality 63mF Mallory in the place of some cheap 10mF caps. The bass and the music details both came out more. As i said very impressive. Maybe it was the result of noise going down more and Voltage rails more stable i can only guess that. The problem usually is that high mF high quality caps are very very big (and expensive also). So there are practical issue. But the sound can change in an astonishing way. I remember the amp sound going on for almost half a minute after the shutdown for the energy reserve stored in the capacitors.
To maintain high power across the full frequency spectrum you will need much larger caps. You should test ripple with a full load on the power supply driven to clipping with clean low frequency source. It’s harder for the caps to fill in the ripple when the power supply is loaded.
Very good explanation, thank you. I've noticed too going from 4700uF to 10000uF on amplifiers only provides minimal gains. Going beyond that would be a waste
What you really want to do is add inductors to the filter network to get the best line filtering, especially under load. Use series L and shunt C. During load the field stored on the inductor provides additional charge without dropping the voltage. This has a few other benefits also, as reducing start up surge. You can repeat this pi network topology for even cleaner results.
Those inductors before the caps will be noisey as all hell until the caps charge up though. That can be dealt with but I would understand why people would avoid putting that in an audio amp. Take a look at this: www.falstad.com/circuit/circuitjs.html?cct=$+1+0.000005+9.001713130052181+41+5+43%0A169+-64+-96+112+-96+0+4+0.25+0+0+0+0.99%0Av+-112+-32+-112+-96+0+1+60+190+0+0+0.5%0Ag+112+-64+144+-64+0%0Ad+192+-32+192+-64+2+default%0Ad+192+-64+192+-96+2+default%0Aw+192+-64+176+-64+0%0Aw+176+-64+176+-96+0%0Aw+176+-96+112+-96+0%0Aw+192+-96+224+-96+0%0Ad+224+-64+224+-96+2+default%0Ad+224+-32+224+-64+2+default%0Aw+192+-32+224+-32+0%0Aw+224+-64+208+-64+0%0Aw+208+-64+208+-16+0%0Aw+208+-16+112+-16+0%0Aw+112+-16+112+-32+0%0A209+304+-96+304+-64+0+0.011+0.001+1%0A209+304+-64+304+-32+0+0.011+0.001+1%0Ag+304+-64+320+-64+0%0Aw+304+-32+400+-32+0%0Aw+304+-96+400+-96+0%0Ar+400+-96+400+-32+0+100%0Al+224+-96+304+-96+0+1+0%0Al+224+-32+304+-32+0+1+0%0Aw+-112+-96+-64+-96+0%0Aw+-112+-32+-64+-32+0%0Ao+20+64+0+4099+0.078125+0.0015625+0+2+20+3%0Ao+7+64+0+4099+40+0.2+1+2+7+3%0A
It is important that low ESR capacitors are selected, this will provide a much better curve. All this was shown by Otto H. Schade in an important paper in the 1930s.
This was a very cool presentation. While I sat through it I had many questions, most of which were addressed at the end of the video. I have one question though, are there any benefits to adding bigger capacitors after you reach the point of diminishing returns in terms of output? Like longevity for example?
For my tri-amplifier project, I am using a computer power supply to which I have modified the output transformer so that instead of giving me 12 Volts with 20 A, it gives me +30 / -30 V with 4 A. The switching frequency is 200 Khz so the ripple falls out of the audio range. Using 10000 µf capacitors I get very good results at 120 Watts power.
Hmm, PA amps I have used tend to have a very large electrolytic power cap in parallel with a high value resister between each voltage rail to reference ground. I always thought the cap was to supply peak power and the resister to allow the cap to drain after power down. After this, there is a sequence of increasingly smaller caps to smooth out various frequency ranges of power rail noise. The smallest filter caps can't handle the voltages so have a resister in series. to protect them. Smaller caps are primarily needed to smooth out the spikes caused by rectifier triggering voltages, these spikes have very steep slope so react like high frequency even though there are only 2 per supply cycle. Could be wrong but that was my take on it.
Just shows even your older videos are still informative. I sent an order off to China and it included a 10,000 uF cap. I have a single rail power supply I am building. It's 27 V at a transformer rated 8.88 A . I'm having a lot of V drop under load it's down to16V. Bad cap ? I'm ordering 2 more Panasonics (hopefully Panasonics) from a UK dealer. Thanks for the help.
27v (DC output?) at no load dropping to 16v (full load?) seems a bit much. Be sure a new cap can handle the ripple current. Three 3300uf caps in parallel may be a better idea for a supply handling this much current.
The results are somewhat counterintuitive. One may assume that 4 Ohm load would draw more current thus discharging power supply caps much faster. And as a result 4 Ohm load should require larger capacity for the same power output. Yet that doesn't seem to be the case. Could it be that a power supply for the 8 Ohm load might also have a higher power rails voltages than a supply for the 4 Ohm load.
Very nice little video . Given the small transformer and small chip amp the 2200 mfd looks like a good knee. with a larger transformer and big current pull . 6800 good start. When your hard core it get up to 100,000 mfd per rail and that's per channel at 200k mfd per rail per channel the power supply is as they say STIFF. You did a fine video with proper size for use cap . Bravo
@@SheikhN-bible-syndrome Two things one they are self forming above 100,000 mfd you will see voltage of 4 or more volts before you form them . Two turn on current is Huge NTC and a relay are a smart move . If a toroid is used the NTC is a must. 200,000 per leg and side 4 total will give a very stiff Class A power supply with a long time constant .
so does power supply effect sound quality when supply your low voltage too your opamps like on voltage regulator lm337 and lm317 too supply +/-15 volts
always enjoy the videos, thanks! If I wanted to run four op amps in parallel off one transformer without a center tap, would I build a power supply for each including their own rectifier, filter caps, and voltage regulator? Would this work well? Thanks
Hello john, i have a question: If for example you use two 4700uf in parallel instead of one 10000uf cap what it would be the result? Thanks for the good video
Does that mean a transformer having 2 or 3 Ampere rated capacity can't deliver more than 50℅ of the capacity and adding a higher value filter capacitor only does Minor improvements ?
This may be a dumb question but here goes, when you say "non inductive load" is that just a pure resistance ? if you had a wire wound resistor, does it have significant inductance over other types of resistor or does it mean something else please...cheers.
Yes, to the first question. Wire wound will have a small amount of inductance. There are non inductive wire wound resisters that have a winding going clockwise and a second winding wound CCW which cancels the inductance effect. Even a wire has some inductance, so let's say minimally inductive. Check my video on resistors. there's some interesting stuff in there.
Using Scope Probes with longer Ground Crocodile connector for testing noise in PSU is a BAD idea. One should pull the small Clamp in the front of the Probe and use the Barrel of the probe which mimics a Ball Point Pen refill .
Can i add capacitor to computer smps output which I use it to drive my car amplifier. Will it help .? Power supply is 12 volt dc and 15 amps current. Car 2 channel Amplifier works within 12 to 14 volt range and demands 15 amps on full performance.
@Dave Micolichek hi, i know capacitors will not increase amperes more than power supply. I would like to ask, will adding caps help maintain voltage and smoothen out load ( as music plays a beat some mili volt drops). When amplifier plays. It pulls a lot of power. I had doubt whether adding Some caps help cushioning load. Some car system owners add huge caps on battery to avoid dimming of headlights.
MrJohhhnnnyyy That depends on many things, but you should get around 65 watts of clean output into 8 ohms give or take. This considers the supply can maintain the 35 volt rails under load.
JohnAudioTech PS is not really capable of handling well over 100W, but will handle up to 100W quite nicely. So, considering efficiency (let's say 60%) of an amp, I ain't gonna get more than 30W into each channel. But it works fine for me, I don't listen to music at full power
This is really only true for the very small scale and low wattage amplifier used in the example, and starting with very little uf value. This whole scene will change to very little to NO, increased output, when you start at 10000uf and go up to 100000uf in capacitence. This fact starts to show already in your example where you go from 4700 to 10000 uf. The real reality for 100 watt to 2000 watt amplifiers, is that there is no change in output from extra capacitance in the smoothing circuit.
Hi, I built 48pcs X Elna 4700uF and then I connect it again to my Capacitance multiplier from Rod Elliot design with 2pcs X Epsilon 22.000uF (sound-au.com/project15.htm) for my 16watt amplifier (Kolibri Amplifier anistardi.wordpress.com/2014/12/15/kolibri-amplifier/) when I switch off, it needs around 70 seconds until it's completely turned off.
For amplifier power supplies use IT = CV where I is current draw, T is time (i.e. 8.3 milliseconds for 60 Hz full wave) equals C for capacitance in farads and V for voltage drop from peak allowed. Solve for the variable you want.
Very good demonstration indeed. Thanks alot. I can only say that my humble experience confirms the importance of power supply caps quality and quantity for the sound. I increased the uF in a commercial amp and the result was impressive.
I remember i put two big very high quality 63mF Mallory in the place of some cheap 10mF caps. The bass and the music details both came out more. As i said very impressive. Maybe it was the result of noise going down more and Voltage rails more stable i can only guess that.
The problem usually is that high mF high quality caps are very very big (and expensive also). So there are practical issue. But the sound can change in an astonishing way.
I remember the amp sound going on for almost half a minute after the shutdown for the energy reserve stored in the capacitors.
Exactly the info i was looking for. Thank you so much.
To maintain high power across the full frequency spectrum you will need much larger caps. You should test ripple with a full load on the power supply driven to clipping with clean low frequency source. It’s harder for the caps to fill in the ripple when the power supply is loaded.
Very good explanation, thank you. I've noticed too going from 4700uF to 10000uF on amplifiers only provides minimal gains. Going beyond that would be a waste
Thanks for the video !
I was tinkering with a board lately and ran into that problem. I setteld for a 2200µF cap in the end.
What you really want to do is add inductors to the filter network to get the best line filtering, especially under load. Use series L and shunt C. During load the field stored on the inductor provides additional charge without dropping the voltage. This has a few other benefits also, as reducing start up surge. You can repeat this pi network topology for even cleaner results.
Those inductors before the caps will be noisey as all hell until the caps charge up though. That can be dealt with but I would understand why people would avoid putting that in an audio amp.
Take a look at this: www.falstad.com/circuit/circuitjs.html?cct=$+1+0.000005+9.001713130052181+41+5+43%0A169+-64+-96+112+-96+0+4+0.25+0+0+0+0.99%0Av+-112+-32+-112+-96+0+1+60+190+0+0+0.5%0Ag+112+-64+144+-64+0%0Ad+192+-32+192+-64+2+default%0Ad+192+-64+192+-96+2+default%0Aw+192+-64+176+-64+0%0Aw+176+-64+176+-96+0%0Aw+176+-96+112+-96+0%0Aw+192+-96+224+-96+0%0Ad+224+-64+224+-96+2+default%0Ad+224+-32+224+-64+2+default%0Aw+192+-32+224+-32+0%0Aw+224+-64+208+-64+0%0Aw+208+-64+208+-16+0%0Aw+208+-16+112+-16+0%0Aw+112+-16+112+-32+0%0A209+304+-96+304+-64+0+0.011+0.001+1%0A209+304+-64+304+-32+0+0.011+0.001+1%0Ag+304+-64+320+-64+0%0Aw+304+-32+400+-32+0%0Aw+304+-96+400+-96+0%0Ar+400+-96+400+-32+0+100%0Al+224+-96+304+-96+0+1+0%0Al+224+-32+304+-32+0+1+0%0Aw+-112+-96+-64+-96+0%0Aw+-112+-32+-64+-32+0%0Ao+20+64+0+4099+0.078125+0.0015625+0+2+20+3%0Ao+7+64+0+4099+40+0.2+1+2+7+3%0A
It is important that low ESR capacitors are selected, this will provide a much better curve.
All this was shown by Otto H. Schade in an important paper in the 1930s.
This was a very cool presentation. While I sat through it I had many questions, most of which were addressed at the end of the video. I have one question though, are there any benefits to adding bigger capacitors after you reach the point of diminishing returns in terms of output? Like longevity for example?
Would be curious how a capacitance multiplier followed by small caps would work out.
For my tri-amplifier project, I am using a computer power supply to which I have modified the output transformer so that instead of giving me 12 Volts with 20 A, it gives me +30 / -30 V with 4 A.
The switching frequency is 200 Khz so the ripple falls out of the audio range. Using 10000 µf capacitors I get very good results at 120 Watts power.
Hmm, PA amps I have used tend to have a very large electrolytic power cap in parallel with a high value resister between each voltage rail to reference ground. I always thought the cap was to supply peak power and the resister to allow the cap to drain after power down. After this, there is a sequence of increasingly smaller caps to smooth out various frequency ranges of power rail noise. The smallest filter caps can't handle the voltages so have a resister in series. to protect them. Smaller caps are primarily needed to smooth out the spikes caused by rectifier triggering voltages, these spikes have very steep slope so react like high frequency even though there are only 2 per supply cycle. Could be wrong but that was my take on it.
Just shows even your older videos are still informative. I sent an order off to China and it included a 10,000 uF cap. I have a single rail power supply I am building. It's 27 V at a transformer rated 8.88 A . I'm having a lot of V drop under load it's down to16V. Bad cap ? I'm ordering 2 more Panasonics (hopefully Panasonics) from a UK dealer. Thanks for the help.
27v (DC output?) at no load dropping to 16v (full load?) seems a bit much. Be sure a new cap can handle the ripple current. Three 3300uf caps in parallel may be a better idea for a supply handling this much current.
Very simple and informative video, thank you for the upload.
brilliant vid....so if I replace a pair of 3300 uF caps in my Amp with 10,000 uF caps will the rest of the circuit work fine ?
Yes
what about dynamic power ?
The results are somewhat counterintuitive. One may assume that 4 Ohm load would draw more current thus discharging power supply caps much faster. And as a result 4 Ohm load should require larger capacity for the same power output. Yet that doesn't seem to be the case.
Could it be that a power supply for the 8 Ohm load might also have a higher power rails voltages than a supply for the 4 Ohm load.
i would love to know if a more normal power such as 100w responds similar. based on your testing 4700uf is basically the best size/cost/power ratio
Thank you John, Very Interesting ....
Very nice little video . Given the small transformer and small chip amp the 2200 mfd looks like a good knee. with a larger transformer and big current pull . 6800 good start. When your hard core it get up to 100,000 mfd per rail and that's per channel at 200k mfd per rail per channel the power supply is as they say STIFF. You did a fine video with proper size for use cap . Bravo
Ive got 8 nichicon gold tune 100v 100000uf capacitors there 4.5lb each i plan on using them on a mega class a amp power suply one day
@@SheikhN-bible-syndrome Two things one they are self forming above 100,000 mfd you will see voltage of 4 or more volts before you form them . Two turn on current is Huge NTC and a relay are a smart move . If a toroid is used the NTC is a must. 200,000 per leg and side 4 total will give a very stiff Class A power supply with a long time constant .
Do power chips like the tda2030a have common mode rejection of noise like in tube amplifiers?
so does power supply effect sound quality when supply your low voltage too your opamps like on voltage regulator lm337 and lm317 too supply +/-15 volts
Thank you so much for sharing this!
Great video!
always enjoy the videos, thanks! If I wanted to run four op amps in parallel off one transformer without a center tap, would I build a power supply for each including their own rectifier, filter caps, and voltage regulator? Would this work well? Thanks
общий принцип понятен. полезное видео
thank you!
Hi John,
Great demonstration, that's helpful to me.
Thanks for sharing.
ATB Adam
6:00 Nice frequency modulator you've built there! 😅
Hello john, i have a question:
If for example you use two 4700uf in parallel instead of one 10000uf cap what it would be the result?
Thanks for the good video
That would be pretty much 9400uF
@@IvanStepaniuk yes, but the ESR would be halved.
@Dave Micolichek ... and often cheaper.
pls test 10khz vs 100hz tones... too and sum them too ....also. try tantalum,, mica,,, hi speed caps on electrolytics too... thicker wires too
Does that mean a transformer having 2 or 3 Ampere rated capacity can't deliver more than 50℅ of the capacity and adding a higher value filter capacitor only does Minor improvements ?
what chip does that amplifier have? i want to build one of these
This may be a dumb question but here goes, when you say "non inductive load" is that just a pure resistance ? if you had a wire wound resistor, does it have significant inductance over other types of resistor or does it mean something else please...cheers.
Yes, to the first question. Wire wound will have a small amount of inductance. There are non inductive wire wound resisters that have a winding going clockwise and a second winding wound CCW which cancels the inductance effect. Even a wire has some inductance, so let's say minimally inductive. Check my video on resistors. there's some interesting stuff in there.
Using Scope Probes with longer Ground Crocodile connector for testing noise in PSU is a BAD idea. One should pull the small Clamp in the front of the Probe and use the Barrel of the probe which mimics a Ball Point Pen refill .
wont the breadboard introduce Major distortion and burning ? that thing cant handle more than 0.1 - 0.2 A , no ??
nice work! thanks
Can i add capacitor to computer smps output which I use it to drive my car amplifier. Will it help .?
Power supply is 12 volt dc and 15 amps current.
Car 2 channel Amplifier works within 12 to 14 volt range and demands 15 amps on full performance.
@Dave Micolichek hi, i know capacitors will not increase amperes more than power supply.
I would like to ask, will adding caps help maintain voltage and smoothen out load ( as music plays a beat some mili volt drops). When amplifier plays. It pulls a lot of power.
I had doubt whether adding Some caps help cushioning load.
Some car system owners add huge caps on battery to avoid dimming of headlights.
How many watts I can get on output with 2x4700uF caps in each rail? Rails are +-35V
MrJohhhnnnyyy That depends on many things, but you should get around 65 watts of clean output into 8 ohms give or take. This considers the supply can maintain the 35 volt rails under load.
JohnAudioTech PS is not really capable of handling well over 100W, but will handle up to 100W quite nicely. So, considering efficiency (let's say 60%) of an amp, I ain't gonna get more than 30W into each channel. But it works fine for me, I don't listen to music at full power
very interesting, thanks for sharing this :-)
This is really only true for the very small scale and low wattage amplifier used in the example, and starting with very little uf value. This whole scene will change to very little to NO, increased output, when you start at 10000uf and go up to 100000uf in capacitence. This fact starts to show already in your example where you go from 4700 to 10000 uf. The real reality for 100 watt to 2000 watt amplifiers, is that there is no change in output from extra capacitance in the smoothing circuit.
after 2200 uf,,, bass magic,,, bass speed,,, hear it...
Hi, I built 48pcs X Elna 4700uF and then I connect it again to my Capacitance multiplier from Rod Elliot design with 2pcs X Epsilon 22.000uF (sound-au.com/project15.htm) for my 16watt amplifier (Kolibri Amplifier anistardi.wordpress.com/2014/12/15/kolibri-amplifier/) when I switch off, it needs around 70 seconds until it's completely turned off.
small caps and no low pass filter setup. not best filter setup.