Also, what limits the current in the TL431? Shunt regulators need a current limited supply (often a resistor or current source) to work against....
Truly it is said, a little knowledge is a dangerous thing.
The danger you have fallen into is that when you finally do get to understand how these things work, you're going to have a very red face at how crass your behaviour has been.
Go away and design a CRCRC PSU filter @ 50 Hz for 2A or whatever, with sufficient headroom to run your selected regulator, select the components from Farnell, including transformer, so I can see them, and come back and tell me the peak-peak ripple you anticipate before the regulator and why it needs to be that low. NO INDUCTORS. You'll lose a few volts across the R's, but nobody in their right mind uses inductors to kill ripple in an LT supply.
I'll want to know the dissipation in the regulator. Don't forget to take into account fluctuations in the mains voltage.
The danger you have fallen into is that when you finally do get to understand how these things work, you're going to have a very red face at how crass your behaviour has been.
Go away and design a CRCRC PSU filter @ 50 Hz for 2A or whatever, with sufficient headroom to run your selected regulator, select the components from Farnell, including transformer, so I can see them, and come back and tell me the peak-peak ripple you anticipate before the regulator and why it needs to be that low. NO INDUCTORS. You'll lose a few volts across the R's, but nobody in their right mind uses inductors to kill ripple in an LT supply.
I'll want to know the dissipation in the regulator. Don't forget to take into account fluctuations in the mains voltage.
Last edited:
I have ran that schematic by an experienced professional engineer that I know personally, and I received no such comments.
My question is, How do these forums work to help anyone, when nobody on here seems to agree with anyone else? That is ofcourse when posters aren't too busy just criticizing people, and are actually trying to help people.
It may be that I am just new to these forums, but can someone explain how I'm supposed to know who's advice I should take?
My question is, How do these forums work to help anyone, when nobody on here seems to agree with anyone else? That is ofcourse when posters aren't too busy just criticizing people, and are actually trying to help people.
It may be that I am just new to these forums, but can someone explain how I'm supposed to know who's advice I should take?
Last edited:
I work on PCBs for certain industries where the EMC requirements go up to 18GHz, I have worked closely with an RF engineer to get the equipment to pass. There is more to it than just filters, layout becomes critical and understanding the coupling methods of the very high frequency noise is critical.
300GHz its rather a high figure and I believe would be quite hard to achieve...18GHz is hard enough and that's for cost no option (almost) equipement.
300GHz its rather a high figure and I believe would be quite hard to achieve...18GHz is hard enough and that's for cost no option (almost) equipement.
I already have, which is why I have the schematic that I do right now. The problem is, that nobody on here seems to agree on anything. The answer is different depending on who replies.
I will likely have to figure what works and what doesn't for myself during the build.
Also look at these because this is the area you are going to have to get into to get such a wide band of filtering....
http://cdn.vicorpower.com/documents/datasheets/Picor/ds_qpi3L.pdf
http://cdn.vicorpower.com/documents/datasheets/Picor/ds_qpi3L.pdf
I will do the best I can with HF attenuation, depending on the board space I have available.
I will however use a series of Ferrite cores on the load end of the power cable coming from the output of the supply, just to keep the attenuation as close to the load as possible.
Most seem to agree that the number of coils and capacitors is overkill, but if you want overkill then go for it. Using a series regulator and a shunt regulator together like that is a mistake though - they're both going to try to set the same output voltage, and end up fighting each other.
Won't the series regulator and the shunt regulator be on different potentials anyway, as a result of being separated by as many coils?
Yes, but it's going to be a dificult balancing act if you want variable output voltage.
e.g. If the first regulator is set to 15V, and the second is variable from 6V to 12V, then the voltage dropped across the coils varies from 3V to 9V.
Then if you want output current up to 2A available at any output voltage, then the total resistance of the 8 coils between the regulators can't be more than 1.5 Ohms (3V/2A).
Problem is when you turn the output voltage down to 6V, there's 6A of current going through the coils even with no load. With a 2A load, the total current will be 8A, which seems horribly wasteful.
Anyway, here's an idea - If you use a dual potentiometer to control both regulators, and organise it so the first regulator is always set to e.g. 2V higher than the second, then the voltage across the coils stays the same when the output voltage is varied, and you can use the resistance of the coils to set the current as you suggested.
Thanks for that, I will have a look into dual potentiometers and see what I can find that will work. Even though the enclosure will be large and well ventilated, I still want to keep the heat down.
Perhaps he didn't look hard enough. We all miss something; for example, I missed that you had added a shunt regulator to the output of a series regulator.JSmith1980 said:
Many of us do agree, but you don't seem to be listening to us. We are trying to help - no, really we are - but you make it difficult for us by ignoring our advice and steaming on with your own wrong ideas.
Make sure that the questions you ask are within your own ability to understand. If you understand the question then you have a fair chance of understanding the solutions offered, and deciding for yourself which ones make sense. Bear in mind that often there will be more than one valid solution but you cannot always combine bits from different solutions. For example, you can regulate voltage with either a shunt regulator or a series regulator but you can't do both at the same time.
This forum is not moderated for technical content, only good manners, so you have to decide for yourself who to believe. Do that by looking at what they say, not who says it. Check in other reliable sources, such as textbooks.
Sadly, the internet is a bit like democracy - everyone gets a vote, however wise or foolish they are.
I am following the advice I am given, but there is a difference between giving advice, and just saying that something can't be done. To just point out problems without offering solutions, is of no use to anyone.
Yes I will start of on a small scale, yes I will take the time to understand the circuit before I build anything, and it will be built and tested in stages to avoid mistakes.
I will still continue to develop the schematic up until the build begins, if you would like to contribute solutions, it would be appreciated.
No, you are not. You are still persisting with a multistage filter which you clearly don't understand, when you said you were going to start with something simple. You have an attempt at a two-stage voltage regulator which you clearly don't understand.JSmith1980 said:
Saying that something can't be done may be excellent advice, if indeed it can't be done.
Don't develop it. Instead, drastically simplify it. Scrap the low value inductors and the high value capacitors, as between them they provide poor performance at all frequencies.
Build a simple PSU with a reservoir cap (learn how to calculate its value, and estimate the likely ripple voltage), a simple voltage regulator, optionally followed by a simple noise filter (learn how to choose and calculate corner frequency, and choose appropriate components). You can probably get most of the circuit from the regulator datasheet. Debug it. Measure its performance. Ask us questions about anything you don't understand. Learn to walk before you attempt running. That is my advice. It is good advice.
Yes, I do need to understand more about LC filters, that's true, which is why I am on these forums. I am sure multistage filters can be done, and have been done before. I don't understand why they can't be done again.
The time I will have available to actually build a power supply will be very limited due to work restrictions, and I don't want to build something that I could just as easily buy pre constructed. This is why I want to understand the circuit before I build anything. Yes there is a learning curve, I am aware of that, but that is part of the challenge.
You mentioned that the multistage filter I already have won't work. I have checked the numbers and abraxalito has posted sim results with that proposed multistage filter back on post #37, and they both say that I will have a cut off frequency of around 5Hz, which is what I am looking for.
If you could explain why the reality would be different from the calculations, that would help me understand multistage filters better, instead of throwing the idea out of the window at the starting post.
Multistage filters can be done, but like all electronics they require understanding.
Real components are not the same as simulated components. Real inductors have stray capacitance in parallel. Real capacitors have stray inductance in series. To effectively filter ripple you need larger inductor values, or resistance instead. To effectively filter RF you need smaller inductor values and smaller capacitance values - and these need to be in separate filter stages. You can't make a filter which works from 5Hz to 30GHz. Actually, you might need an LF rolloff lower than 5Hz to get low ripple and you certainly don't need an HF rolloff much above a few MHz in the PSU. Any GHz filtering must be in the equipment to be powered, not the PSU, as a few inches of wire will undo it all.
Why not design a filter which works from, say, 2Hz to 200kHz? That would be easier.
- Status
- Not open for further replies.