One caution on using a very large amount of capacitance for the main filter: The greater the capacitance there the higher the ratio of RMS current from the transformer to DC output current. This must be considered when specifying the transformer. Usually there isn't a huge difference between "enough" capacitance and an moderate excess, but it is something worth keeping in mind. What happens: Current can only flow from the transformer when the instantaneous voltage out of the rectifier is higher than the capacitor voltage. With most of the ripple "gone" that means current only flows very briefly during each half-cycle. However, the total amount of energy delivered must make up for the total amount of energy "removed" by the load each half cycle. You get "spikes" of current that are narrow but high in amplitude. That means higher RMS current. Transformers are rated for RMS current because the biggest part of their losses and consequent heating are in the resistance of the windings. It is RMS current that "counts" in the equation. It is quite easy to measure the RMS current with a good multimeter with a "true RMS" function. Measure between the transformer and rectifier. Keep in mind there will be a very large "surge current" when the cap first charges. Shorting the metere leads together until after the surge can protect the meter.
Thanks for this demonstration. Seeing the effect of different loads and different input capacitors has really helped me get a handle on how voltage regulators work.
Those 3 terminal regs can become unstable without the small bypass caps, especially if the connecting leads are long, as this starts to add inductance into the circuit. Even if you have a few thousand microfarads on the input, the bypass is still advisable, as the small cap is much more effective at higher frequencies, than the larger ones. The big electo's are good at removing mains ripple, not stopping high frequencies. The manufacturers recommend the two bypass caps be as close to the reg as practical.
The main reason for a small capacitor, with good high frequency characteristics (unlike a electrolytic), is to prevent high frequency noise from several sources. The voltage regulator itself could have high frequency rf (oscillations) on the output. In a high RF environment i.e. near transmitters - it prevents RF coming back down the power leads. It is a simple, cheap and effective cure to a lot of possible issues.
So true... the number of times i've found commercially produced PSUs with the common garden 3 term reg that breaks into oscillating at some point (& gets all hot) only to discover they saved 0.0001cent per unit by not adding the normal 0.1 caps.... !!!!
This video was very helpful to me because I love these types of regulators and build them regularly, the LM317 is one of my favorites. However I did experiment with the output capacitors but I don't have an oscilloscope so I can't do any of the testing that you show in the video, I can only measure things with a lets say middle class DMM so I just can't see the ripple the way an oscilloscope makes it visible, so I did what I thought was a good idea and put a much bigger capacitor on the output as they recommend as a minimum capacitor on the datasheet. So I need to thank you for showing this in your video, I'm now gonna change all my DIY power supplies and put the minimum recommended capacitor on the outputs and I'll keep doing that from now on, I'm sure that you helped me to prevent lots of problems in the future with my DIY power supplies. Best regards, Ricardo Penders
Excellent. Answered all sorts of questions about the heat created by the volt drop and a clue to the circuit I need. I like the way you don’t waffle and just get to the point. Nice balance between theory and practice.
linear regulator also has bandwidth regulation which means if you put transient load then regulator can't compensate for that fast transient . that transient going to draw current from capacitor at output , so some capacitor actuality needed
Just a side note: Some regulators dislike having low-ESR caps on their output legs. This will be noted in the manufacturer's data sheet, but those often don't ship with your order. READ THOSE SPECS! Also, if you're the type that "just has to see what happens" when you exceed the specs be sure you're wearing eye protection or a face shield. I picked bits of component parts out of my beard more than once in my youth. 😉
This was an amazing video showing me why I would or wouldn't want to use a larger microfard capicitor at the the beginning, the amount of voltage drop before you start seeing ripple, it also helped me to understand the input and out resistances to make a perfectly regulated power supply starting with about 9 volts dc so that I could derive a 5 volt output to be perfect for an A supply to light my filaments on antique radio, that has been a hobby of mine for maney years, I thank him fully and peace be with you always. Sincerely Eric Mootz
Great video, very well explained and demonstrated! I would just like to add that some viewers may notice somewhat large capacitors, 220uf - 470uf, on the regulated voltage traces on the device circuit board, especially if the power supply is on another PC board or separate from the device main board. This is due to a different situation. It is to compensate for the resistance and AC inductance of the lead wires from the power supply to the board, or even resistance losses in the main PC board power traces. The AC inductance deals with the transient response of the regulator as you noted, but now we are getting into some rather complex EE theory. Save that for another video.
I heard a quite different explanation: If you switch off the power, high capacitance of the circuit behind the regulator would act as a voltage supply that could damage the regulator. This is even mentioned in the LM78xx datasheets, with recommendation to use shunt diodes as protection.
That's completely different thing to what is discussed in this video. The discussion here is about low frequency input ripple, and that no output capacitor is required for that... the 78** still needs small capacitors for STABILITY. What the data sheets are talking about is that if you add large caps on the output to act as an energy reserve, then you must dealwith back feeding like you mentioned.
Be aware that 0.1uF is not always enough when dynamic loads are on the regulated line, the regulator can not adequately handle the peak demands of all functioning circuits. Certainly no very large capacitor should be on regulated end, or a diode MUST be across the output to input cathode on input side anode on output side ( 79xx reverse that), the 78xx can be destroyed by input falling below output , for example on turning input power off. However for good transient output function say 10uF or so is very desirable in any circuit with dynamic loads, not just resistive static loads.
Thank you from me too. I'm trying to learn best way to regulate cheap and nasty Aliexpress Daytime Running lights to, I'm thinking 12V. But now seeing what you've shown, I'm wondering what will happen when the DRL controller dims them (when headlights come on) what will happen to my DRLs. They draw about 129.9mA @ 12.323V for the pair. (1.600W) Each one has 6×LEDs which i have no idea if they're even being overdriven, or not, at only 12V, The DRL controller output Voltage, when the controller dimmed them was right down to 7.6V. While it's input was only pulling 68.2mA @ 12.332V giving us 0.841W. This is the point that I'm concerned that if i was using an LM7812, to stop it overdriving them when the batteries charging < or > 14.8V, then what will happen controller drops it down to 7.6 on to the regulator? How can i make this work and correct sized components. Caps diodes resistors etc. ps the DRL controller turns on the lights when the battery is charging (when V goes above 13V) and shuts the lights off 20sec after dropping under 13V
When I worked for a power supply manufacturer, the rule was roughly 1000µF per ampere. With acceptable ripple, of course. If there is too much ripple on the filter capacitor, it will heat up and age very quickly. Now, with switching power supplies that operate at 200 kHz, the problem is not the same !
With switching supplies...here I beg to differ.. the same issue is still there and has to be treated in the same way.. In general ( at all frequencies..in both , linear and switching ) the ripple current flow rate that charges/ discharges the cap. is the cause of capacitor heating even when caps. working voltage is well higher than the supply instantaneous peak voltage.... This is more critical in switching supplies.. because now the ESR & ESL of the cap become dominant .. therefore some times caps are derated.. The capacitor leads at h.f. can behave more like an inductor and to reduce this more than two caps. in parallel are required to achieve the targeted capacitor value..
For Audio (line amp, pre-amp etc) circuits the LM317/117 series "adjustable" regulators are bit better. I've built "audio-grade" PSU's with both. I wish this video was around 10+ years ago when I was building PSU units for my audio projects.
Thank you for the explanation, I have a question, how do you know what capacitance to use at the input and input? in other words, how do you select the capacitors? is it just by guessing different values of capacitors or through calculations? I'd really appreciate a feedback, I am doing a project on Undervoltage/Overvoltage protection system and I'm using a 7812 voltage regulator, I'm going to use capacitors in the circuit. Thank you
That is why we have such a thing as a SPSU=Switching PSU which uses a very high switching frequency which means we can use smaller coils and smaller capacitors and less heat is generated.
Always very useful to go back to basics .... especially where capacitors are concerned! More of these please. Noise suppression is an extremely complex subject
Adding a large capacitor after a linear regulated power supply isn't always pointless. Some also have current limiting, which is great, unless you need to occasionally exceed that current limit for very short amounts of time, like kicking over a motor. A capacitor after the current limiting would allow a short pulse of high current when necessary.
Can you do a video on use of switching mode power supplies in analog audio equipment, since modern switching PSs are so much more efficient, not to mention lighter and cheaper, than old linear supplies.
Tony, can a busted voltage regulator create some 120hz humm? On my mixer i have a B+18 and B-18. And my +18 measure almost 19 volt (18.85 volt). The negative is good.
Very well explained video. I would just add that in the past I kept a range of fixed linear regulators (7805, 7905, 7812 etc.) and if I needed some intermediate voltage I would put a zener between the GND pin and the 0V rail. Since adjustable regulators so cheap now, I just buy 20 or so LM317s and LM350s, which are rated at 1.5 and 3 A respectively in the TO-220 package. One trick with these is to bypass the ADJ pin with a capacitor of about 10uF to reduce the ripple rejection even more. You also should run another wire from the ADJ ground directly to the negative output terminal, separate from wire running from the main filter capacitor to the negative terminal. BTW I have built a dual power supply using a mains transformer with twin 20V windings, both independently rectified and filtered. One goes to a DPS5005 switching buck converter and the other to an LM350 linear regulator. The linear regulator has noticeably less ripple and generates zero switching noise (and EMF noise).
Generally, that's not a very good idea, since the zener's characteristics (like temperature dependency) will be 100% deducted from the stabilization, i.e. the zener variation will be added to / subtracted from the output voltage. Often, it's better to drive the gnd lead from the tap of a voltage divider between output and gnd. Regulation characteristics are dependent on the current going through the gnd tap as a fraction of the current through the voltage divider. A larger current through the voltage divider (smaller resistors) reduces the influence of varying current flowing out of the device's gnd pin to gnd.
Let's start another argument: I think that identical windings should be rectified separately and then joining two of the outputs for the 0V, rather than joining the windings with the bridge rectifier. It's down to the phase of the windings...
@@jonathanpalmer155 If the (separate) windings are identical, then it is safe to connect them in parallel. The transformer I used was made so you could connect the two secondary windings in series to double the voltage or in parallel to double the current rating, Or just use them independently, (and isolated). If you are making a general purpose PS, it does make sense to have separate rectifiers since that gives you maximum flexibility if you later decide to change the circuit. But if you just want to increase the current rating it is cheaper to parallel the windings, since you only need one bridge rectifier.
I really like this presentation. Xraytonyb you are always so good with your videos! I must bring up the subject of power supply impedance at audio frequencies for audio circuits. Everything is nice and predictable with a resistive load on your power supply but things are different if you are supplying the voltage for audio amplifier circuits. Even at the basic single stage op amp circuit the op amp is amplitude modulating the power supply rails. So larger capacitors after the voltage regulators lower the power supply source impedance for the audio waveforms. Think of the transient of the crack of the snare drum and the op amp needs to reproduce it at it's output but if the power supply can't source that transient fast enough the sound suffers.
The small output caps discussed here are for stability of the Regulator. If you need capacity as a "power reserve" then that can be added onto the output as well. In other words, you add some large electrolytics AS WELL as the small 100nF - 330nF caps.
Agregando un capacitor de bajo valor a la salida de los reguladores previene autooscilaciones del propio regulador que significará un voltaje diferente al deseado por la alta frecuencia.
Just rec'd a Technics SA-EX510 from my Finace'e in NC. Was her Dad's..Mint not a scratch. As I mentioned in a earlier emmail I once owned the STA-2100 when forty-years younger-lol! Tony do you have any thought on the Tech SA-EX510? Thanks.
For others using these 78/79 3 pin regulators.... There's an optimum supply voltage with these, not enough and you run into the issues outlined in this video, too much and the reg is working too hard. Ideally you want a transformer with the same rated voltage, EG 15v AC RMS for a 7815, which when rectified gives you about 20v DC, so a 5v difference. Also the caps to stop oscillation and noise suppression should be as close to the pins as poss and are different for 79XX types, being bigger. From the datasheet application note - 0.33u and 0.1u for the 78XX, 2,2u and 1u for the 79XX. Anyhoo, grand video as always...DA.
If you added a larger reservoir capacitor on the input to the regulator, the input ripple would be far less and you wouldn't see the ripple on the output. You can also put a dropping resistor on the input to reduce the wattage dropped in the regulator. Not useable on every application but something to bear in mind if you want to use smaller heat sinks.
If you have too much wattage to dissipate, the dropping resistor will just report the problem as it also needs to dissipate heat. The most efficient way is to use a buck regulator before the linear stage. If the filtering between the two is properly done, there won't be any ripple or transient at the output and this will be advisable for audio or high precision equipment.That's very common in modern electronic devices to have an unregulated switching power supply then some local LDOs or more precision ICs for the most voltage sensitive chips or opamps.
2n3055 or amosfet in the negative rail as regulator is the best as far as low dropout is concerned, An opamp or a transistor circuit with tl431 ref can control output voltage
Great video and explanation. I would like to ask you for your opinion. Do you recommend 9V linear power supply (LM7809) as power supply for LCR meter DER DE-5000 identical to your example or SMPS power supply? I do not care about efficiency, only the stability of the measurements.
I don't know how much of a difference it would make. I use a basic cheap SMPS power plug on mine and I haven't had any problems at all. It seems to work the same as when it is on battery power.
@@xraytonyb I was worried about SMPS voltage ripple. But given the low current of the device (I have not checked it yet), I hope it should not be a problem.
Excellent explanation w/demonstration, Thanks again for (all) videos Tony! ..Always enjoyable to benchDown with coffee at hand and listen/watch. ..High-quality channel :) >Paypal-support (if able to sometime) = highly recommendable for XrayTony's electronics-videos.
Actually, a chop stick is a good tool for probing. I have several in my kit and they have been in there from the start. Much safer than a screwdriver, which is conductive. I see them in YT videos a good bit. That's where I got the idea. Great for pressing on things to see if something is loose. A cinch stick is also common, for this purpose. I also DIY'd a needle probe with one. Great to reach in tight places and get a measurement.
Do u have any in depth videos on switch mode power supplies? Or do u know of any other really good quality videos on switch mode power supplies that u would use or recommend?
Great video! I was watching EEVBlog and he was saying that these voltage regulators weren’t good at removing ripple. I guess he was making the mistakes you outlined
In fact they explained different phenomenons. The low frequency ripple from a linear AC/DC is effectively removed with a linear regulator and a barely sufficient input cap but the high freq ripple is not as they pass through its transistors driven in linear mode and the feedback of the regulation circuitry is not that fast. Even worse, if you look at datasheets you'll see that those linear chips produce some ripple by their own (not as much as switching regulators but still) and that's one of the reason there is an output capacitance on schematics (the other is to keep stability and avoid the regulator to oscillate). They both told that the capacitance is crucial and without that a linear regulator can't do miracles.
The simple answer is you have to use a cap that is sized to prevent higher frequency ripple. On the ramp wave shown, the leading edge is almost infinite frequency. So, a smaller cap is better at responding to the higher frequency. Big caps are sluggish in responding to high frequencies.
It's not the main reason. The simple answer is that the output resistance of the voltage regulator is very low, thus the cutoff frequency of the equivalent lowpass filter would exceed 50/60 Hz frequencies (and its harmonics), residuals of which one would hope to filter. So an additional electrolythic capacitor would not make a difference. As you said, the low capacitance/low inductance 0.1 uF capacitor blocks high frequency currents of the consumer, so the voltage regulator doesn't need to react to those frequencies.
Faster Rise time of cap. charge ing rate only stresses diode there's no change in ripple frequency , only the ( p- p) ripple voltage , can / may stress the regulator..whereas capacitor ripple current must be within its specified limits..
What you call "ramp wave" actually is part of the (50 or 60 Hz) input sine wave; the last part until the voltage reaches its peak value. Far from "almost infinite frequency".
lets say you need a 5V, 3A regulated dc voltage power supply, is it possible to build with 2 blocks of 7805 in parallel? any coupling problem between the 2 blocks?
The two regulators will fight, and (worst case) fast-oscillate at full power. You can boost up a regulator with a big external transistor, often called a "pass transistor"
The tolerances in manufacturing will mean one reg will put out a smidgen more volts, and as such take most of the load, so it will be overloaded and stressed, and the other just coasting along. You are better off adding a high power transistor to increase the output current. Unfortunately I cannot pop a suitable circuit into these comments, use Google, or check out the data sheets, sometimes they show you how in the application notes.
Both Pauls are essentially correct here. The best solution is a single regulator with a series pass transistor and don't forget to put the pass transistor on a nice big heat sink.
Use a buck converter, far better efficiency and higher current. Switching frequency and type of buck converter depends on application. I have not used linear regulators in years unless subregulated from a switch mode power supply (subregulation with Low Drop Out linear regulators are low noise and have pretty good power supply rejection ratio....good for radio transceivers).
Hi tONY.. i have a question that sorta fits with the video and your explaination... with vintage audio equipment that has a stated mains input of 110 volts, and no soft start circuit, wouls it be advisable to add in a 'PROTECTION ' circuit ... i realize that with the volume all the way down that the speakers are protected from the initial power up with out one, but with the difference in mains supply we have today versus the 110 v stated on the equipment i am thinking that the ac spikes would cause premature compoenent wear, anybodies input to this question would be welcome ..thanks....
RMS value is a DC equivalent of an AC signal. With normal mains voltage, which is a sine wave, the peak voltage will be √2 (or 1.414) times the RMS value. The capacitor, with no load on the circuit, will charge up to this peak value. So a 12 volt AC RMS signal will charge the capacitor up to just under 17 volts DC. Check out the video I have linked at the end of this one. I go over it on that video. Thanks for the comment!
Linear regulators are regulated, so their outputs are very stable. A zener diode is "open loop" and drifts with load. You can use a zener if you don't need precision, but linears are pretty cheap, so you might as well take the better regulation.
This video explains more of what a 78xx expects than cap value smoothing ripples. By the look of it, the 78xx does a very good job of smoothing ripples with a small physical package, as long as one gives it +1.5 working volt above output. 78xx linear while easy to use is wildly inefficient, whoever invented the switching power supply should get a Nobel.
1.5V drop out is way lower than the typical 2V of the 78xx series, 1.5V is the best case scenario at few mA.Otherwise I agree. LDO regulators are still great for small current applications but switching PSUs were a revolution for everything else.
garbage in garbage out hmmmm what happens if we run these regulators in parallel, say we need 2 or 3 branches that would be more amps than one alone could handle???
You could design a linear regulator using a mosfet (or a bank of 'em) on a humungous heat sink. Then there's the control circuit. But this whole arrangement is expensive and it wastes a lot of B+ voltage which could otherwise go toward producing power output. Quite honestly, you would be much further ahead just doing some conventional tube-era maintenance. I am the proud owner of a Bell Sound Carillon 6060 amp. As-is condition, it sounded good but not spectacular. I replaced the EL34 output tubes with "re-issues" of a well known British brand, the 5V3 rectifier with a "new-old-stock" unit I found at a reasonable price. Just renewing all the power tubes turned this amp into a real room-rattling beast! I also replaced the electrolytic filter caps with new units and that added even more "punch" to the low end. New "orange drop" coupling caps too. But the new tubes alone made the biggest difference. These things are all recommended whether you're into hi-fi or guitar. Hope you find this helpful. 8D
My intuition is that regulating the high voltage supply would diminish the qualities people value in tube amps. Voltage sag under load is often cultivated intentionally in the name of tone.
@@brocktechnology Tbh it's the same with solid state... high current egulated supplies need to be highly engineered and tested a lot. Subjectively my amps sound way nicer with linear supplies.
The term (on the diagram) "smoothing" is incorrect, it is a low pass filter. (passing DC, but not the ac ripple frequency. (Pi networks are king here) The regulating IC works hard to cancel out the ripple, it uses a lot of internal gain to do that, which makes three terminal regulators nosier than properly designed discrete regulators with moderated gain. Ron W4BIN
There's noting "incorrect" about the term "smoothing capacitor" in this application, as it is a standard term that is on much information available if you bothered to research or even do a web search. A true "Pi network" also uses other components such as inductors and resistors to create either a high, low, or bandpass filter circuit for AC signals. A single capacitor in parallel with a DC source with an AC ripple component could really not be considered a "network" onto itself, so, the term "smoothing capacitor" is a good description of a single components function in this particular circuit.
@@poormanselectronicsbench2021 Yup. I agree. The cap is there to store and deliver energy and "smoothing" is a good term for that. It is commonly used in the UK.
I've seen an old amp with a lot of big caps in it which seem to run in parallel with ceramic resistors on the output channel to the speaker (i assume its a single channel amp). whats the reason for so many caps and could i improve the speaker output with a similar set up on my new Yamaha home cinema amp. The reason for this is there's a distinct distortion on low notes even at low volume. I'm an electronics assembly orerative so opening the lid and modifying or adding circuits wouldn't be a problem. many thanks. jon.
One caution on using a very large amount of capacitance for the main filter:
The greater the capacitance there the higher the ratio of RMS current from the transformer to DC output current. This must be considered when specifying the transformer. Usually there isn't a huge difference between "enough" capacitance and an moderate excess, but it is something worth keeping in mind.
What happens:
Current can only flow from the transformer when the instantaneous voltage out of the rectifier is higher than the capacitor voltage. With most of the ripple "gone" that means current only flows very briefly during each half-cycle. However, the total amount of energy delivered must make up for the total amount of energy "removed" by the load each half cycle. You get "spikes" of current that are narrow but high in amplitude. That means higher RMS current. Transformers are rated for RMS current because the biggest part of their losses and consequent heating are in the resistance of the windings. It is RMS current that "counts" in the equation.
It is quite easy to measure the RMS current with a good multimeter with a "true RMS" function. Measure between the transformer and rectifier. Keep in mind there will be a very large "surge current" when the cap first charges. Shorting the metere leads together until after the surge can protect the meter.
Thanks for this demonstration. Seeing the effect of different loads and different input capacitors has really helped me get a handle on how voltage regulators work.
Those 3 terminal regs can become unstable without the small bypass caps, especially if the connecting leads are long, as this starts to add inductance into the circuit.
Even if you have a few thousand microfarads on the input, the bypass is still advisable, as the small cap is much more effective at higher frequencies, than the larger ones. The big electo's are good at removing mains ripple, not stopping high frequencies.
The manufacturers recommend the two bypass caps be as close to the reg as practical.
The main reason for a small capacitor, with good high frequency characteristics (unlike a electrolytic), is to prevent high frequency noise from several sources. The voltage regulator itself could have high frequency rf (oscillations) on the output. In a high RF environment i.e. near transmitters - it prevents RF coming back down the power leads. It is a simple, cheap and effective cure to a lot of possible issues.
So true... the number of times i've found commercially produced PSUs with the common garden 3 term reg that breaks into oscillating at some point (& gets all hot) only to discover they saved 0.0001cent per unit by not adding the normal 0.1 caps.... !!!!
Very interesting and well demonstrated.
This video was very helpful to me because I love these types of regulators and build them regularly, the LM317 is one of my favorites. However I did experiment with the output capacitors but I don't have an oscilloscope so I can't do any of the testing that you show in the video, I can only measure things with a lets say middle class DMM so I just can't see the ripple the way an oscilloscope makes it visible, so I did what I thought was a good idea and put a much bigger capacitor on the output as they recommend as a minimum capacitor on the datasheet.
So I need to thank you for showing this in your video, I'm now gonna change all my DIY power supplies and put the minimum recommended capacitor on the outputs and I'll keep doing that from now on, I'm sure that you helped me to prevent lots of problems in the future with my DIY power supplies.
Best regards,
Ricardo Penders
Very good explanation. Where were you when i was studying electronics 30 odd years ago. You make it so simple to understand. Thank you
Excellent. Answered all sorts of questions about the heat created by the volt drop and a clue to the circuit I need. I like the way you don’t waffle and just get to the point. Nice balance between theory and practice.
linear regulator also has bandwidth regulation which means if you put transient load then regulator can't compensate for that fast transient . that transient going to draw current from capacitor at output , so some capacitor actuality needed
Just a side note: Some regulators dislike having low-ESR caps on their output legs. This will be noted in the manufacturer's data sheet, but those often don't ship with your order. READ THOSE SPECS! Also, if you're the type that "just has to see what happens" when you exceed the specs be sure you're wearing eye protection or a face shield. I picked bits of component parts out of my beard more than once in my youth. 😉
Thanks, I'll keep that in mind next time that I'm building a new power supply.
Tantalums reverse biased every time please and stand well back!
This was an amazing video showing me why I would or wouldn't want to use a larger microfard capicitor at the the beginning, the amount of voltage drop before you start seeing ripple, it also helped me to understand the input and out resistances to make a perfectly regulated power supply starting with about 9 volts dc so that I could derive a 5 volt output to be perfect for an A supply to light my filaments on antique radio, that has been a hobby of mine for maney years, I thank him fully and peace be with you always. Sincerely Eric Mootz
Same to you, Sir!Your explanations are clear as and demonstrations couldn't be better.Thank you!
Great video, very well explained and demonstrated! I would just like to add that some viewers may notice somewhat large capacitors, 220uf - 470uf, on the regulated voltage traces on the device circuit board, especially if the power supply is on another PC board or separate from the device main board. This is due to a different situation. It is to compensate for the resistance and AC inductance of the lead wires from the power supply to the board, or even resistance losses in the main PC board power traces. The AC inductance deals with the transient response of the regulator as you noted, but now we are getting into some rather complex EE theory. Save that for another video.
I'm learning a lot from the comments today, thanks for adding the extra bit of information. To me this is very helpful.
Great video. Learned a lot on voltage regulation here.
I really appreciate your work... took me back to 1972
This the kind of information that is golden to noobies like me. Just earned yourself a subscriber!
I heard a quite different explanation: If you switch off the power, high capacitance of the circuit behind the regulator would act as a voltage supply that could damage the regulator. This is even mentioned in the LM78xx datasheets, with recommendation to use shunt diodes as protection.
That's completely different thing to what is discussed in this video. The discussion here is about low frequency input ripple, and that no output capacitor is required for that... the 78** still needs small capacitors for STABILITY.
What the data sheets are talking about is that if you add large caps on the output to act as an energy reserve, then you must dealwith back feeding like you mentioned.
NEAR 80 STILL LEARNING NICE VIDEO
and now you should learn, what it means writing in capital letters ;-)
Be aware that 0.1uF is not always enough when dynamic loads are on the regulated line, the regulator can not adequately handle the peak demands of all functioning circuits. Certainly no very large capacitor should be on regulated end, or a diode MUST be across the output to input cathode on input side anode on output side ( 79xx reverse that), the 78xx can be destroyed by input falling below output , for example on turning input power off. However for good transient output function say 10uF or so is very desirable in any circuit with dynamic loads, not just resistive static loads.
Thank you from me too. I'm trying to learn best way to regulate cheap and nasty Aliexpress Daytime Running lights to, I'm thinking 12V. But now seeing what you've shown, I'm wondering what will happen when the DRL controller dims them (when headlights come on) what will happen to my DRLs. They draw about 129.9mA @ 12.323V for the pair. (1.600W) Each one has 6×LEDs which i have no idea if they're even being overdriven, or not, at only 12V, The DRL controller output Voltage, when the controller dimmed them was right down to 7.6V.
While it's input was only pulling 68.2mA @ 12.332V giving us 0.841W.
This is the point that I'm concerned that if i was using an LM7812, to stop it overdriving them when the batteries charging < or > 14.8V, then what will happen controller drops it down to 7.6 on to the regulator?
How can i make this work and correct sized components. Caps diodes resistors etc.
ps the DRL controller turns on the lights when the battery is charging (when V goes above 13V) and shuts the lights off 20sec after dropping under 13V
Excellent demo. Thank you.
Good video. Thanks Tony. These 78xx and 79xx and 338/337/350/317... have short circuit protection too, and that's good to have.
When I worked for a power supply manufacturer, the rule was roughly 1000µF per ampere. With acceptable ripple, of course. If there is too much ripple on the filter capacitor, it will heat up and age very quickly.
Now, with switching power supplies that operate at 200 kHz, the problem is not the same !
@MichaelKingsfordGray Beefier rectification
With switching supplies...here I beg to differ.. the same issue is still there and has to be treated in the same way..
In general ( at all frequencies..in both , linear and switching ) the ripple current flow rate that charges/ discharges the cap. is the cause of capacitor heating even when caps. working voltage is well higher than the supply instantaneous peak voltage....
This is more critical in switching supplies.. because now the ESR & ESL of the cap become dominant .. therefore some times caps are derated..
The capacitor leads at h.f. can behave more like an inductor
and to reduce this more than two caps. in parallel are required to achieve the targeted capacitor value..
Brilliantly explained A compendium of some of the videos like these and who needs to go to college to learn electronics???
Electronics implies way serious maths than what's explained here. Remember that YT videos are rarely courses but more discoveries or entertainments.
For Audio (line amp, pre-amp etc) circuits the LM317/117 series "adjustable" regulators are bit better. I've built "audio-grade" PSU's with both. I wish this video was around 10+ years ago when I was building PSU units for my audio projects.
Awesome, your explanations are always so good!
Wow...I just viewed that same circuit diagram to experiment building my first rectifier. I only smoked one 1N4007 lol.
To be on the safe side ..REPLACE ALL with new ones..
Well done. Explained what I hadn't figured out on my own. Thank you sir!
Thank you for the explanation,
I have a question, how do you know what capacitance to use at the input and input? in other words, how do you select the capacitors? is it just by guessing different values of capacitors or through calculations?
I'd really appreciate a feedback, I am doing a project on Undervoltage/Overvoltage protection system and I'm using a 7812 voltage regulator, I'm going to use capacitors in the circuit.
Thank you
That is why we have such a thing as a SPSU=Switching PSU which uses a very high switching frequency which means we can use smaller coils and smaller capacitors and less heat is generated.
They can also be used for a inexpensive constant current regulator in a different configuration.
Always very useful to go back to basics .... especially where capacitors are concerned! More of these please. Noise suppression is an extremely complex subject
Regulator type 78xx for positive voltages.
Regulator type 79xx for negative voltages.
The last two digits indicates "how many" volts.
Thanks Tony. How about using a regulator for class A power amps? Also are you going to cover capacitor multiplier supplies?
Adding a large capacitor after a linear regulated power supply isn't always pointless. Some also have current limiting, which is great, unless you need to occasionally exceed that current limit for very short amounts of time, like kicking over a motor. A capacitor after the current limiting would allow a short pulse of high current when necessary.
This is a really really useful video! Everything's so clear. Thank you very much!
Can you do a video on use of switching mode power supplies in analog audio equipment, since modern switching PSs are so much more efficient, not to mention lighter and cheaper, than old linear supplies.
Excellent work Tony, thanks for your efforts 🙏
What an EXCELLENT explanation!!! THANKS A BUNCH!
good demonstration! clear
Nothing to add, great explanation Tony, stay safe...
This kind of schematic diagram information is interesting & educational for me to review/recall after 25 years ago.
Tony, can a busted voltage regulator create some 120hz humm? On my mixer i have a B+18 and B-18. And my +18 measure almost 19 volt (18.85 volt). The negative is good.
Very well explained video.
I would just add that in the past I kept a range of fixed linear regulators (7805, 7905, 7812 etc.) and if I needed some intermediate voltage I would put a zener between the GND pin and the 0V rail.
Since adjustable regulators so cheap now, I just buy 20 or so LM317s and LM350s, which are rated at 1.5 and 3 A respectively in the TO-220 package. One trick with these is to bypass the ADJ pin with a capacitor of about 10uF to reduce the ripple rejection even more. You also should run another wire from the ADJ ground directly to the negative output terminal, separate from wire running from the main filter capacitor to the negative terminal.
BTW I have built a dual power supply using a mains transformer with twin 20V windings, both independently rectified and filtered. One goes to a DPS5005 switching buck converter and the other to an LM350 linear regulator. The linear regulator has noticeably less ripple and generates zero switching noise (and EMF noise).
Generally, that's not a very good idea, since the zener's characteristics (like temperature dependency) will be 100% deducted from the stabilization, i.e. the zener variation will be added to / subtracted from the output voltage. Often, it's better to drive the gnd lead from the tap of a voltage divider between output and gnd. Regulation characteristics are dependent on the current going through the gnd tap as a fraction of the current through the voltage divider. A larger current through the voltage divider (smaller resistors) reduces the influence of varying current flowing out of the device's gnd pin to gnd.
Let's start another argument: I think that identical windings should be rectified separately and then joining two of the outputs for the 0V, rather than joining the windings with the bridge rectifier. It's down to the phase of the windings...
@@jonathanpalmer155 If the (separate) windings are identical, then it is safe to connect them in parallel.
The transformer I used was made so you could connect the two secondary windings in series to double the voltage or in parallel to double the current rating, Or just use them independently, (and isolated). If you are making a general purpose PS, it does make sense to have separate rectifiers since that gives you maximum flexibility if you later decide to change the circuit. But if you just want to increase the current rating it is cheaper to parallel the windings, since you only need one bridge rectifier.
That was very simple and intuitively explained.
I really like this presentation. Xraytonyb you are always so good with your videos! I must bring up the subject of power supply impedance at audio frequencies for audio circuits. Everything is nice and predictable with a resistive load on your power supply but things are different if you are supplying the voltage for audio amplifier circuits. Even at the basic single stage op amp circuit the op amp is amplitude modulating the power supply rails. So larger capacitors after the voltage regulators lower the power supply source impedance for the audio waveforms. Think of the transient of the crack of the snare drum and the op amp needs to reproduce it at it's output but if the power supply can't source that transient fast enough the sound suffers.
That's something many people don't understand - a very low bass note with a fast leading edge actually begins as a high frequency signal.
The small output caps discussed here are for stability of the Regulator. If you need capacity as a "power reserve" then that can be added onto the output as well.
In other words, you add some large electrolytics AS WELL as the small 100nF - 330nF caps.
Very informative, thanks!
Very interesting and well explained video. Thank you for sharing Tony!
Hey Tony I was told by one of your subscribers to contact you about possible repairs on a sansui receiver so giving it a shot :)
Then you need to CONTACT him, not waste time typing UA-cam comments.
Huge amount of good info and really well explained. Nice one.
odpoved je jednoducha: maly kondik typicky 10 - 100nF zabranuje kmitaniu stabilizatora na relativne vysokych frekvenciach 1 - 2MHz, na ktore su 78XX (79XX) nachylne :-)
Agregando un capacitor de bajo valor a la salida de los reguladores previene autooscilaciones del propio regulador que significará un voltaje diferente al deseado por la alta frecuencia.
Just rec'd a Technics SA-EX510 from my Finace'e in NC. Was her Dad's..Mint not a scratch. As I mentioned in a earlier emmail I once owned the STA-2100 when forty-years younger-lol! Tony do you have any thought on the Tech SA-EX510? Thanks.
I love it when I get those "Aha" moments. Great explanation, thank you!
Crystal clear explanation. Thank you 🙏
This was a great vidoe I learned alot here thanks
For others using these 78/79 3 pin regulators.... There's an optimum supply voltage with these, not enough and you run into the issues outlined in this video, too much and the reg is working too hard. Ideally you want a transformer with the same rated voltage, EG 15v AC RMS for a 7815, which when rectified gives you about 20v DC, so a 5v difference. Also the caps to stop oscillation and noise suppression should be as close to the pins as poss and are different for 79XX types, being bigger. From the datasheet application note - 0.33u and 0.1u for the 78XX, 2,2u and 1u for the 79XX.
Anyhoo, grand video as always...DA.
Thank you for the information!
Thanks for the extra bit of information
If you added a larger reservoir capacitor on the input to the regulator, the input ripple would be far less and you wouldn't see the ripple on the output. You can also put a dropping resistor on the input to reduce the wattage dropped in the regulator. Not useable on every application but something to bear in mind if you want to use smaller heat sinks.
If you have too much wattage to dissipate, the dropping resistor will just report the problem as it also needs to dissipate heat. The most efficient way is to use a buck regulator before the linear stage. If the filtering between the two is properly done, there won't be any ripple or transient at the output and this will be advisable for audio or high precision equipment.That's very common in modern electronic devices to have an unregulated switching power supply then some local LDOs or more precision ICs for the most voltage sensitive chips or opamps.
So for an audio preamp section would it be better to use the old zener and darlington bjt configuration for slew rate reasons?
Thank you very much Tony!
Super cute tips sir very nice work
2n3055 or amosfet in the negative rail as regulator is the best as far as low dropout is concerned, An opamp or a transistor circuit with tl431 ref can control output voltage
I like your videos a lot. I was curious what country you are based in.
Looks like I learned something here😁👍🏻
So well exp[lained. Thank you.
Very well explained. Thank you!!
Great video and explanation. I would like to ask you for your opinion. Do you recommend 9V linear power supply (LM7809) as power supply for LCR meter DER DE-5000 identical to your example or SMPS power supply? I do not care about efficiency, only the stability of the measurements.
I don't know how much of a difference it would make. I use a basic cheap SMPS power plug on mine and I haven't had any problems at all. It seems to work the same as when it is on battery power.
@@xraytonyb
I was worried about SMPS voltage ripple. But given the low current of the device (I have not checked it yet), I hope it should not be a problem.
Ideally (for robustness), a diode should be placed across the linear regulators out / in. See spec sheets for details.
@Karen See www.ti.com/lit/ds/symlink/lm340.pdf --- Section 8.1.4. This (also) solves issue when the voltage at VOUT is greater than at VIN.
Nicely done.
The capacitors are designed to store all the smoke that is needed makes everything run and when you let the smoke out everything quits working.
Great explanation.
Excellent explanation w/demonstration, Thanks again for (all) videos Tony! ..Always enjoyable to benchDown with coffee at hand and listen/watch. ..High-quality channel :)
>Paypal-support (if able to sometime) = highly recommendable for XrayTony's electronics-videos.
Excellent! Thank you.
Great explanation thanks. Where would we be without oscilloscopes? Though I must admin I never had a chop stick in my tool kit !
Actually, a chop stick is a good tool for probing. I have several in my kit and they have been in there from the start. Much safer than a screwdriver, which is conductive. I see them in YT videos a good bit. That's where I got the idea. Great for pressing on things to see if something is loose. A cinch stick is also common, for this purpose. I also DIY'd a needle probe with one. Great to reach in tight places and get a measurement.
You need to order Chinese to go more often
Very nice explanation
Another interesting video. Thanks a lot
Do u have any in depth videos on switch mode power supplies? Or do u know of any other really good quality videos on switch mode power supplies that u would use or recommend?
check out eevblog and Mr Carlson's lab. the latter I believe repairs them for cnc and other machinery. eevblog goes on depth on several videos.
Brilliant, glad I watched, new subscriber. :) (was in recommended finally)
Great video! I was watching EEVBlog and he was saying that these voltage regulators weren’t good at removing ripple. I guess he was making the mistakes you outlined
In fact they explained different phenomenons. The low frequency ripple from a linear AC/DC is effectively removed with a linear regulator and a barely sufficient input cap but the high freq ripple is not as they pass through its transistors driven in linear mode and the feedback of the regulation circuitry is not that fast.
Even worse, if you look at datasheets you'll see that those linear chips produce some ripple by their own (not as much as switching regulators but still) and that's one of the reason there is an output capacitance on schematics (the other is to keep stability and avoid the regulator to oscillate).
They both told that the capacitance is crucial and without that a linear regulator can't do miracles.
The simple answer is you have to use a cap that is sized to prevent higher frequency ripple. On the ramp wave shown, the leading edge is almost infinite frequency. So, a smaller cap is better at responding to the higher frequency. Big caps are sluggish in responding to high frequencies.
It's not the main reason. The simple answer is that the output resistance of the voltage regulator is very low, thus the cutoff frequency of the equivalent lowpass filter would exceed 50/60 Hz frequencies (and its harmonics), residuals of which one would hope to filter. So an additional electrolythic capacitor would not make a difference.
As you said, the low capacitance/low inductance 0.1 uF capacitor blocks high frequency currents of the consumer, so the voltage regulator doesn't need to react to those frequencies.
Faster Rise time of cap. charge ing rate only stresses diode there's no change in ripple frequency , only the ( p- p) ripple voltage , can / may stress the regulator..whereas capacitor ripple current must be within its specified limits..
What you call "ramp wave" actually is part of the (50 or 60 Hz) input sine wave; the last part until the voltage reaches its peak value. Far from "almost infinite frequency".
Thanks. Very helpful.
Great video. Like to hear about current pass transistor for 13.8v linear power supply. Thanks in advance. S21OD
Excellent!
Thank you - very useful!
lets say you need a 5V, 3A regulated dc voltage power supply, is it possible to build with 2 blocks of 7805 in parallel? any coupling problem between the 2 blocks?
The two regulators will fight, and (worst case) fast-oscillate at full power. You can boost up a regulator with a big external transistor, often called a "pass transistor"
The tolerances in manufacturing will mean one reg will put out a smidgen more volts, and as such take most of the load, so it will be overloaded and stressed, and the other just coasting along.
You are better off adding a high power transistor to increase the output current.
Unfortunately I cannot pop a suitable circuit into these comments, use Google, or check out the data sheets, sometimes they show you how in the application notes.
Both Pauls are essentially correct here. The best solution is a single regulator with a series pass transistor and don't forget to put the pass transistor on a nice big heat sink.
Use a buck converter, far better efficiency and higher current. Switching frequency and type of buck converter depends on application. I have not used linear regulators in years unless subregulated from a switch mode power supply (subregulation with Low Drop Out linear regulators are low noise and have pretty good power supply rejection ratio....good for radio transceivers).
true, too much capacitance can actually take too long to charge
The price you pay for regulation is 2 volts.
Hi tONY.. i have a question that sorta fits with the video and your explaination... with vintage audio equipment that has a stated mains input of 110 volts, and no soft start circuit, wouls it be advisable to add in a 'PROTECTION ' circuit ... i realize that with the volume all the way down that the speakers are protected from the initial power up with out one, but with the difference in mains supply we have today versus the 110 v stated on the equipment i am thinking that the ac spikes would cause premature compoenent wear, anybodies input to this question would be welcome ..thanks....
Banjo do you mean Vintage as in tube type or solid state ??
6:59 Does the rectifying and capacitor filter cause and in increase of the outputed DC voltage? The Vrms of the AC is lower
RMS value is a DC equivalent of an AC signal. With normal mains voltage, which is a sine wave, the peak voltage will be √2 (or 1.414) times the RMS value. The capacitor, with no load on the circuit, will charge up to this peak value. So a 12 volt AC RMS signal will charge the capacitor up to just under 17 volts DC. Check out the video I have linked at the end of this one. I go over it on that video. Thanks for the comment!
Thank you
what is the difference between a voltage regulator and a zener diode?
it is much easier and simple with f.e. a 7805
Linear regulators are regulated, so their outputs are very stable. A zener diode is "open loop" and drifts with load.
You can use a zener if you don't need precision, but linears are pretty cheap, so you might as well take the better regulation.
This video explains more of what a 78xx expects than cap value smoothing ripples. By the look of it, the 78xx does a very good job of smoothing ripples with a small physical package, as long as one gives it +1.5 working volt above output. 78xx linear while easy to use is wildly inefficient, whoever invented the switching power supply should get a Nobel.
1.5V drop out is way lower than the typical 2V of the 78xx series, 1.5V is the best case scenario at few mA.Otherwise I agree. LDO regulators are still great for small current applications but switching PSUs were a revolution for everything else.
garbage in garbage out hmmmm what happens if we run these regulators in parallel, say we need 2 or 3 branches that would be more amps than one alone could handle???
Nah doesnt work with 78xx. Use lm317 and put a high current bootstrap transistor around it. Works great.
Thank you!
Can you use a similar circuit to supply a regulated high voltage DC supply for a tube amp? Say 300 to 500 VDC? What devices are needed?
You could design a linear regulator using a mosfet (or a bank of 'em) on a humungous heat sink. Then there's the control circuit. But this whole arrangement is expensive and it wastes a lot of B+ voltage which could otherwise go toward producing power output. Quite honestly, you would be much further ahead just doing some conventional tube-era maintenance. I am the proud owner of a Bell Sound Carillon 6060 amp. As-is condition, it sounded good but not spectacular. I replaced the EL34 output tubes with "re-issues" of a well known British brand, the 5V3 rectifier with a "new-old-stock" unit I found at a reasonable price. Just renewing all the power tubes turned this amp into a real room-rattling beast! I also replaced the electrolytic filter caps with new units and that added even more "punch" to the low end. New "orange drop" coupling caps too. But the new tubes alone made the biggest difference. These things are all recommended whether you're into hi-fi or guitar. Hope you find this helpful. 8D
My intuition is that regulating the high voltage supply would diminish the qualities people value in tube amps. Voltage sag under load is often cultivated intentionally in the name of tone.
@@brocktechnology Tbh it's the same with solid state... high current egulated supplies need to be highly engineered and tested a lot. Subjectively my amps sound way nicer with linear supplies.
Very interesting content. I have subscribed.
Cheers
Nice Demonstration, good job, K1RSU
Excellent !!!!!!!!!!!!!!!!
The term (on the diagram) "smoothing" is incorrect, it is a low pass filter. (passing DC, but not the ac ripple frequency. (Pi networks are king here) The regulating IC works hard to cancel out the ripple, it uses a lot of internal gain to do that, which makes three terminal regulators nosier than properly designed discrete regulators with moderated gain. Ron W4BIN
There's noting "incorrect" about the term "smoothing capacitor" in this application, as it is a standard term that is on much information available if you bothered to research or even do a web search. A true "Pi network" also uses other components such as inductors and resistors to create either a high, low, or bandpass filter circuit for AC signals. A single capacitor in parallel with a DC source with an AC ripple component could really not be considered a "network" onto itself, so, the term "smoothing capacitor" is a good description of a single components function in this particular circuit.
@@poormanselectronicsbench2021
Yup. I agree. The cap is there to store and deliver energy and "smoothing" is a good term for that. It is commonly used in the UK.
I've seen an old amp with a lot of big caps in it which seem to run in parallel with ceramic resistors on the output channel to the speaker (i assume its a single channel amp). whats the reason for so many caps and could i improve the speaker output with a similar set up on my new Yamaha home cinema amp. The reason for this is there's a distinct distortion on low notes even at low volume. I'm an electronics assembly orerative so opening the lid and modifying or adding circuits wouldn't be a problem. many thanks. jon.
There is an old EEVblog video on why putting caps in parallel, you should watch it. TL:DW it's often a matter of ESR and ESL.